ကူးယူချက် 🔗
WEBVTT
1
00:00:00.120 --> 00:00:03.759
In this episode of data framed,
a data camp podcast, I'll be speaking
2
00:00:03.759 --> 00:00:09.589
with evehillpish about how data science is
disrupting finance. Eve's name is synonymous with
3
00:00:09.750 --> 00:00:14.310
Python for finance and he is the
founder and managing partner of the Python Quants,
4
00:00:14.589 --> 00:00:19.230
a group focusing on the use of
open source technologies for financial data science,
5
00:00:19.550 --> 00:00:25.820
artificial intelligence, algorithmic trading and computational
finance. Why a bank such as
6
00:00:26.019 --> 00:00:30.899
Bank of America and JP Morgan adopting
the open source data science ecosystem? What
7
00:00:31.059 --> 00:00:35.729
are the major subdisciplines of finance that
data science is and can have a large
8
00:00:35.729 --> 00:00:40.250
impact in? How has the rise
of data science change the financial world and
9
00:00:40.329 --> 00:00:44.490
how the work is done and thought
about? Stick around to find out.
10
00:00:45.130 --> 00:00:50.920
I'm Hugo bound Anderson, a data
scientist a data camp and this is data
11
00:00:51.000 --> 00:00:59.359
frame. Welcome to data framed,
a weekly data camp podcast exploring what data
12
00:00:59.399 --> 00:01:03.399
science looks like on the ground for
working data scientists and what problems it can
13
00:01:03.479 --> 00:01:07.430
solve. I'm your host, Hubo
bound Anderson. You can follow me on
14
00:01:07.469 --> 00:01:11.870
twitter at Hugo bound and data camp
at data camp. You can find all
15
00:01:11.950 --> 00:01:18.390
our episodes and show notes at data
Campcom community slash podcast. For those interested,
16
00:01:18.579 --> 00:01:22.340
we've also got a special offer this
week for data framed listeners the opportunity
17
00:01:22.500 --> 00:01:26.219
to try data camp yourself. All
you need to do is email sales at
18
00:01:26.260 --> 00:01:30.140
data Campcom, that sales at data
campcom, with the subject line podcast and
19
00:01:30.260 --> 00:01:38.890
redeem your free two week trial.
Hi, they ave, and welcome to
20
00:01:38.930 --> 00:01:42.810
data framed. Hi there, and
thanks for having me. It's a real
21
00:01:42.930 --> 00:01:46.290
pleasure to have you on the show
and I'm really excited to have you here
22
00:01:46.370 --> 00:01:49.680
to talk about your work in finance, how you think about the use of
23
00:01:49.760 --> 00:01:53.799
Python in finance and the implications of
all of this with respect to data science
24
00:01:53.840 --> 00:01:56.840
in general. But before we get
into that, I just want to get
25
00:01:56.840 --> 00:01:59.840
a bit of context and learn a
bit about you. So maybe you can
26
00:01:59.879 --> 00:02:04.189
start by telling us what you're most
known for in the data community. I
27
00:02:04.349 --> 00:02:07.349
think there is a there is an
easy one. So I'm known for Python
28
00:02:07.590 --> 00:02:10.509
for finance. So I started using
python
Load full transcript
more than ten years ago for finance.
29
00:02:10.590 --> 00:02:14.710
Back then we started with computational finance
and people say, well, you
30
00:02:14.789 --> 00:02:17.900
can do that is to slow and
all these kind of pujudices and reasons why
31
00:02:17.979 --> 00:02:23.060
you can do something. And today
the biggest institutions around the world use python
32
00:02:23.219 --> 00:02:25.419
for exactly that, in a particular
for iber trading. So it has moved
33
00:02:25.419 --> 00:02:30.490
a little bit from the comp finance
side to the ARGO trading side, and
34
00:02:30.729 --> 00:02:34.969
I think people now know me for
this as well. So whenever somebody thinks
35
00:02:35.009 --> 00:02:38.610
of coding up their strategy and trying
to deploy it automatically for trading, it's
36
00:02:38.689 --> 00:02:43.569
yeah when they use python. And
many people book go online courses or listen
37
00:02:43.610 --> 00:02:46.080
to talks already the books. So
fantastic. And how did you get there?
38
00:02:46.120 --> 00:02:50.560
How did you get into Python,
finance, now data science? What
39
00:02:50.639 --> 00:02:53.719
type of journey I led you to
where you are now? Well, our
40
00:02:53.879 --> 00:02:58.039
after our German I'd be tour,
which is how we finish school, high
41
00:02:58.080 --> 00:03:01.590
school, so to say, on
German Gymnasium. I studied business administration and
42
00:03:01.990 --> 00:03:06.110
already focus on the final side of
things. Everything that it did to in
43
00:03:06.229 --> 00:03:12.789
my diploma in Business Administration was centered
around finance, financial markets, banking and
44
00:03:12.900 --> 00:03:16.139
all these topics that interest me from
the beginning. I later on started,
45
00:03:16.340 --> 00:03:22.219
immediately after my started working on my
PhD thesis, which was a mathematical finance.
46
00:03:22.219 --> 00:03:25.139
So it got more formal overtime.
Actually, I had a topic about
47
00:03:25.180 --> 00:03:30.370
Aynamic catching and feedback in markets due
to hatching of options, and so really
48
00:03:30.409 --> 00:03:34.250
evolved in terms of math, what
you need. But I really enjoyed it,
49
00:03:34.490 --> 00:03:39.090
the analytical challenges and writing some theoretical
stuff and using law take and all
50
00:03:39.169 --> 00:03:43.199
these things. But in the meantime
I started working as a menation because ide
51
00:03:43.199 --> 00:03:46.000
because I thought, well, math, finance, it's nice and charging,
52
00:03:46.080 --> 00:03:47.680
but if you later want to work
in any kind of institute, any kind
53
00:03:47.680 --> 00:03:52.080
of like organization, Corporation or whatnot, to have seen a little bit more
54
00:03:52.199 --> 00:03:54.360
might be good. So I started
as a general management consultant, working for
55
00:03:54.479 --> 00:03:59.669
different companies of different types of projects, but again the focus was on finance.
56
00:03:59.870 --> 00:04:03.669
Didn't leave the financial industry that much. Yeah, after that, finished
57
00:04:03.669 --> 00:04:09.229
my PhD, I moved first time
within Germany to hand work. Back then
58
00:04:09.270 --> 00:04:14.259
it was around the hype of the
Internet startups and the web was then the
59
00:04:14.340 --> 00:04:16.699
hot thing, and so we did
some with the company I joined with it,
60
00:04:16.819 --> 00:04:20.459
some consulting around wet topics and so
forth. But after the burst of
61
00:04:20.500 --> 00:04:25.089
the bubble this was no work anymore. So we needed to look for something
62
00:04:25.089 --> 00:04:27.889
else. And there's something else.
Nots we or what we did. We
63
00:04:28.050 --> 00:04:30.329
started our own company. So I
pound of my first company in two thousand
64
00:04:30.370 --> 00:04:34.089
and one actually pretty nice because last
Friday I visited the company for their summer
65
00:04:34.170 --> 00:04:39.040
first. So I left the company
into two person reasons, moved back to
66
00:04:39.360 --> 00:04:43.279
my hometown in this island area due
to the family, but we're still good
67
00:04:43.319 --> 00:04:46.079
speaking terms and the company is going
pretty pretty well. I must say that
68
00:04:46.120 --> 00:04:49.560
must have been an interesting experience.
Seventeen years later, bring back. Actually
69
00:04:49.800 --> 00:04:53.629
the company is going, I didn't
know, to more than seventy consultants these
70
00:04:53.670 --> 00:04:57.430
days. has been taken over by
a bigger, large company. So I
71
00:04:57.629 --> 00:05:00.069
hotly knew all the people around there, but of course the founders and got
72
00:05:00.149 --> 00:05:03.350
of the older ones. This was
pretty nice for me. Had a nice
73
00:05:03.389 --> 00:05:08.220
evening there and when I moved back, this was actually the starting point where
74
00:05:08.259 --> 00:05:12.860
I use for first time Python for
finance, because I discovered it before moving
75
00:05:12.899 --> 00:05:15.220
back and I found it my own
company, which I still own, which
76
00:05:15.339 --> 00:05:19.180
is now the python ones, and
immediately I got started with some side projects
77
00:05:19.220 --> 00:05:24.290
which I couldn't pursue in the other
contacts with other co founders that didn't want
78
00:05:24.329 --> 00:05:26.970
to do something like that. So
now I got the freedom to pursue my
79
00:05:27.050 --> 00:05:30.769
own stuff and this was, among
others. This was python, and we
80
00:05:30.050 --> 00:05:33.610
started pick them where even number wasn't
around, or we use no more ray
81
00:05:33.769 --> 00:05:36.360
and no marry, not the well, that was going to be one of
82
00:05:36.399 --> 00:05:40.439
my question like. This is before
number, this is before pandas, this
83
00:05:40.560 --> 00:05:43.720
is before so many of the technologies
that people are quite with the data science,
84
00:05:43.759 --> 00:05:46.439
Python Stack now. So it must
have been a wild landscape. Exactly,
85
00:05:46.519 --> 00:05:50.269
exactly, so dumb things that are
now kind of like standard and that
86
00:05:50.389 --> 00:05:54.389
you might teach. I had to
know in the third hour of training or
87
00:05:54.470 --> 00:05:57.670
what with. And there's we needed
to code up on our own. Lifetime
88
00:05:57.750 --> 00:06:00.269
series manager wanted to do something in
computation finders and all these things. But
89
00:06:00.350 --> 00:06:03.980
I'm more than happy that. For
examples, mckinney started the finance project,
90
00:06:04.100 --> 00:06:08.860
and have you bought the project having
chroned that much and providing us with all
91
00:06:08.939 --> 00:06:13.939
the Nice cupabilities that we today use
and finance. But for sure we was
92
00:06:14.620 --> 00:06:17.019
a different landscape. Many people can't
imagine how it looked like then, but
93
00:06:17.220 --> 00:06:20.970
I was convinced, due to the
beauty of the language and the whole approach,
94
00:06:21.170 --> 00:06:25.089
that this might be the future.
And Yeah, well, this time
95
00:06:25.129 --> 00:06:29.610
I think I was prone right,
given this success off Python in our industry
96
00:06:29.689 --> 00:06:32.199
and today we don't do anything else. It's all about Python and finance.
97
00:06:32.240 --> 00:06:34.439
And I got traded. Yeah,
and particularly with, you know, the
98
00:06:34.519 --> 00:06:38.480
emergence of I python as well,
and you butter notebooks, which you know,
99
00:06:38.560 --> 00:06:41.600
back in two thousand one days went
around either. So that's very exciting
100
00:06:41.680 --> 00:06:45.079
that all these developments have converged in
this landscape that allow us to do what
101
00:06:45.160 --> 00:06:47.750
we do. That's fantastic us,
I think, from scientific point of view,
102
00:06:47.910 --> 00:06:50.430
from developers point of view, they
de signs point of view, or
103
00:06:50.550 --> 00:06:54.230
financially, data science, if we
want to put it that way, for
104
00:06:54.350 --> 00:06:58.470
sure. So what happened then in
your trajectory? When I was talking about
105
00:06:59.029 --> 00:07:01.100
Python being a side project? Indeed
was the side process. We couldn't make
106
00:07:01.139 --> 00:07:05.459
any money out of that, but
we did regular other consulting work that I
107
00:07:05.579 --> 00:07:11.100
was doing before, and I think
it was maybe like six years ago probably
108
00:07:11.420 --> 00:07:15.339
all we started getting real traction maybe
seven years I think it was two thousand
109
00:07:15.339 --> 00:07:18.930
and level. We've got the first
big client in Germany actually toutsch a brewser
110
00:07:19.569 --> 00:07:26.810
the derivatives exchange, your x approach
to stay actually one of the executives.
111
00:07:26.889 --> 00:07:30.399
Have seen on Youtube a talk of
myself at Europe Python and Florence, and
112
00:07:30.519 --> 00:07:33.079
this is how things work. And
as a well, do you do training
113
00:07:33.079 --> 00:07:35.759
in this reek guard and can you
support us? And I said, well,
114
00:07:35.959 --> 00:07:40.759
for sure, this is what we
were waiting for it. And so
115
00:07:41.040 --> 00:07:44.629
this was more less than the former
starting point. And Yeah, it's been
116
00:07:44.709 --> 00:07:47.389
crowing. If you're working with many
banks, we catch funds around the world
117
00:07:47.430 --> 00:07:54.470
and having hundreds of training assume thousands
of people on our pun platform that use
118
00:07:54.509 --> 00:07:58.230
python. And Yeah, so today
we don't do anything else. If people
119
00:07:58.310 --> 00:08:01.420
look up python, finance or whatever, I think Google or wherever they they
120
00:08:01.579 --> 00:08:05.060
usually find us and get to our
trainings, to my talks. I've given
121
00:08:05.100 --> 00:08:07.819
I don't know how many talks,
more than a well, more than a
122
00:08:07.860 --> 00:08:11.699
hundred of the last five years,
and conferences around the world. And for
123
00:08:11.779 --> 00:08:15.930
sure. so you write books,
you can, so, you develop training,
124
00:08:16.490 --> 00:08:18.850
you host made ups. The other
thing I mean that these, not
125
00:08:18.970 --> 00:08:22.329
that these aren't enough, but the
right these are the main things, that
126
00:08:22.449 --> 00:08:26.009
where your focused lies currently. Yeah, actually, so, I basically see
127
00:08:26.050 --> 00:08:31.480
us these days as more less content
oriented company. So this is what I
128
00:08:31.600 --> 00:08:37.159
think our core is. Indeed,
so writing books of courses about content creation,
129
00:08:37.320 --> 00:08:41.629
but also designing and delivering online trainings. Also, I put in the
130
00:08:41.710 --> 00:08:46.789
same category like events, where of
course you also need some content. We
131
00:08:46.909 --> 00:08:50.830
used to do also more conference but
these days you focus, for example,
132
00:08:50.870 --> 00:08:54.990
with our partner hitch learning, on
the boot camp side. So this is
133
00:08:54.070 --> 00:08:58.659
also more about training and what we
need to or deliver the content. So
134
00:08:58.779 --> 00:09:01.779
content of courses at the core and
also, I think, skills and know
135
00:09:01.940 --> 00:09:07.460
how which is then used in addition
to the content, where we consult clients
136
00:09:07.500 --> 00:09:11.929
around the world, like working with
a big broker in New York, Manhattan,
137
00:09:11.090 --> 00:09:15.610
or working with Hatch Fund in London. This, of course, requires
138
00:09:15.809 --> 00:09:18.409
certain skill set. Now, how
so, but this is more less where
139
00:09:18.450 --> 00:09:22.850
I'm coming from. The content side
has become more and more important over recent
140
00:09:22.929 --> 00:09:26.399
years, in particular with our online
training, which has growing at tremendously,
141
00:09:26.519 --> 00:09:30.600
which I'm really happy about. And
you mentioned the meat ups. Running a
142
00:09:30.679 --> 00:09:35.000
few Med ups actually in London,
Berlin, Frankfurt, per resultso one in
143
00:09:35.320 --> 00:09:37.990
New York, and whenever I'm there
try to do their something as well.
144
00:09:39.149 --> 00:09:43.149
So it's keeping me busy. Definitely
sounds like it. So you, you're
145
00:09:43.190 --> 00:09:46.149
known as the Python quant you work
with a team of people and you call
146
00:09:46.190 --> 00:09:50.190
yourselves the Python quants, and I
think myself and our listeners, no,
147
00:09:50.389 --> 00:09:52.580
python is but I don't know whether
everyone has a lot of context around what
148
00:09:52.700 --> 00:09:56.820
a quant is. I definitely don't
feel I have enough. So maybe you
149
00:09:56.940 --> 00:10:01.059
can kind of unpack this term for
us. quant. Yeah, now,
150
00:10:01.139 --> 00:10:03.139
if you want to get to live
with academy, who was that? Well,
151
00:10:03.179 --> 00:10:07.169
we have different types of quants,
but I think these days and has
152
00:10:07.450 --> 00:10:11.889
narrow down a little bit. So
back in the days when I started with
153
00:10:11.129 --> 00:10:16.809
the que side of things and finance, what people typically understood and the quant
154
00:10:16.970 --> 00:10:20.250
was kind of the model quie.
Somebody was sitting down and comes up with
155
00:10:20.850 --> 00:10:24.320
specific model, for example, to
price a specific type of financial derivative,
156
00:10:24.399 --> 00:10:30.799
so complex financial instrument that might depend
on a couple of risk factors like interest
157
00:10:30.840 --> 00:10:35.279
rates or, let's say, a
stock index or maybe a basket of stocks,
158
00:10:35.320 --> 00:10:37.470
for example. So if people sitting
down and really doing research on their
159
00:10:37.509 --> 00:10:41.710
plackboard or these days probably the white
boards or pain and paper. Back in
160
00:10:41.750 --> 00:10:45.710
the day, is a little bit
maybe on the computer, but more that's
161
00:10:45.750 --> 00:10:48.389
the document. But these were really
kind of what mathematicians, yeah, and
162
00:10:48.629 --> 00:10:52.740
and physicists as well. Right a
lot of physicists this is what people called
163
00:10:52.820 --> 00:10:58.179
back then the rocket scientists because,
yeah, the fact that many physicists came
164
00:10:58.259 --> 00:11:01.820
to was because the mathematics that you
need to price financially roots is pretty similar
165
00:11:01.860 --> 00:11:05.490
to what physicists use on a daily
basis and and many of the areas,
166
00:11:05.690 --> 00:11:11.009
so engineers and so forth. So
this was the the origin or to see
167
00:11:11.009 --> 00:11:13.409
if the Quan's the rocket scientists and
so forth. But these days I think
168
00:11:13.610 --> 00:11:18.679
it's much more data driven. So
I would say these days a quant is
169
00:11:18.799 --> 00:11:24.440
more like a
financial data scientists,
because they they need to prune huge numbers
170
00:11:24.480 --> 00:11:28.360
of data, and I think it
might differ to some extend to the area.
171
00:11:28.639 --> 00:11:31.639
Have a look at, for example, if there is, let's say,
172
00:11:31.799 --> 00:11:35.830
an equity research analyst who crunches numbers
of certain companies, let's say apple,
173
00:11:37.110 --> 00:11:39.629
you typically are not faced with that
amount of data. But if you
174
00:11:39.669 --> 00:11:45.710
have others that do kind of like
more systematic analysis, for example for systematic
175
00:11:45.870 --> 00:11:50.019
trading strategies, they might call themselves
even data engineers. So one of the
176
00:11:50.220 --> 00:11:54.820
the biggest hatch funds out there,
more successful ones to Seema, headquartered in
177
00:11:56.059 --> 00:12:00.620
York, they call these guys and
the data engineers and because they are,
178
00:12:01.220 --> 00:12:07.490
they are produce more or less independent
of what the financial professors, people coming
179
00:12:07.529 --> 00:12:13.169
up with financial theories say how the
market should behave they have kind of like
180
00:12:13.370 --> 00:12:16.399
a pretty mutual approach and saying,
well, let's apply whatever technique to the
181
00:12:16.440 --> 00:12:20.559
data that we can get our hands
on and see how we can profit from
182
00:12:20.600 --> 00:12:24.399
that. So this is what I
call in all my talks these days data
183
00:12:24.480 --> 00:12:30.950
triven finance instead of kind of this
equation driven and praying driven finance. So
184
00:12:31.070 --> 00:12:35.909
people sit down and think of how
the financial world should behave they rather have
185
00:12:35.990 --> 00:12:37.950
a look at the data and try
to figure out something? Maybe not coming
186
00:12:37.990 --> 00:12:43.950
up with kind of this fantastic,
nice single equation which might award you Nobel
187
00:12:43.990 --> 00:12:48.059
Prize in economics someday in the future, know, but with things that might
188
00:12:48.179 --> 00:12:50.659
work? Similar project. Every big
data company out there, like all the
189
00:12:50.779 --> 00:12:54.980
social base companies like twitter or spotty
by fan, how they come up with
190
00:12:54.980 --> 00:12:58.730
their recommendations? It simply have a
look at the data for the recommend the
191
00:12:58.769 --> 00:13:03.289
engines use machine learning approaches and recommend
a song number three. You and people
192
00:13:03.409 --> 00:13:07.450
usually hear three in combination with one
and two, and this is what these
193
00:13:07.490 --> 00:13:11.250
data engineers, these ones these days, do as well. So having a
194
00:13:11.490 --> 00:13:13.759
data driven mindset and say, well, let's have a look at the data
195
00:13:13.919 --> 00:13:18.799
and apply whatever technique might bring us
something in this regard. So people working
196
00:13:18.080 --> 00:13:22.519
with quantitative things usually in a miracle
data and more and more unstructured text based
197
00:13:22.559 --> 00:13:26.120
data. This is what the ones
do these days. So we might have
198
00:13:26.389 --> 00:13:31.789
still a few handful of modelic ones
that are started with, but it's just
199
00:13:31.950 --> 00:13:35.789
a handful compared to thousand others that
work with financial data in this area.
200
00:13:37.230 --> 00:13:39.789
Okay, and so to reframe that
slightly, what one of the ideas there
201
00:13:39.950 --> 00:13:46.299
is that in the large data limit, theory laden models may be unnecessary essential.
202
00:13:46.340 --> 00:13:48.580
Essentially, yes, because when you
have a look at the history,
203
00:13:48.860 --> 00:13:52.899
people have been pretty successful with coming
up with nights models when they made let's
204
00:13:52.899 --> 00:13:58.490
say, appropriate assumptions, and the
forum typically of normally distributed returns and linear
205
00:13:58.610 --> 00:14:03.250
relationships. But this is already where
all these theories are doomed to fail,
206
00:14:03.330 --> 00:14:07.649
because we don't have normally distributed returns
and markets across all asser classes. You
207
00:14:07.730 --> 00:14:11.200
can analyze the returns on the markets, they are not normally distributed and in
208
00:14:11.320 --> 00:14:15.559
general, the relationships that you face
are nonlinear and not linear, and therefore
209
00:14:15.919 --> 00:14:20.799
having a different look at the nonlinear
changing world with different algorithms white prove more
210
00:14:20.840 --> 00:14:26.029
fruitful than to rely on the old
theories from the sixty s s s that
211
00:14:26.070 --> 00:14:31.230
are still in use today. Yeah, absolutely so. In general, what
212
00:14:31.350 --> 00:14:35.710
are the major subdisciplines of finance that
you think data science is and can have
213
00:14:35.789 --> 00:14:39.580
a large impact in? So I
must say I'm in general, only on
214
00:14:39.139 --> 00:14:43.940
the investment and corporate banking site,
so to say. I hardly have any
215
00:14:43.340 --> 00:14:48.340
point of contact with the retail side. So most people, think everybody who
216
00:14:48.340 --> 00:14:52.500
will listening to what we are talking
about do. Of course they are financial
217
00:14:52.500 --> 00:14:56.370
stuff to maybe on a daily basis
with their APPs on the phones or online
218
00:14:56.409 --> 00:14:58.169
banking on the web, but this
is not the part I'm involved in.
219
00:14:58.370 --> 00:15:03.529
Typically, there I think there are
tremendous success stories of kind of like data
220
00:15:03.570 --> 00:15:09.240
signs, for example, and credit
lending, which is mostly automated these days
221
00:15:09.279 --> 00:15:13.679
using machine learning algorithms like scoring there
and so for but again, this is
222
00:15:13.759 --> 00:15:16.240
not my field of experten so what
I'm mostly concerned with this the corporate
and
223
00:15:16.240 --> 00:15:20.429
investment banks are and they are particular
to the routus and the trading side and
224
00:15:20.710 --> 00:15:24.750
in general, and this is how
I think of this little world, smaller
225
00:15:24.830 --> 00:15:28.429
than the word others, but it's
financial data signs or wherever you need to
226
00:15:28.590 --> 00:15:33.470
crunch the Pata Bites of the Aa
that or a wailabile. These days,
227
00:15:33.549 --> 00:15:35.340
it is the financial data signs,
and you might find, we discussed this
228
00:15:35.460 --> 00:15:39.539
with regarded ones who and find people
in different areas of bank, for example,
229
00:15:39.620 --> 00:15:43.620
or Hatch Fund or other by side
company like an as sub management company,
230
00:15:43.740 --> 00:15:48.100
that are concerned with financial data signs, one of smaller or, in
231
00:15:48.220 --> 00:15:52.889
general, these days, on a
larger scale, so crunching any type of
232
00:15:52.970 --> 00:15:56.370
financial data, market data, unstructured
data like news data and so for then
233
00:15:56.409 --> 00:16:00.490
we have the trading side of things, where people try to come up with
234
00:16:00.889 --> 00:16:04.600
algorithms, and they are different types
of algorithms, and more and more you
235
00:16:04.679 --> 00:16:10.799
see people are trying to apply aiyebased
algorithms. Contrary, let's say, for
236
00:16:10.840 --> 00:16:15.559
example, to some deterministic algorithms that
you need to execute larger trades, they
237
00:16:15.720 --> 00:16:21.590
try to come up with some machine
learning eyebase algorithms that predict markets and if
238
00:16:21.629 --> 00:16:25.669
they are well enough in predicting markets, that might benefit of that. So
239
00:16:25.870 --> 00:16:29.309
we recently started such an endeavor selves
as well. And then, of course,
240
00:16:29.389 --> 00:16:33.379
computation of finance, which includes areas
such as derivatives and options pricing.
241
00:16:33.620 --> 00:16:37.059
This also includes risk management, for
example. Risk to management is still the
242
00:16:37.259 --> 00:16:41.299
big, big topics of when you
have a big investment bank that sitting on
243
00:16:41.419 --> 00:16:47.019
let's say a million plus derivatives positions
overnight, one of the major tasks over
244
00:16:47.059 --> 00:16:52.330
night is to come up with some
risk numbers for the complete port for you,
245
00:16:52.450 --> 00:16:56.450
that the face. So maybe people
have food about valued risk or credit
246
00:16:56.490 --> 00:17:00.570
value risk or the x Raas.
We have few more these days and all
247
00:17:00.610 --> 00:17:06.279
field and it's really computationally demanding and
such jobs are running on huge clusters with
248
00:17:06.519 --> 00:17:11.319
thousands of computers overnight to crunge the
numbers in an appropriate way for a big
249
00:17:11.440 --> 00:17:14.839
bank to get, let's say,
at easy view on the risk position.
250
00:17:14.960 --> 00:17:18.670
So computation of finance usually is kind
of the most demanding in this field,
251
00:17:18.670 --> 00:17:22.109
and this is mentionally before we get
started on a small scale. But these
252
00:17:22.150 --> 00:17:25.829
days, for example, the biggest
banks in the world, like Bank of
253
00:17:25.950 --> 00:17:30.420
American reline for their trading and risk
mention platform. They mainly use python,
254
00:17:30.539 --> 00:17:34.940
for example, as the implementation language, although the had core calculations are still
255
00:17:34.980 --> 00:17:40.339
done and C plus plus not so
much since pice ones too slow, but
256
00:17:40.420 --> 00:17:44.569
they started developing their pricing libraries like
decades a going back then, there was
257
00:17:44.609 --> 00:17:48.009
no way around c plus plus in
this computationally demanding area. So financial data
258
00:17:48.009 --> 00:17:52.049
science are goot training computation of finance
are at least our areas. Will we
259
00:17:52.130 --> 00:17:57.250
focus on and apply data science techniques
in the financial field? Fantastic. And
260
00:17:57.450 --> 00:18:03.319
could you just slightly unpack the difference
between financial data signs and computational finance a
261
00:18:03.400 --> 00:18:07.079
bit more? You typically, what
you what to do in data science is
262
00:18:07.160 --> 00:18:08.160
that you have a look at the
data. That is they are, meaning
263
00:18:08.240 --> 00:18:12.710
historical data. Be on a simple
level, end of day data for Apple
264
00:18:12.829 --> 00:18:17.150
Stock. Over ten years, then
you have probably is some two hundred fifty
265
00:18:17.190 --> 00:18:21.710
two data points per year. After
ten years you have two thousand five hundred
266
00:18:21.710 --> 00:18:23.869
data points. So this is not
really challenging these days, as we know,
267
00:18:25.190 --> 00:18:27.380
but this is basically wherever we financial
theory is based on. But rather,
268
00:18:27.420 --> 00:18:32.980
when you have a look at apple
tick data, which submitted and provided
269
00:18:33.019 --> 00:18:36.220
by Nastag, why are data provide
us, such as Bloomberg and Tom's Roy
270
00:18:36.259 --> 00:18:40.579
do's with which we work, then
you might get some two thousand tick points
271
00:18:40.700 --> 00:18:44.329
per quarter of an hour or eight
thousand per hour. So this is then
272
00:18:44.490 --> 00:18:48.690
where we get to bigger data and
people need some different techniques than, let's
273
00:18:48.690 --> 00:18:52.289
say, an excel spretchy, for
example, to work with such data.
274
00:18:52.970 --> 00:18:59.359
But no matter what, it's typically
historical data and you might try to come
275
00:18:59.400 --> 00:19:03.839
up with some predictions, some forward
looking numbers or whatever based on the historical
276
00:19:03.920 --> 00:19:08.079
data. Computation of finance it's more
or less with the guard to the areas
277
00:19:08.079 --> 00:19:11.829
that I've been describing, like risk
measurement over time is based typically a moder
278
00:19:11.869 --> 00:19:18.349
caror simulation, where which is by
definition a forward looking technique. So while
279
00:19:18.390 --> 00:19:22.430
I might have a look backwards ten
years and apple stock in computation of finance,
280
00:19:22.470 --> 00:19:25.339
when I want to prize the derivative, I have a look forward,
281
00:19:25.380 --> 00:19:29.940
let's say over three months or twelve
months or two years, three years,
282
00:19:30.140 --> 00:19:36.539
and try to simulate the markets and
model correlations between different risk factors to come
283
00:19:36.579 --> 00:19:40.970
up with somehow good understanding of what
the future might look like in terms of
284
00:19:41.049 --> 00:19:45.890
market prices and other relevant quantity.
So my thinking is rather is in that
285
00:19:45.049 --> 00:19:49.289
sense that data sign sup looks at
the existing data and tries to come up
286
00:19:49.289 --> 00:19:53.720
with certain points to Dick, but
computation finance in and of itself has a
287
00:19:53.960 --> 00:19:59.839
forward looking element that is dominant.
They're trying to better understand what the future
288
00:19:59.880 --> 00:20:03.960
might look like. Not Coming up
with the single, let's say, forecast
289
00:20:03.039 --> 00:20:07.150
for the Apple Stock Price and twelve
months. No, rather with the distribution
290
00:20:07.430 --> 00:20:12.630
of possible apple stock prices intworth months
based on hundred thousand or five hundred thousand
291
00:20:12.710 --> 00:20:18.109
simulations of the Apple Stock Price.
Fantastic. Now there are two terms that
292
00:20:18.390 --> 00:20:21.430
you've used it. I just wanted
you to explain briefly for me the listeners,
293
00:20:21.539 --> 00:20:25.059
derivatives and options, because none of
us are necessarily know much about fine.
294
00:20:25.059 --> 00:20:27.819
Yeah, sure, these are actually
they are involved in, typically as
295
00:20:27.859 --> 00:20:32.180
a let's say, general investor.
When you when you say you, you
296
00:20:32.299 --> 00:20:34.500
want to say for a time and
they typically don't get in touch with these
297
00:20:34.619 --> 00:20:37.769
instruments. But they are used in
many different areas. They are, for
298
00:20:37.930 --> 00:20:41.849
is, used to do some risk
management. You can use derivatives in general.
299
00:20:42.009 --> 00:20:45.490
Options is a subclass of derivatives,
to do risk management. You can
300
00:20:45.529 --> 00:20:51.039
also use to speculate and so forth. But basically what they typically are is
301
00:20:51.559 --> 00:20:56.440
is where the name comes from.
Their price is derived from another financial instrument.
302
00:20:56.599 --> 00:21:00.799
So, for example, the apple
stock straight in on NASTAC and you
303
00:21:00.880 --> 00:21:04.150
can buy it and price might go
up down. This is a
straightforward thing.
304
00:21:04.470 --> 00:21:10.349
But there are options traded on things
like apple stocks or on the SMP
305
00:21:10.470 --> 00:21:15.069
five hundred or on other instruments to
this end that they derived their prize directly
306
00:21:15.109 --> 00:21:18.579
from what the underlying this is how
it's called, and this case the apple
307
00:21:18.660 --> 00:21:22.579
stock is doing and the option,
for example, a call option for example,
308
00:21:22.819 --> 00:21:27.700
represent the right to buy the apple
stock at some certain point in the
309
00:21:27.779 --> 00:21:34.490
future at a predetermined price. So
options and that sense typically represent some rights
310
00:21:34.930 --> 00:21:41.450
which you can exercise, but typically
you are not required to exercise them.
311
00:21:41.569 --> 00:21:45.009
So this is where the optionality comes
from. The other derivatives like futures,
312
00:21:45.089 --> 00:21:48.960
that are unconditional, so you buy
them. The price is also derived from
313
00:21:48.960 --> 00:21:55.200
something else, but this is more
or less live or die. Once in
314
00:21:55.359 --> 00:21:57.119
your in you can only sell this
thing, but with the option you have
315
00:21:57.200 --> 00:22:02.829
the right to buy something at the
predetermined price at the future, a predetermined
316
00:22:02.869 --> 00:22:04.829
date or over a certain period of
time, or to sell it. That
317
00:22:04.950 --> 00:22:08.309
we speak of a put option,
put options to sell, call options to
318
00:22:08.470 --> 00:22:14.150
call, and the pricing of these
in Siman's might get really tricky and evolved
319
00:22:14.269 --> 00:22:18.140
and you need advanced financial mathematics in
order to come up with a proper prize.
320
00:22:18.259 --> 00:22:22.500
The pricing of Apple Stock is actually
pretty simple for a single trader.
321
00:22:22.740 --> 00:22:26.099
You simply open your promser and you
look up the price and it's there.
322
00:22:26.259 --> 00:22:30.250
So but to price derivatives, it's
not straightforward. So this is where many,
323
00:22:30.329 --> 00:22:34.049
many people and many many books have
been written in this we God they
324
00:22:34.130 --> 00:22:40.890
callor this topic. For sure.
Great, that makes perfect sense. Will
325
00:22:40.890 --> 00:22:44.319
jump right back, you know.
Interview with Ave helpish after a short segment.
326
00:22:48.319 --> 00:22:52.720
Now it's time for a segment called
statistical lesson of the week. I'm
327
00:22:52.759 --> 00:22:56.279
here with Emily Robinson, a data
scientist on the growth team here at data
328
00:22:56.319 --> 00:22:59.549
camp. Hi there, emily,
Hi hug thanks for having me on.
329
00:23:00.069 --> 00:23:03.230
I want to discuss an important yet
often misunderstood topic and statistics, the P
330
00:23:03.349 --> 00:23:07.630
value. I love it. First, for our listeners who may not have
331
00:23:07.990 --> 00:23:11.980
a strong statistical
background, what is
a pay value? Well, let's start
332
00:23:12.019 --> 00:23:18.059
by talking about the statistical hypothesis testing
process. If you have a hypothesis about
333
00:23:18.099 --> 00:23:21.740
differences between groups, for the rate
in a population, you can test that.
334
00:23:22.420 --> 00:23:26.140
For example, maybe your hypothesis is
that making the sign up button on
335
00:23:26.259 --> 00:23:30.809
your website. Bigger will increase the
percentage of visitors signing up. That would
336
00:23:30.849 --> 00:23:36.089
be your alternative hypothesis. You're no. Hypothesis which you're trying to disprove is
337
00:23:36.170 --> 00:23:38.250
that there will be no difference in
the rate of sign ups. Let's say
338
00:23:38.289 --> 00:23:41.559
you run an AV test, sending
half your people to the old version and
339
00:23:41.720 --> 00:23:45.559
half to the new version, and
measure the percentage of people signing up in
340
00:23:45.599 --> 00:23:49.200
each group. After a week,
maybe you get forty five percent signed up
341
00:23:49.240 --> 00:23:52.880
in the old version and fifty percent
in the new sounds like it was a
342
00:23:52.960 --> 00:23:57.589
success. Let's launch it. Not
Quite yet, Hugo. If you flipped
343
00:23:57.589 --> 00:24:00.990
a coin twice and it came up
heads both times, would you immediately say
344
00:24:02.069 --> 00:24:06.509
the coin was unfair? No,
you understand that even fair coins don't always
345
00:24:06.509 --> 00:24:10.380
turn up tails fifty percent of the
time. You need to take into account
346
00:24:10.420 --> 00:24:14.819
how often the results you see would
happen if you're an all hypothesis that there
347
00:24:15.059 --> 00:24:18.339
is no difference was true and ipothesis
testing. We do this with a p
348
00:24:18.460 --> 00:24:22.579
value. The P value is a
probability that, if an all hypothesis is
349
00:24:22.819 --> 00:24:27.809
true, you get a result as
extreme as the one you did to calculate
350
00:24:27.849 --> 00:24:33.130
your p value, you find the
appropriate statistical test for your data. In
351
00:24:33.329 --> 00:24:37.690
this case that would be a proportion
test. We're interested if the proportion of
352
00:24:37.769 --> 00:24:41.519
people signing up in the old versus
the new version is different. Maybe we
353
00:24:41.680 --> 00:24:47.440
enter our numbers and find that the
P value is two seven. We generally
354
00:24:47.480 --> 00:24:51.960
use P equals point zero or five
as a threshold for rejecting then all hypothesis.
355
00:24:52.680 --> 00:24:56.630
So in this case we'd fail to
reject then all hypothesis and we wouldn't
356
00:24:56.630 --> 00:25:00.230
be able to say that there's a
difference between the old and the new version.
357
00:25:00.269 --> 00:25:03.309
Oh shocks, I guess you can't
win them all. So what are
358
00:25:03.349 --> 00:25:08.380
some of the most common misconceptions that
people have about pay values? Well,
359
00:25:08.619 --> 00:25:11.940
what is that? It's the probability
of your result being a false positive.
360
00:25:12.420 --> 00:25:15.299
So if you run a test and
you get a P value of point zero
361
00:25:15.500 --> 00:25:19.500
four, people think that that means
that there's a four percent chance of an
362
00:25:19.579 --> 00:25:26.410
all hypothesis being true. But imagine
you had a thousand people all running the
363
00:25:26.450 --> 00:25:30.049
same test to see if people are
happier when they see the color red than
364
00:25:30.089 --> 00:25:33.369
the color blue. Sadly, unknown
to them, there's actually no difference.
365
00:25:33.490 --> 00:25:37.480
And then all hypothesis is true,
but we know that about five percent of
366
00:25:37.519 --> 00:25:41.079
them will have a p value of
less than point zero five and that they'll
367
00:25:41.079 --> 00:25:47.000
reject an all hypothesis. But all
of those are false positives. There wasn't
368
00:25:47.000 --> 00:25:49.349
a less than five percent chance of
a false positive, there was a hundred
369
00:25:49.349 --> 00:25:56.029
percent. That's because the probability of
your result being a false positive depends on
370
00:25:56.109 --> 00:26:00.390
the ground truth of whether than all
hypothesis is or isn't true. So are
371
00:26:00.470 --> 00:26:07.500
there any other misconceptions that people generally
have? Yes, sometimes people think that
372
00:26:07.700 --> 00:26:11.900
the smaller the P value, the
bigger and thus more important the effect in
373
00:26:11.059 --> 00:26:15.299
actuality. The P value can vary
just as a function of the sample size.
374
00:26:15.819 --> 00:26:19.210
For example, let's go back to
our B test one, where forty
375
00:26:19.250 --> 00:26:23.170
five percent of people signed up in
the old version and fifty percent in the
376
00:26:23.289 --> 00:26:27.009
new. If you would a thousand
people in each group, the P value
377
00:26:27.049 --> 00:26:30.809
of the difference is point zero three, but if you had tenzero people in
378
00:26:30.890 --> 00:26:36.480
each that p value is now less
than point zero zero zero one. The
379
00:26:36.640 --> 00:26:40.400
percentage difference between in the groups is
the same in both cases, but we
380
00:26:40.480 --> 00:26:45.599
get different P values just because of
the sample sized difference. So what most
381
00:26:45.680 --> 00:26:48.789
people should really be interested in is
the effect size. The P value just
382
00:26:48.910 --> 00:26:52.150
tells you whether you can look at
it. For example, let's say you
383
00:26:52.269 --> 00:26:56.549
have a website with lots of traffic. You Run an a bit test on
384
00:26:56.630 --> 00:27:00.549
a new feature and you increase click
the rate P lesson points your five,
385
00:27:00.269 --> 00:27:03.859
but maybe only increase it by point
one percent, and you have to weigh
386
00:27:03.900 --> 00:27:08.900
that incremental increase against the engineering maintenance
costs of your new feature. Thanks,
387
00:27:08.940 --> 00:27:14.220
Emily. I think this will be
very helpful to our listeners. Thank you
388
00:27:14.420 --> 00:27:18.329
for having me on and looking forward
to doing more statistical lessons of the week.
389
00:27:22.049 --> 00:27:27.450
Time to get straight back into our
chat with Ave. he'll fish now.
390
00:27:27.569 --> 00:27:32.240
The other thing you mentioned a couple
of times when talking about, you
391
00:27:32.319 --> 00:27:34.519
know, the major subdisciplines of finance
that data science is having an impact in.
392
00:27:34.759 --> 00:27:38.279
You talked about machine learning and artificial
intelligence. So I was wondering what
393
00:27:38.359 --> 00:27:41.440
you see the role of, you
know, these two, you know coupled
394
00:27:41.640 --> 00:27:45.990
technologies and ways of thinking about modeling
the world, what impact they're actually having
395
00:27:47.029 --> 00:27:49.910
or whether you know, we have
a healthy skepticism as well concerning things that
396
00:27:51.029 --> 00:27:52.549
are buzz terms as well as things
that provide a lot of value. Right.
397
00:27:52.710 --> 00:27:56.950
So how does this supply in find
it? Well, actually getting to
398
00:27:56.430 --> 00:28:00.059
trading side of things, so just
today a book arrive which is called pedant
399
00:28:00.099 --> 00:28:03.579
trading. So I just covered this
in a magazine on the weekend. I
400
00:28:03.579 --> 00:28:04.660
said, well, let's have a
look at this book. Let's pretty nice
401
00:28:04.819 --> 00:28:08.819
and already the names, such as
pattern trading, means trading based on some
402
00:28:10.059 --> 00:28:14.890
patterns or price formations that you see
with regard to financial instrument. So be
403
00:28:15.009 --> 00:28:17.730
it, for example, apple stock, or this can be let's say,
404
00:28:17.769 --> 00:28:21.650
the good prize, or it can
be the Urus all exchange rate, and
405
00:28:22.130 --> 00:28:26.250
the theory goes that when you see
certain patterns in the prices, this might
406
00:28:26.650 --> 00:28:32.400
in one case signal further upwards movement
or in another case downwards movement, or
407
00:28:32.640 --> 00:28:37.400
that the market is most probably to
move sidewards. But again, it's all
408
00:28:37.440 --> 00:28:41.240
based on patterns, and I think
most of the listeners are completely aware of
409
00:28:41.279 --> 00:28:45.470
the fact that machine learning techniques are
pretty good in learning about, let's say,
410
00:28:45.470 --> 00:28:49.029
the value of patterns, first of
first of all and recognizing patterns and
411
00:28:49.150 --> 00:28:53.109
second of all, coming up with
predictions based on pets. Make example from
412
00:28:53.150 --> 00:29:00.619
before spotify, people listen to Song
ABCDF g. Then the fatastic new song
413
00:29:00.779 --> 00:29:06.420
might be something for you because you
have also been listening to Song Abcd of
414
00:29:06.539 --> 00:29:07.940
G. and this is the same
with patents. If you see patterns and
415
00:29:08.019 --> 00:29:11.970
markets and you might say, well, this goes, let's say, up,
416
00:29:11.089 --> 00:29:15.289
down, up, down, then
the machine might learned that with a
417
00:29:15.410 --> 00:29:21.970
higher likelihood it's more probable that the
market goes down afterwards or up afterwards.
418
00:29:21.970 --> 00:29:22.809
So this is done. What the
machines, of course, should do.
419
00:29:22.930 --> 00:29:26.640
They should learn what is happening.
And when I give talks, a typically
420
00:29:26.680 --> 00:29:32.519
have a couple of pictures showing s
patterns and people starting to not as yeah,
421
00:29:32.559 --> 00:29:33.599
I know, I know this one, I know this one and I've
422
00:29:33.680 --> 00:29:37.599
been trading on that one. But
my argument is, and you've been asking
423
00:29:37.599 --> 00:29:41.230
you about how machine learning deep learn
all these steps. My influence marks I'm
424
00:29:41.269 --> 00:29:45.549
saying. Usually I'm not saying that
there is nothing in these patents, nor
425
00:29:45.750 --> 00:29:49.109
that there is something in these pendants. What I'm saying to people is that
426
00:29:49.710 --> 00:29:55.380
if there is something in these pedants, then for sure machines are better and
427
00:29:55.619 --> 00:30:02.460
recognizing these pedants, at learning these
pedants and then at executing trades based on
428
00:30:02.539 --> 00:30:04.900
these parts. Because, I mean, we all know maybe a human being
429
00:30:04.980 --> 00:30:08.490
might learn over the course of this
trading lifetime, I can know twenty,
430
00:30:08.809 --> 00:30:12.609
forty patterns, maybe machine doesn't have
any issue, and learning patterns which are
431
00:30:12.730 --> 00:30:17.089
pretty complex, let's say a hundred
based on a hundred features, for example,
432
00:30:17.250 --> 00:30:22.160
and maybe a hundred fiftyzero relevant patterns
that it immediately recognizes. And of
433
00:30:22.279 --> 00:30:27.240
course, when we are about trading, it's about saising opportunities. The fast
434
00:30:27.279 --> 00:30:30.079
that you can trade on what you
see, the better it usually is,
435
00:30:30.400 --> 00:30:33.920
and the less emotionally you are,
the better it usually is. And this
436
00:30:34.119 --> 00:30:38.029
is, I think, were the
advance, the advantages of the machines are
437
00:30:38.509 --> 00:30:41.789
compared to human beings. I think
we are not yet there. That then,
438
00:30:41.910 --> 00:30:47.829
every single area the machines and the
algorithms will replace the traders. But
439
00:30:48.029 --> 00:30:52.140
there are good examples like golden sucks. You know always read this cool of
440
00:30:52.220 --> 00:30:56.619
using adults are from the economists,
for example, that in two thousand sucks
441
00:30:56.660 --> 00:31:00.619
had the six hundred equities traders on
the single
trading floor and of the six
442
00:31:00.700 --> 00:31:04.410
hundred it just to left. These
days just two people and the ast as
443
00:31:04.410 --> 00:31:10.130
kind of replaced by technology. And
of course technology doesn't build itself. So
444
00:31:11.049 --> 00:31:17.890
let's say to your resources have been
replaced from the training skill to the technological
445
00:31:18.049 --> 00:31:21.000
skill. I mean you need people, of course, who built the systems
446
00:31:21.079 --> 00:31:22.759
and so for. So it's when
people say, well, this job is
447
00:31:22.799 --> 00:31:27.079
about to be replaced by machines,
but there must be people who build the
448
00:31:27.240 --> 00:31:30.839
machines and who built to write the
software that will replace the people. And
449
00:31:30.880 --> 00:31:34.589
there's what we see in the in
the financial markets, I think, in
450
00:31:34.670 --> 00:31:40.549
spectacular fashion, that they are looking
for more and more technologies, data scientists,
451
00:31:40.630 --> 00:31:44.589
programmers that are able to build the
machines that you acquire these days in
452
00:31:44.670 --> 00:31:48.380
order to be successful in markets.
But other stuff that has been done manually
453
00:31:48.619 --> 00:31:52.779
in the recent past is then not
on walk anymore, and even high paying
454
00:31:52.900 --> 00:31:56.940
jobs are suffering in this regard,
as the equities traders that are mentioned here
455
00:31:57.059 --> 00:32:01.329
in the Goldman Sucks em for sure. And actually this this reminds me.
456
00:32:01.529 --> 00:32:05.849
You've got a great two page piece
on your website, which will link to,
457
00:32:06.369 --> 00:32:08.490
which you written, called computational finance. White Python is taking over,
458
00:32:08.690 --> 00:32:13.529
and you actually you quote Robin Wigglesworth
of the Financial Times there. And what
459
00:32:13.609 --> 00:32:16.640
Robin Road is traders used to be
first class citizens of the financial world,
460
00:32:16.640 --> 00:32:20.640
which is exactly what you're just saying
about them, so many thousands being on
461
00:32:20.680 --> 00:32:23.000
the floor, but that's not true
anymore. Robin Robin continues. He writes
462
00:32:23.240 --> 00:32:27.759
technologists are the priority now, and
this was three years ago. So I
463
00:32:27.880 --> 00:32:30.750
wanted what you've seen now in terms
of technologist being the priority, and I
464
00:32:30.829 --> 00:32:36.269
presume by technologists you actually include working
data scientists in that. So I'm wondering
465
00:32:36.309 --> 00:32:40.109
how data science and technology is continuing
to disrupt finance. Yeah, I mean
466
00:32:40.230 --> 00:32:45.460
this is exciting. This subscribes it
and it's already quite a while ago when
467
00:32:45.539 --> 00:32:50.339
this same was made actually, and
what we see now these I mean this
468
00:32:50.460 --> 00:32:53.700
is usually when you have something new
than
people try to rush in one particular
469
00:32:53.779 --> 00:32:57.980
direction. But I think now we
are getting back to a point where I
470
00:32:58.059 --> 00:33:01.930
said, well, we might need
the market savvy people still. You know,
471
00:33:02.009 --> 00:33:06.930
when you hire somebody who is pretty
good at programming but has no experience
472
00:33:07.009 --> 00:33:10.970
in markets, what what you expect
from these people to program into the trading
473
00:33:12.009 --> 00:33:16.400
applications in terms of like risk and
and, yeah, safety measures and so
474
00:33:16.559 --> 00:33:20.960
forth. You know, a little
bit of understanding is simply require in that
475
00:33:21.119 --> 00:33:24.279
sense. And what we see,
and think this will be kind of the
476
00:33:24.440 --> 00:33:30.430
near future at least, is that
these days they try to merge the world's
477
00:33:30.470 --> 00:33:34.470
and the sense of that people doing
let's say simple, or rather, from
478
00:33:34.509 --> 00:33:38.630
a data point of view, simple
equity research. They start using our fantastic
479
00:33:38.670 --> 00:33:44.220
technologies like pandas for data crunching,
a visualization with, let's say, I
480
00:33:44.259 --> 00:33:46.579
didn't know what the couplings and all
these nice things that we use on a
481
00:33:46.619 --> 00:33:52.019
daily basis to become better at their
jobs and maybe, at the same time,
482
00:33:52.140 --> 00:33:55.609
accomplish more or being able to prune
the ever increasing amount of data.
483
00:33:57.009 --> 00:34:00.049
And also the traders. I mean
history was kind of like in a way
484
00:34:00.089 --> 00:34:02.930
that there was one trader and maybe
two people on the left and the right
485
00:34:04.009 --> 00:34:07.889
and side or traders, they were
in real time programming their excel spreadsheet applications
486
00:34:07.969 --> 00:34:12.960
if the trader had a new idea. So these days the trader and people
487
00:34:13.079 --> 00:34:16.320
were responsible there are probably quiet come
up with their own solutions and and use
488
00:34:16.360 --> 00:34:21.480
different techniques than excel spreadsheet, where
some people are sitting on the left and
489
00:34:21.519 --> 00:34:25.110
the right and doing kind of like
real time tweaks while the trader is training.
490
00:34:25.269 --> 00:34:30.829
So this is one example that I
recently retreated from the fortune well,
491
00:34:30.630 --> 00:34:36.469
forget what's reeling at the language City
Group. Once it's incoming investment bank analyst
492
00:34:36.510 --> 00:34:39.260
to know was python. So even
in a feel like investment banking, or,
493
00:34:39.260 --> 00:34:45.380
let's say, people working for consulting
companies, they are these days required
494
00:34:45.019 --> 00:34:50.659
to have some programming knowledge. Just
hasn't been a case before which MNA banker
495
00:34:50.980 --> 00:34:54.769
use python like ten years ago,
fight nobody there, but
these days people
496
00:34:54.769 --> 00:35:00.409
are instead of expected to know a
little bit about excel, they are request
497
00:35:00.449 --> 00:35:05.610
because in every field they are facing
these huge amounts of data and people now
498
00:35:05.730 --> 00:35:09.920
know that it's much more efficient to
crunge these numbers and data by technology such
499
00:35:09.920 --> 00:35:14.639
as Python. They are expected to
know a little bit about programming as well.
500
00:35:14.679 --> 00:35:17.360
And I wouldn't say that everybody these
days should become a software developer,
501
00:35:17.400 --> 00:35:22.510
architect or engineer, but you know, kind of like, with a little
502
00:35:22.510 --> 00:35:27.070
bit of training you can accomplish quite
a lot compared to the traditional approaches in
503
00:35:27.110 --> 00:35:30.550
this regard. So more like emerging
like high breeds. I think I'd be
504
00:35:30.869 --> 00:35:35.110
this kind of trendy word anyways.
So like the hybrid skill set that is
505
00:35:35.190 --> 00:35:38.699
required. Market Knowledge, your background
maybe more banking side, market side,
506
00:35:38.699 --> 00:35:44.260
or ever which department you belong to, but nevertheless to know coding. After
507
00:35:44.340 --> 00:35:46.579
all, it's a little bit like
math. You know math, never heard
508
00:35:47.420 --> 00:35:51.969
and I think to know about programming
and they a science doesn't hurt either these
509
00:35:51.969 --> 00:35:54.090
states. So English, math and
python. These are, I think,
510
00:35:54.130 --> 00:36:00.690
the three most important language that anybody
should master before getting a job exactly.
511
00:36:00.769 --> 00:36:05.719
And that actually reminds me that I've
heard you say, when asked why Python
512
00:36:06.119 --> 00:36:08.360
for finance so much, you've spoken
to the fact that you can see the
513
00:36:08.440 --> 00:36:14.159
python to be the English of the
data and financial world, and I'm wondering
514
00:36:14.239 --> 00:36:16.760
why, why that's the case.
Should mean? Many people say, well,
515
00:36:17.079 --> 00:36:22.110
you know, this language has this
fantastic feature, or Julia's faster or
516
00:36:22.190 --> 00:36:27.309
this and that. I think most
people, and just having talked about the
517
00:36:27.550 --> 00:36:31.750
Investment Bank analyst that are supposed to
learn python, how many language should such
518
00:36:31.789 --> 00:36:37.500
a person in their expect capacity,
should learn? I mean it's hard to
519
00:36:37.739 --> 00:36:44.340
master a single language properly. So, and I'm coming more from a time
520
00:36:44.539 --> 00:36:46.889
constrained, resource constrained point of view, where I said, well, if
521
00:36:46.969 --> 00:36:52.889
you only have time to learn a
single programming language, then it should be
522
00:36:52.170 --> 00:36:58.210
python. If you only have time
to learn one foreign language, spoken language
523
00:36:58.570 --> 00:37:01.679
for almost everybody around the world,
it is these days English and this is
524
00:37:02.239 --> 00:37:06.360
where I see kind of the parallels. To say, well, not too
525
00:37:06.440 --> 00:37:12.079
many people easily learn three four languages, neither spoken language nor programming language.
526
00:37:12.920 --> 00:37:15.469
But why Python? What? What
is it about python? Python? Why
527
00:37:15.590 --> 00:37:19.150
Python, I mean, of course, is a proper one. When I
528
00:37:19.269 --> 00:37:22.989
started back in the days, this
was for me the first top of contact.
529
00:37:22.030 --> 00:37:24.989
I'm stay with the scripting language which, on a high level, allowed
530
00:37:25.429 --> 00:37:30.260
yet fast interactivity. Even back then, even without eye python, you could
531
00:37:30.300 --> 00:37:34.460
do on the mall amazing things and
so for so I wasn't used to that.
532
00:37:34.579 --> 00:37:37.900
So when I crew up I started
actually coding some land, basically on
533
00:37:38.019 --> 00:37:40.139
the code or see sixty four.
You know, this is where I came
534
00:37:40.219 --> 00:37:45.289
from. Then I did see at
university compile cycles and soul force and then,
535
00:37:45.329 --> 00:37:49.570
for first a couple this fantastic scripting
language. And this was on top
536
00:37:49.610 --> 00:37:52.690
of the interactivity was so close to
mathematical language. So when you have a
537
00:37:52.769 --> 00:37:58.719
financial theorem, equation or whatever,
you have you equations there. It's pretty
538
00:37:58.760 --> 00:38:01.639
straightforward and you without that much of
a train, you can translate what the
539
00:38:01.760 --> 00:38:06.880
math, the finance says to Python. This is what got me hooked there.
540
00:38:07.440 --> 00:38:10.119
I think this is not the major
argument these days. Anymore form.
541
00:38:10.199 --> 00:38:14.710
From my point of your python is
kind of the orchestrating language for all the
542
00:38:14.789 --> 00:38:19.389
technologies that you need. I think
it's the best language to use for data
543
00:38:19.469 --> 00:38:22.630
signs. It's a first class citizen
in the air world. So when people
544
00:38:22.750 --> 00:38:25.670
these days speak of tensorflow, they
use python as their interface. We have
545
00:38:25.750 --> 00:38:30.139
the fantastics, I could learn good
and many, many others. I can
546
00:38:30.300 --> 00:38:35.219
even get to some what's comprehensive list
in this regard, and I think this
547
00:38:35.420 --> 00:38:40.059
is what actually differentiates python compared to
all the other competitors, like, let's
548
00:38:40.099 --> 00:38:45.449
start with the pythones and the established
ones in our fiel C plus Plus Java,
549
00:38:45.889 --> 00:38:50.849
or with c Sharp, or with
Julia, which is a typical competitor,
550
00:38:51.010 --> 00:38:54.369
not in terms of numbers, but
in terms of being pretty close with
551
00:38:54.449
--> 00:38:59.159
the guard to the syntax and their
approach and so forth. What the ecosystem
552
00:38:59.239 --> 00:39:01.519
is missing, and I think the
ecosystem is what makes python unique. And
553
00:39:01.760 --> 00:39:07.719
today everybody who's wants to enter the
financial field, I think, on top
554
00:39:07.760 --> 00:39:13.349
of all my arguments, which might
be subjective, is simply almost every institution
555
00:39:13.789 --> 00:39:17.389
has chosen python as their core language. So if you have some kind of
556
00:39:17.590 --> 00:39:22.269
career, the tack of the air. Then our field, python is simply
557
00:39:22.349 --> 00:39:27.179
a good thing to have on your
CV because you might prefer some other,
558
00:39:27.500 --> 00:39:31.820
more exotic, may be faster or
whatever language, but if your potential employer
559
00:39:31.900 --> 00:39:37.889
doesn't use this language, probably you
won't get too many plus points on your
560
00:39:38.170 --> 00:39:43.130
evaluation afterwards. So this is more
the career aspect, that if you know
561
00:39:43.250 --> 00:39:46.409
python you can work in many,
many fields and many companies these days.
562
00:39:46.769 --> 00:39:52.170
So exactly having python on your CV
and resume is incredibly important. But also
563
00:39:52.329 --> 00:39:57.880
now having python in your github repositories
also a huge step in, you know,
564
00:39:57.960 --> 00:40:00.079
the interview process as well, right
when we're applying for jobs in finance.
565
00:40:00.119 --> 00:40:05.079
Sure, I mean there's this one
of the fantastic aspects of living in
566
00:40:05.280 --> 00:40:08.150
open source age that you really can
showcase what you've been working on. Of
567
00:40:08.230 --> 00:40:13.949
course, most people having kind of
a professional job, they are usually not
568
00:40:14.070 --> 00:40:17.590
allowed them to talk about what they're
doing. The financial industry is really executive
569
00:40:17.630 --> 00:40:23.420
in this regards, and this is
why the industry loses many fantastic people,
570
00:40:23.500 --> 00:40:29.099
we must say, to the more
open companies. But due to regulation,
571
00:40:29.179 --> 00:40:32.860
legal and all these things, people
are hardly ever allowed to really talk about
572
00:40:32.980 --> 00:40:37.449
what they are doing. But of
course you can do stuff on the side
573
00:40:37.570 --> 00:40:43.769
and you can build your let's say, programming or data science ceasily on platforms
574
00:40:43.849 --> 00:40:47.409
such as get up, and if
you want to pull the reputation beyond what
575
00:40:47.530 --> 00:40:51.760
you do at work, there's of
course, task platform. On the other
576
00:40:51.840 --> 00:40:53.920
hand, of course, these days, these outside effects, there are some
577
00:40:54.079 --> 00:41:01.320
companies specialized crawling pages like Github or
platforms like get up and looking for people,
578
00:41:02.239 --> 00:41:07.269
so replacing maybe
the surge on Linkedin, or why are other mazs for
579
00:41:07.510 --> 00:41:10.670
talent in this regard? So every
once in a while I get approached there
580
00:41:10.710 --> 00:41:15.230
as well. If I would be
interested in changing jobs. Maybe they should
581
00:41:15.469 --> 00:41:17.510
improve their research. When they would
have seen that I'm running my own company,
582
00:41:17.550 --> 00:41:22.500
maybe they wouldn't even ask so.
But for other people this mnded it
583
00:41:22.780 --> 00:41:27.300
represent opportunity to present themselves, to
showcase what they can do and also to
584
00:41:27.420 --> 00:41:30.539
learn, to get feedback from the
community. They're working on something of interest
585
00:41:30.579 --> 00:41:35.650
to others. They might even win
over contributors or get feedback. And Yeah,
586
00:41:35.769 --> 00:41:38.969
it's maybe some fam even the community, for sure, and something we've
587
00:41:39.010 --> 00:41:43.969
really been talking around in the emergence
of Python as as such an important tool
588
00:41:43.969 --> 00:41:46.480
in finance and data signs in the
use of Github and Jupiter notebooks. We've
589
00:41:46.480 --> 00:41:51.519
actually seen a huge shift in the
past several decades from the use across the
590
00:41:51.679 --> 00:41:54.159
board, not only in finance,
when academic research in all types of quantitative
591
00:41:54.199 --> 00:42:00.000
computational disciplines and moved from proprietary tools
to open source tools. Right. Yes,
592
00:42:00.039 --> 00:42:02.070
sure, and I mean this is
also one of the benefits. If
593
00:42:02.070 --> 00:42:07.989
you have a huish community behind it, I think hardly any commercial company can
594
00:42:07.030 --> 00:42:10.389
keep out, because such community to
an effort. If you have millions of
595
00:42:10.469 --> 00:42:14.909
users of Jupiter notebooks, and not
only for Python, of course. We
596
00:42:14.949 --> 00:42:19.739
have multiple other kernels that can use
like our kernels or junior kernels and use
597
00:42:19.780 --> 00:42:23.860
the same environment. This is fantastic
and the project has been growing tremendously.
598
00:42:23.940 --> 00:42:28.579
I'm really happy about it. Also
for US providing content, as that we
599
00:42:28.699 --> 00:42:31.530
are more or less a content creation
company, as a fantastic means of grading
600
00:42:31.610 --> 00:42:36.809
content and of sharing it. For
example, my currently riding on a second
601
00:42:36.849 --> 00:42:39.809
edition of my Python for Finance Book. It's basically all Jupiter notebook based old
602
00:42:39.889 --> 00:42:43.730
codes such to be the notebook,
and I have one of my books list
603
00:42:43.769 --> 00:42:45.360
of what a tity Vara is.
The Roots was actually written to a hundred
604
00:42:45.360 --> 00:42:49.840
percent and it to be the notebook, and then I programmed a little workflow
605
00:42:49.920 --> 00:42:53.480
behind it, translated it into a
lottach and finally, a while they published
606
00:42:53.639 --> 00:42:57.800
what I got all started and written
down. And you be the notebook.
607
00:42:57.880 --> 00:43:00.190
So many, many things that you
can do also with you. God what,
608
00:43:00.309 --> 00:43:05.389
I'm a big fan of reproducible research. So I was growing up doing
609
00:43:05.510 --> 00:43:09.190
my research, you just said,
back then, mostly printed research papers and
610
00:43:09.269 --> 00:43:13.820
people presented just the results, but
you never could somehow get to the point
611
00:43:13.860 --> 00:43:15.139
as at well, but they that
have, they were working with how to
612
00:43:15.420 --> 00:43:20.619
crunch the data. Are these results
of their present reliable? And so for
613
00:43:20.659 --> 00:43:24.900
wars and today, with access to
many data sources and providing for something who
614
00:43:24.900 --> 00:43:28.570
should be the notebook, not only
with text but with the code that crunches
615
00:43:28.610 --> 00:43:32.090
the data and presents the results,
it's fantastic when you're after reproducible research,
616
00:43:32.369 --> 00:43:36.849
again, which I'm a big fan
of. Exactly. So, for someone
617
00:43:36.929 --> 00:43:39.570
who wanted to get started in financial
datasciens and these, they such things we've
618
00:43:39.570 --> 00:43:44.079
been talking about. What are the
main types of skills and technologies that they
619
00:43:44.199 --> 00:43:46.280
would need to know to work at
the intersection of finance and data signs.
620
00:43:46.719 --> 00:43:52.400
Yeah, I mean Python for sure. I mean they are may if you
621
00:43:52.480 --> 00:43:55.590
believe in kind of like polls and
overviews and surveys and so for people typically
622
00:43:55.630 --> 00:44:01.309
say is that the most common combination
is python, are and sequel. So
623
00:44:02.070 --> 00:44:07.030
you need somehow, some databases.
I think we have multiple other options these
624
00:44:07.070 --> 00:44:09.340
days, like a g five.
We don't need a sequel, but Python
625
00:44:09.539 --> 00:44:14.699
for sure. Maybe some are here
and there. If you don't find let's
626
00:44:14.699 --> 00:44:19.300
say, statistic package in the python
UN worse, but this has hardly about
627
00:44:19.340 --> 00:44:22.739
in the case for what we have
been doing for ourselves and also for clients
628
00:44:22.780 --> 00:44:29.050
and of course, machine learning.
I in general to know about the basics
629
00:44:29.130 --> 00:44:32.809
of statistics. You know about the
algorithms. What unsuper warts learning is super
630
00:44:32.849 --> 00:44:36.929
warts learning, all the techniques.
They are like to put back the machines
631
00:44:36.969 --> 00:44:40.920
of just a repression gus in in
you've based whatever jumps to mind. I
632
00:44:42.360 --> 00:44:46.559
I'm pretty sure you don't need to
be an expertinary single area of this field.
633
00:44:46.679 --> 00:44:51.360
So knowing all
the the theories more
like from an apply perspective, to
634
00:44:51.440 --> 00:44:54.869
know what is there and to know
what to apply when given certain data sets.
635
00:44:55.230 --> 00:45:00.269
Then actually the points which I usually
summarize on the basics. So we
636
00:45:00.389 --> 00:45:06.750
have, example, in our certificate
programs, the largest longest running online training
637
00:45:06.789 --> 00:45:10.460
program that we have a complete area
tools and skills where we teach basic tools
638
00:45:10.539 --> 00:45:15.460
and skills that people, from our
point of view, to know at least
639
00:45:15.500 --> 00:45:20.300
a little bit about, like using
editors, processes, kind of setting up
640
00:45:20.300 --> 00:45:23.250
environments, deploying stuff at the cloud, working a little bit with Dacca containers.
641
00:45:23.530 --> 00:45:27.489
Of course, I learned the topics
that I've mentioned right now. They
642
00:45:27.769 --> 00:45:34.610
probably require a complete study over years, but to know the basics typically help
643
00:45:34.730 --> 00:45:37.480
you, pretty much like basics of
Linux, if you command line tools,
644
00:45:37.599 --> 00:45:43.639
some death tools, python, packaging, publishing, testing, all these topics
645
00:45:43.719 --> 00:45:47.639
an soft engineering and basics in general. Not being an expert. Maybe after
646
00:45:47.679 --> 00:45:52.710
that, but I think first ten
percent helpe you already with the teenty percent
647
00:45:52.789 --> 00:45:55.590
of your of your problems. Then
about data storage. Working with data is
648
00:45:55.630 --> 00:45:59.989
important, of course, but they
could in the financial field, thinking of
649
00:46:00.150 --> 00:46:04.110
trading, if you implement back testing
programs, you'll be able to fast work
650
00:46:04.190 --> 00:46:07.860
with huge amounts of data, to
crunge the data, to store it correctly,
651
00:46:07.980 --> 00:46:12.659
to story your results and all these
things, but also, for example,
652
00:46:12.820 --> 00:46:16.420
to work with streaming data which is
actually pretty important in all case.
653
00:46:16.500 --> 00:46:21.690
Not then every area you have the
need to process data and really turn but
654
00:46:22.289 --> 00:46:25.449
in finance generally there is a case
that, for example, in trading,
655
00:46:25.530 --> 00:46:28.969
when you do I'll go trading,
you need to be able to digest tick
656
00:46:29.050 --> 00:46:31.610
data, streaming data, in real
time, to crunch it, to maybe
657
00:46:31.730 --> 00:46:36.159
resemble it, to come up with
your signals in real time based on your
658
00:46:36.239 --> 00:46:39.480
train models, and to act in
real time. So this is, from
659
00:46:39.559 --> 00:46:45.360
experience, something that people at least
having our time getting started with. But
660
00:46:45.639 --> 00:46:49.869
it's simply require, because otherwise,
how do you want to automate things with
661
00:46:49.989 -->
00:46:52.030
the guard to Algo trading, if
you don't know how to work with sockets
662
00:46:52.070 --> 00:46:58.269
and streaming and maybe also even streaming
visualization, for example, yet to implement
663
00:46:58.349 --> 00:47:00.860
your trading strategies? To keep with
this examples, but I think these are
664
00:47:01.300 --> 00:47:06.059
kind of like, from my point
of view, the skills and technologies that
665
00:47:06.139 --> 00:47:08.980
people should know about. It it
great, that's very useful, and I
666
00:47:09.139 --> 00:47:13.179
think it isn't as so you just
need to learn all of these things straight
667
00:47:13.219 --> 00:47:15.130
away, right. I mean essentially
you can on them on the go in
668
00:47:15.289 --> 00:47:20.130
a project based way. They're sure. This is what we usually tell people
669
00:47:20.130 --> 00:47:22.329
as well that we so well,
we provide you with a basic overview,
670
00:47:23.050 --> 00:47:27.170
maybe skip some details in the beginning, so with the gut to our program
671
00:47:27.210 --> 00:47:30.360
with let's say, a twelve week
structured schedule, so to say, where
672
00:47:30.360 --> 00:47:34.880
we try to cover the basis.
But afterwards it's more about practical things,
673
00:47:35.039 --> 00:47:37.679
practical modules, and people are supposed, for example, to do a final
674
00:47:37.800 --> 00:47:43.159
project and there they can select a
topic, but they are then expected to
675
00:47:43.239 --> 00:47:46.110
apply the different things, and this
is from experience where really learn about the
676
00:47:46.150 --> 00:47:51.750
stuff. Maybe you know about it
after the former education part, but you'll
677
00:47:51.789 --> 00:47:54.309
learn about it once you get started
applying it. And this is what people
678
00:47:54.550 --> 00:47:59.579
write me back as well. I
couldn't even imagine that one day I would
679
00:47:59.820 --> 00:48:05.619
sitting here spinning up cloud instances and
deploying Python Code and the cloud and doing
680
00:48:05.739 --> 00:48:08.820
remote monitoring based on socket streaming and
so forth. Now it feeds like second
681
00:48:08.860 --> 00:48:14.050
nature, but I'm pretty sure when
these people first time saw what this is
682
00:48:14.090 --> 00:48:16.090
all about, it Oh, I
will never master that. So this is,
683
00:48:16.210 --> 00:48:20.849
of course, I think this is
kind of nature. This reaction would
684
00:48:21.489 --> 00:48:24.449
of course, to something that interests
you. When you have a purpose,
685
00:48:24.610 --> 00:48:30.039
this is when you get started learning
about stuff and learning to master it exactly,
686
00:48:30.239 --> 00:48:32.639
and you've half answered this, but
in my next ration really was where
687
00:48:32.679 --> 00:48:37.519
would you suggest beginners who want to
work in financial data science? Look,
688
00:48:37.000 --> 00:48:40.469
yeah, of course, I mean
this is I can hardly say anything else,
689
00:48:40.510 --> 00:48:45.030
because there's at the core for our
business that we provide online trainings,
690
00:48:45.190 --> 00:48:50.510
also life trainings, events and the
form of Wood Cams in London and York
691
00:48:51.030 --> 00:48:53.389
usually. But if that look at
our pages, we have kind of a
692
00:48:53.710 --> 00:48:58.619
broad offering which has been growing over
years and it is of course, influence
693
00:48:58.739 --> 00:49:00.380
or what we have been done over
more than ten years. I've been working
694
00:49:00.420 --> 00:49:05.699
for the biggest hatch funds, for
big BANGS and other financial institutions in this
695
00:49:05.860 --> 00:49:09.809
regard. But we know what is
kind of expected and this is under one
696
00:49:09.849 --> 00:49:15.409
of our subpages, training to TPQ, Todoh and the certificate that I mentioned
697
00:49:15.449 --> 00:49:19.650
before. This is really our flagship
offering where we have a sixteen week program
698
00:49:20.329 --> 00:49:23.719
after which we hope that you're able
to use python for I go trading or
699
00:49:23.880 --> 00:49:28.639
for other financial things, but I
go trading is the focus indeed, and
700
00:49:28.800 --> 00:49:34.159
we even are able to award a
German university certificate because we are co operating
701
00:49:34.199 --> 00:49:39.030
here with all local university and if
you are doing a master program within Europe,
702
00:49:39.070 --> 00:49:44.630
this is seemed good. For the
five cts points. European credit transfer
703
00:49:44.750 --> 00:49:49.510
system might be interesting, then only, of course, for like current students
704
00:49:49.590 --> 00:49:53.380
of future students in this regelt.
But it's a former certificate from university for
705
00:49:53.619 --> 00:49:58.980
Python and I go trading and or
finance, depending on you. Fantastic,
706
00:49:59.019 --> 00:50:01.179
and we'll make sure to link to
those in the show notes as well.
707
00:50:01.739 --> 00:50:06.820
So Great. So in general,
what does the future of data science in
708
00:50:06.900 --> 00:50:09.570
finance look like to you? Yeah, I mean this is, from my
709
00:50:09.650 --> 00:50:15.130
point of view, an easy one. They designs will become a core discipline
710
00:50:15.250 --> 00:50:19.929
or finance. I said it before. We come from an area where the
711
00:50:20.130 --> 00:50:25.719
brains and math equations have been driving
finance, and this regard, data triven
712
00:50:25.800 --> 00:50:30.039
finance is what we'll replace it.
So we might lose quite a bit of
713
00:50:30.360 --> 00:50:36.429
beauty and financial equations and modeling and
so forth, but on the other hand,
714
00:50:36.550 --> 00:50:39.150
we might get back what I like
to call the scientific method in this
715
00:50:39.349 --> 00:50:44.389
context, that we start with the
data, we have a deep look at
716
00:50:44.429 --> 00:50:47.630
the data in any area, every
area that we have in the financial
industry,
717
00:50:49.070 --> 00:50:52.539
and apply the new algorithms and that
sense, I think finance will be
718
00:50:52.699 --> 00:51:00.019
more driven by stuff that is developed
outside of our industry than ever before,
719
00:51:00.260 --> 00:51:05.090
so meaning that of course these nice
theories that are still around and are applied
720
00:51:05.329 --> 00:51:09.090
and some of them successful, others
not that much. They usually came from
721
00:51:09.090 --> 00:51:15.889
finance, finance professors or finance practitioners, but now people start using stuff maybe
722
00:51:16.050 --> 00:51:22.719
developed within let's say Google, and
major point was to have a good algorithm
723
00:51:22.880 --> 00:51:27.480
to play go or to build a
self driving car. So completely different background.
724
00:51:27.519 --> 00:51:31.639
But in that sense the background can
easily be changed to a financial background
725
00:51:31.639 --> 00:51:37.389
with some adjustments and the disregard.
I'm happy that we have, since Mars
726
00:51:37.550 --> 00:51:42.909
this year, the first proper book
about financial machine learning, because many algorithms
727
00:51:42.949 --> 00:51:46.780
can easily be transferred. But we
have some specifics to consider when we apply
728
00:51:46.940 --> 00:51:51.179
the algorhythms with the guard to the
data, how we crunch data, how
729
00:51:51.219 --> 00:51:53.539
we manage the data, what a
kind of special things. When we look
730
00:51:53.539 --> 00:51:58.860
at the financial time series in contrast
to let's say a physical time series or
731
00:51:58.940 --> 00:52:02.289
whatever kind, and that sends,
it will become place, not something special,
732
00:52:02.489 --> 00:52:06.289
it will become a core thing,
as you say. Well, what
733
00:52:06.409 --> 00:52:10.329
do you think of technology in the
banking industry? Many people say, banks
734
00:52:10.369 --> 00:52:19.039
are essentially technology companies and I think
the data driven AI first finance future is
735
00:52:19.119 --> 00:52:22.719
not too far away from my point
of view. So my final question is,
736
00:52:22.000 --> 00:52:24.440
do you have a final call to
action for all the listeners out there?
737
00:52:24.480 --> 00:52:29.679
Yeah, sure, I mean focusing
again on the Python for finance side.
738
00:52:29.760 --> 00:52:32.949
And it started some let's say three
years ago, when people each other
739
00:52:32.949 --> 00:52:36.389
and said, well, I'm interested
in machine and you want apply it in
740
00:52:36.469 --> 00:52:38.309
finance. I, having this and
that background, want to make a move.
741
00:52:38.590 --> 00:52:42.469
And today, when people try,
still today, I must say,
742
00:52:42.510 --> 00:52:45.940
when people try to get into the
field and to apply data signs and machine
743
00:52:45.980 --> 00:52:51.820
learning and try to profit, be
it within corporate contacts or as a retail,
744
00:52:51.900 --> 00:52:55.219
I'll go trader, let's say,
trading their own cash precisions, there
745
00:52:55.900 --> 00:53:00.050
it's still hard to get up and
running. What I can only recommend you
746
00:53:00.170 --> 00:53:04.090
to experience, and since we are
doing it, is to look for a
747
00:53:04.090 --> 00:53:09.530
kind of an opera, appropriate integrated
training program you have that many things for
748
00:53:09.690 --> 00:53:15.400
free? Are they of that many
university based things that might be remotely related
749
00:53:15.400 --> 00:53:20.760
to what you're looking for, but
many people told me and confirmed that they
750
00:53:20.880 --> 00:53:23.960
have wasted months and months trying to
look for that stuff. On their own.
751
00:53:24.039 --> 00:53:28.829
So we have done a living out
of putting this all together into integrated
752
00:53:28.869 --> 00:53:32.389
training programs. Documentations, of example
that our program you get one hundred page
753
00:53:32.389 --> 00:53:37.829
documentation only about Python for finance,
and I got trading. So it's quite
754
00:53:37.829 --> 00:53:40.739
a bit. And once you have
found something like this is suddenly to have
755
00:53:42.460 --> 00:53:45.179
to do with what we do.
Not Everybody's interested in I go trading,
756
00:53:45.300 --> 00:53:47.300
but something like this, where I
said, well, this is a good
757
00:53:47.340 --> 00:53:51.019
starting point, this provides me with
the product of you, but also with
758
00:53:51.099 --> 00:53:52.780
the details that I need to get
started. Then you should focus on the
759
00:53:52.860 --> 00:53:57.889
basics. And this is where many
people in our program I wouldn't say really
760
00:53:57.969 --> 00:54:00.650
complain, but where I said,
well, I'm having a really hard time,
761
00:54:00.849 --> 00:54:05.130
and it's with marthing the basics.
So, for example, for the
762
00:54:05.210 --> 00:54:09.199
first time people maybe just having a
windows background, having used the regular tools
763
00:54:09.239 --> 00:54:14.119
there and so forth, setting up
a drop let moving around in the Linux
764
00:54:14.159 --> 00:54:19.639
environments of it might see kind of
a really difficult thing. But I mentioned
765
00:54:19.679 --> 00:54:22.559
it before some people after a few
weeks at well, now it feels like
766
00:54:22.679 --> 00:54:25.469
second natures. This means you can
get to the point where you're where you
767
00:54:25.550 --> 00:54:30.750
know how to use him wire as
as age access to a cloud instance and
768
00:54:30.550 --> 00:54:35.309
you deploy your Automated Code and work
with sockets and whatever. So master the
769
00:54:35.349 --> 00:54:38.099
basics, have a look at the
nice tools that are there in the Linux
770
00:54:38.179 --> 00:54:43.179
world and master the processes there.
And I think then, once you have
771
00:54:43.900 --> 00:54:46.739
get around the raining, have the
basics ready, being more in the financede.
772
00:54:46.780 -->
00:54:51.539
More on the programming side, you
should get started with implementing many little
773
00:54:51.739 --> 00:54:54.929
project so that interest you. There're
so well, this is what we typically
774
00:54:55.010 --> 00:54:59.489
do. A straining samples that you
might say, well, I want to
775
00:54:59.570 --> 00:55:04.090
have a little APP where I simply
put in, let's say, a stock
776
00:55:04.170 --> 00:55:08.800
symbol and my flask APP that I
host on digitalotion troublet then shows me a
777
00:55:08.960 --> 00:55:13.679
craft with some simple moving average.
So for somebody who knows what this is,
778
00:55:14.119 --> 00:55:16.159
this might be just takes less than
an hour if you know what you
779
00:55:16.239 --> 00:55:20.039
do. But if you get started
with these things, this might take like
780
00:55:20.159 --> 00:55:22.670
a week or two. But to
have these projects where you have indeed,
781
00:55:22.710 --> 00:55:28.869
at the end, result that you
might even showcase your colleagues friends, that
782
00:55:28.989 --> 00:55:30.949
you might put on Github or whatever. This is, I think, where
783
00:55:31.110 --> 00:55:36.619
the learning curve gets really steep and
towards the end, I'm pretty sure everybody
784
00:55:36.659 --> 00:55:39.539
who wants to enter this field and
wants to learn about all these fantastic technologies
785
00:55:39.699 --> 00:55:44.059
and the challenges in the field.
They might have something in their head like
786
00:55:44.300 --> 00:55:50.010
building their own algorithmic training operation or
coming up with their own derivatives pricer or
787
00:55:50.170 --> 00:55:53.570
coming up with the machine learning application
for put full your management, for example.
788
00:55:53.690 --> 00:55:58.329
Then they should after that. They
should come up with such a projects,
789
00:55:58.409 --> 00:56:02.050
Cope it specified and get started and
then work over weeks or even months
790
00:56:02.250 --> 00:56:06.800
and will something huge. I think
this is done where your where you collect
791
00:56:06.960 --> 00:56:09.400
on the way all the pieces that
might have been missing before because you have
792
00:56:09.559 --> 00:56:14.559
been really focusing on a few things. But once you get to the point
793
00:56:14.599 --> 00:56:17.079
of where your master such a Mute
Project, I think you can say,
794
00:56:17.199 --> 00:56:21.869
well, now I'm I'm proficient and
work in the industry, I'll get my
795
00:56:22.349 --> 00:56:25.630
major project and up and running.
Fantastic. So to recap, to select
796
00:56:25.670 --> 00:56:30.550
an appropriate integrated training program then master
the basics, then get your hands dirty
797
00:56:30.590 --> 00:56:35.059
with as many little python projects as
possible and then do a larger, more
798
00:56:35.099 --> 00:56:38.780
challenging project. Perfect, fantastic.
Eve, it has been such a pleasure
799
00:56:38.820 --> 00:56:42.659
having you on the show. Thanks
for me, it's pleasure as well.
800
00:56:46.019 --> 00:56:50.849
Thanks for joining our conversation with Eve. We saw that one of the main
801
00:56:50.929 --> 00:56:54.130
drivers for the adoption of python and
open source software and finance has been the
802
00:56:54.210 --> 00:57:00.210
amount of tools available and in constant
development, from numpire to Pandus and beyond.
803
00:57:00.809 --> 00:57:04.280
E've made clear that all the big
banks and funds are now using such
804
00:57:04.400 --> 00:57:07.760
tools. On top of this,
we saw that banks are essentially technology companies
805
00:57:07.880 --> 00:57:13.440
and that the data driven, ai
first finance future is approaching quickly. With
806
00:57:13.559 --> 00:57:16.429
the masses of data, computational power
and tools we now have, along with
807
00:57:16.469 --> 00:57:22.389
a pattern recognition capacities of machine learning, there is less need for mathematical modeling
808
00:57:22.389 --> 00:57:25.269
and more for Data Science Pipelines.
And for all you learners out there,
809
00:57:25.469 --> 00:57:30.420
if you're interested in this type of
work, select an appropriate integrated training program
810
00:57:30.739 --> 00:57:35.820
then master the basics, then get
your hands dirty with as many little python
811
00:57:35.900 --> 00:57:39.420
projects as possible and then do a
larger, more challenging project. Also,
812
00:57:39.500 --> 00:57:44.059
make sure to check out our next
episode, a conversation with Christy Ballet,
813
00:57:44.329 --> 00:57:49.329
Assistant Professor at Kent State University,
About Data Science, ecology and the adoption
814
00:57:49.369 --> 00:57:52.730
of techniques such as machine learning in
academic research. What are the biggest challenges
815
00:57:52.809 --> 00:57:57.409
in ecology that data science can help
to solve. What does the intersection of
816
00:57:57.530 --> 00:58:01.280
open science and data science look like
in scientific research? What is happening at
817
00:58:01.320 --> 00:58:06.280
the interface between data science and machine
learning methods, which are pattern based,
818
00:58:06.480 --> 00:58:10.760
and traditional research methods, which are
classically hypotheticod deductive? Is there a paradigm
819
00:58:10.800 --> 00:58:15.190
shift occurring here? Join US next
week to find out. Now, don't
820
00:58:15.230 --> 00:58:19.710
forget that, for those interested,
we've also got a special offer this week
821
00:58:19.710 --> 00:58:23.469
for data framed listeners the opportunity to
try data camp yourself. All you need
822
00:58:23.510 --> 00:58:29.500
to do is email sales at data
Campcom, that sales at data campcom,
823
00:58:29.900 --> 00:58:34.460
with the subject line podcast and redeem
your free two week trial. I'm your
824
00:58:34.579 --> 00:58:37.099
host, Hugo bound Anderson. You
can follow me on twitter, that Hugo
825
00:58:37.179 --> 00:58:42.010
bound and data camp. At data
camp, you can find all our episodes
826
00:58:42.130 --> 00:58:45.730
and show notes at data Campcom.
Community podcast.