The Machine Learning Revolution

Published on
September 10th, 2018
Topic
Technology, Financial System, Sentiment
Duration
25 minutes

The Machine Learning Revolution

Discoveries ·
Featuring Dr. Dario Villani

Published on: September 10th, 2018 • Duration: 25 minutes • Topic: Technology, Financial System, Sentiment

Dr. Dario Villani, CEO of Duality Group and the former global head of portfolio strategy and risk at Tudor Investment, explains the major hurdles and opportunities companies face when implementing machine learning. While machine learning is the backbone of many of the world’s biggest projects, from self-driving cars to the development of virtual assistants, Dario explains why it has yet to be successfully applied to the world of finance and investing. Filmed on June 19, 2018 in New York.

Comments

  • DC
    D C.
    9 February 2019 @ 03:06
    Amazingly insightful. One of the best conceptual exposition of machine learning and its application. Most others I have heard don't make a convincing case of why and how machine learning or AI will evolve in finance, or seem muddled with their analogies and presumptions. Dr. V's explanations just made intuitive sense to this seemingly non-intuitive subject. And that to be is gold.
  • JX
    Jack X.
    2 February 2019 @ 20:28
    This is an great interview. But it is again an optimistic one. The multi-dimension and contemplative things that machine learning will unlock will only come after an world-wide suffering and job decimation.
  • AE
    Alex E.
    16 September 2018 @ 23:15
    Seems to me that machines will take a very long time to learn how humans behave in finance and markets. While they may have predictive capabilities, I'm not aware of any machine programs that use Chaos Theory to predict illogical and irrational actions as fomented by human traders. As was demonstrated by the death of a female cyclist struck by a Tesla in auto-pilot, the human driver was given less than several seconds to respond to what the car thought was an anomaly. Don't see insurance rates going down for machine drivers either...
  • VS
    Victor S. | Contributor
    12 September 2018 @ 19:47
    Now this was a “thinking piece”... Dario is a superior thinker indeed!!! One of the best you have ever shown. Thank you..
  • CK
    Charles K.
    12 September 2018 @ 12:51
    Tyz
  • ET
    Eduard T.
    12 September 2018 @ 07:35
    The speaker, Dr. Villani, should discuss some of the technical aspects of cutting edge machine learning. It would be nice to hear his thoughts on applications of techniques such as, (deep) reinforcement learning/Monte Carlo tree search and Bayesian neural networks to developing a algorithms akin to DeepMind's AlgoGo for quantitative trading portfolio management. And also would be interesting to know what kind of time series techniques his firm, Duality Group, is developing and some uses case examples. The video is good intent but lacks technical depth.
    • RI
      R I.
      13 September 2018 @ 01:11
      Completely agree, Eduardo.
  • AC
    Andrew C.
    12 September 2018 @ 04:41
    Short the horse; daily noise versus signal in market movements; and others. On top of good information on machine learning, so much more in here to take away about general investing. Thanks Dr Dario and RVTV!
  • RN
    Robert N.
    11 September 2018 @ 20:48
    As an aside on how far you can take a model approach in the world of horse racing, this story from a while back is worth remembering: https://www.bloomberg.com/news/features/2018-05-03/the-gambler-who-cracked-the-horse-racing-code
  • JM
    Jan M.
    11 September 2018 @ 20:09
    Great stuff
  • RK
    Robert K.
    11 September 2018 @ 14:59
    Nice interview. The machines are competing with humans for "jobs" since the 18th century. It happens in bursts and the next brings probably more stress than the previous. Remember the Luddites (https://en.wikipedia.org/wiki/Luddite). The marginal cost of low value added menial (what this means is continuously redefined) tasks is zero.
  • bf
    bart f.
    11 September 2018 @ 12:31
    So to avoid an implosion of our system we need to move into a new societal structure where humans and machine do not compete as machines are much more efficient. But humans will not derive wealth from labour anymore (which is kind of already the case) so will capital be fairly shared between this “above the machine cast”? On top humans in their current state are linear, 2 dimension beings how will that fit with the quantum multiverse which machines will be evolving in? I am of limited intellect to find the answers but somehow I do see this ending with a bang.
    • CM
      Carl M.
      12 September 2018 @ 04:41
      ...or a whisper so quiet?
  • RO
    Robert O.
    11 September 2018 @ 04:20
    When computers were first starting to get used by the general public, the question was how to get more done with this new tool. The question that later became more relevant was how to prevent bad actors from creating chaos or disrupting productive activities. I can imagine that once self-driving cars reach a critical mass with relatively uniform sensor and decision making technology that bad actors will again emerge with small inexpensive devices that could confuse these cars, similar to recent ransomware attacks. In the end the same advancing technology that can make cars safer could also be used to make cars more dangerous. This reminds me of an old quote concerning freedom and security; giving up the first does not guarentee the second.
  • CC
    Christopher C.
    11 September 2018 @ 02:10
    Great interview. The time will come very shortly where insurance for human driven cars is more expensive than that for self-driving cars. When systems which are progressing with geometric gains in power or efficiency by the time to you get to 2% done you are halfway home.
    • DS
      David S.
      11 September 2018 @ 22:39
      The insurance may be the lowest for self-driving cars with a driver ready to take the controls. DLS
    • PJ
      Peter J.
      13 September 2018 @ 07:54
      Yes, but will there also become mandatory driverless control of vehicles further down the road, especially in cities to 'control' traffic flow?
    • DC
      D C.
      9 February 2019 @ 03:03
      People should really watch the history of aviation (or more precisely, the accidents of aviation over the last few decades) In the 70's there were approx 2,000 accidents per year and now it's approx 50. At the beginning, there were a lot of mechanical and engineering design problems (much like the "O" Ring for Challenger), and proportionally speaking fewer pilot error. Then somewhere along the line, that proportion started to shift, when they refined all the electronic and mechanical system of the planes, and pilot error became more and more prevalent. (Pilots and ATC trying to override built in systems have led to disasters). So this will likely be the case for self driving cars. I can almost bet that the insurance rate for self driving cars alone will be much much lower than human drivers. As for DLS's comment if a ready to take control driver along with self driving algos would be the lowest, I would think we have to define who is that driver. If it's the average driver, than still the refined algo is still way better. (Again, we see that the top top top pilots did in some cases avert disaster when systems malfunctioned.) But from my daily commute, I can assure you that the average driver is no Lewis Hamilton.
  • MA
    Mike A.
    10 September 2018 @ 21:25
    I like his optimism on machine learning but it seems to me that at this early stage, the limited successes of ML have been on problems with a massive amount of data that a human can do very easily. I am also not sure many of the discovered relationships machine learning has unearthed in some of these problems isn't just coincidence and not proven out over time to be a true discovery. There is a lot of promising work being done and it's another tool to try but whether it is a massive breakthrough or just a better statistical method when given a massive amount of data is inconclusive.
  • CD
    Cheryl D.
    10 September 2018 @ 19:58
    Excellent, really enjoyed that video. The autonomous vehicle space will be really interesting.
  • DS
    David S.
    10 September 2018 @ 19:51
    Great discussion. I am ready for self-driving cars with someone ready at the stirring wheel like a pilot on an airplane with autopilot. I really would not like to be a passenger on a plane without a pilot. I would also like a driver to step in and drive in case of the car being hacked, or other non-normal circumstances. DLS
  • DS
    David S.
    10 September 2018 @ 19:40
    Extraordinary introduction on machine learning. I am sure Mr. Villani has explained machine learning to the uninformed many times. His use of graphics, metaphor and personal experience really helped me understand his positions. Mr. Villani and RealVision are doing a great job. Thanks. DLS
  • SS
    Sam S.
    10 September 2018 @ 18:29
    The Jetsons, Lost In Space, 2001 Space Odyssey ---it's coming true and certainly has taken a lot of time. Now time is being compressed with the tools of technology changing so fast. Will we be better off in the journey of life? Can't fight it, so let's profit from it and stay engaged. Well done.
  • Sv
    Sid v.
    10 September 2018 @ 17:50
    Nice presentation. One comment for a Non-native English speaker is that adaptation and evolution are not the same word and have different meanings.
  • NG
    Nick G.
    10 September 2018 @ 15:16
    "The horse" is in fact the macro guy who is not data dependent. The outcomes are changing before his very eyes but he continues to shout "It will all end badly" into his ever diminishing account value.
    • my
      markettaker y.
      27 September 2018 @ 00:55
      Mark Dow, is that you?
  • V!
    Volatimothy !.
    10 September 2018 @ 14:05
    Humans creating machines to solve problems they can't solve. So isn't he saying short humans? Makes absolutely no sense to me.
  • TY
    Tyler Y.
    10 September 2018 @ 12:55
    You had me at, "There is no model. There are relationships that are elusive and they evolve over time. Very much like us evolving through evolution." Excellent walkthrough!
    • MH
      Michael H.
      10 September 2018 @ 16:30
      One point I'd suggest is important to understand is the tradeoff between the parametrization of a model and the volume of data needed to identify "truth." Traditional statistical models (more specifically, parametric models) impose relationships between the variables. Those asserted relationships, derived from theory, allow an analyst to get an answer from only a little bit of data. That's a good outcome if your assertions are correct (in other words, your theory is reliable). It's a bad outcome if not. Machine learning algorithms in which there is "no model" (more specifically, non-parametric models, or no theory) are free from any imposed constraints on the relationships between variables. However, people don't appreciate just how much data is needed to compensate for that freedom. It's vastly more than most people understand. If you pipe too little data into a machine learning model, the software will still spit out some answer. It's just likely to be very wrong. The fields in which machine learning has a promising future are those that generate nearly-unimaginable amounts of data and the existing theory is bunk. In those situations, traditional statistics don't offer much value, so why not trust the black box? However, in fields in which data production is relatively slow (like medicine) or the existing theory is relatively robust, then the value of machine learning is limited. Personally, I'm not informed enough to have an opinion as to whether finance is a good candidate to be meaningfully disrupted by machine learning strategies or not. At first blush it seems obvious that it should be, but I could imagine counter arguments, too.
    • PC
      Philip C.
      11 September 2018 @ 00:21
      Often, the difference between the machine learning approach on one hand, and traditional knowledge engingeering, or even statistical methods, boils down to whether you need the system to explain its reasoning. With machine learning, the system learns directly from data and typically cannot explain to a human user why it reached a particular decision. Knowledge based systems use a representation of knowledge, often formal logic or probability theory, that can be used to yield an explanation. So the suitability of machine learning in finance would boil down to the question: would you be happy to let your machine make trading decisions that you could not fathom and would you let it run even if it made trades that make no sense to you?
  • YB
    Yuriy B.
    10 September 2018 @ 10:56
    Bravo Real Vision! A masterpiece interview.