400 % returns in 1.5 years trading a NASDAQ 100 stock per day (part 1)

Quantonomy Econophysics algorithms

A little history

Online Portfolio Selection

Predicting the orbit of a comet from past data is a deterministic process with relatively small random noise.
Performance of different OLPS algorithms. In our experience, they seem to produce strong results for relatively artificial data (no slippage, no fees) but they need to be highly modified to perform well in real markets.

Selecting best performer stock

Sorting of NASDAQ 100 stocks in terms of one of our performance metrics. Low-value positions are winning stocks. A clear trend is visible when compared with a random sorting of the same stocks.

Walk forward Optimization

Walk-forward design of training set and test set data.
Final cumulative gain graph and statistical results for one of our algorithms after walk-forward testing using third-party software [8].

Statistical Properties of the algorithms

Distribution of the daily returns from the previously described algorithm. The best fit is the heavy-tailed distribution t-location-scale.
A 1000 simulation of 1.5 years of trading with a system with similar distribution to one of Quantonomy’s algorithms; the x-axis is trading days and the y-axis is the final cumulative gain
Final gain of 1000 simulation of trading with Quantonomy algorithms. While there is a wide range of final results, the most likely result is about 5.91x (red straight line). None was below 0.
The behavior of simulated cumulative gain curves vs real cumulative curve. The y axis is cumulative PL and the x-axis is trading days.
Distribution of the final cumulative gains for a 1000 run Monte Carlo simulation.

AlphaHub

An example of the signals produced by the Quantonomy algorithms. They are available to subscribed members on the AlphaHub trading platform. The large jump at the beginning of June is not a glitch, but a large win associated with AAL stock that recovered dramatically after being heavily oversold due to the COVID crisis. The algo cleverly picked this good opportunity. Such large gains are actually relatively common in our trading system given the heavy tail of the distribution of the returns (on the right side of the distribution, so large gains are more common than large losses).
The GUI for the AlphaHub Trader, a fully automated app, that executes using the Alpaca Markets broker the signals from different Quantonomy algorithms.

Conclusion of Part 1

References

Physicist, neuroscientist, financial analyst. CEO and Director of Research at Quantonomy: https://www.quantonomy.fund/giovanni-santostasi-phd