'A Data Science Approach to Managing Crowd-Sourced Systematic Trading Strategies” by Dr. Jess Stauth

This talk, titled, "Statistical Algorithm Selection: A Data Science Approach to Managing Systematic Trading Strategies Developed by "The Crowd"" was given by Dr. Jessica Stauth at QuantCon NYC 2018.

The field of quantitative finance has experienced rapid adoption of machine learning techniques at nearly every stage of the typical workflow, e.g. new signal identification, signal combination, portfolio optimization, and trading.

We explore how the techniques of machine learning might be applied to a unique new problem: identifying trading strategies which are likely to produce "alpha" in a real market setting from a pool of millions of backtests created by Quantopian's 180,000+ member online community. We will review challenges including:
- Compiling our (imbalanced) data set of simulation results
- Designing features based on transactions, returns, holdings, and more?
- Defining a goal (what should we select for?)
- Facing the problems of non-stationarity, overfitting, and interaction terms

Dr. Jessica Stauth is former Managing Director of Portfolio Management, Research, and Trading at Quantopian, where she and her team were in charge of selecting the algorithms from the Quantopian community. Dr. Stauth holds a Ph.D. in Biophysics from UC Berkeley, where her research focused on computational neuroscience.

Disclaimer
Quantopian provides this presentation to help people write trading algorithms - it is not intended to provide investment advice.

More specifically, the material is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory or other services by Quantopian.

In addition, the content neither constitutes investment advice nor offers any opinion with respect to the suitability of any security or any specific investment. Quantopian makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change and may have become unreliable for various reasons, including changes in market conditions or economic circumstances.

[viridianhawk]

The correlation-causation disconnect seems nowhere more apparent than in how Quantopian views turnover. High turnover strategies of course tend to perform worse because they are penalized more often by the slippage model. It seems to me that Quantopian takes this to mean that they should generally avoid high turnover strategies, including (seemingly erroneously) strategies that do perform well and survive the cost model. I don't see how they can draw that conclusion. To me that seems like mistaking correlation with causation. Any help understanding this would be appreciated. Is there any evidence that high-turnover strategies are more likely to exhibit degraded OOS performance over low-turnover strategies?

[tuathadedanann]

There are so many problems with this approach I don't even know where to begin and it clearly won't even out perform buy and hold with a bear market filter.