Tampering with Universes: Correction and Improvement
In our previous post regarding a flexible universe selection, we presented a more flexible universe selection model. The results we obtained exhibited unusual beta values, and we thought to have found new avenues of research for certain universe slices. We had to check publications such as:
- "Trading Volume and Beta Stability"
- "Symmetric and Asymmetric Market Betas and Downside Risk"
- "Tail Risks, Investment Horizons, and Asset Prices"
We could not find evidence to support that volume slices (as a proxy of size factor) could deliver negative or low beta values in any market condition. The fact is that there is an issue with the automatic calculation of beta in Quantconnect while using low-resolution data. The solution is to increase the resolution to hourly and make some minor modifications to our base algorithm to obtain similar portfolio positions. By doing this, the high volume and low volume universes yield the following equity curves:
And the beta values are respectively 1.068 and 0.54, values much more in line with what is expected by common sense. Note that we have not limited the minimum volume in this low-vol case, and replicating the strategy may well be impossible due to the unavailability of stock.
The strange beta values are now at least fixed, and no easy low-beta strategy has been found with simple cross-sections. It is worth double-checking by research every suspicious result even before deploying a paper account trial algorithm. 3-4 months of erroneous paper trials can be saved with an additional week of research. Here is the corrected wrapping algorithm for the flexible universe slices at hourly resolution:
Information in ostirion.net does not constitute financial advice; we do not hold positions in any of the companies or assets that we mention in our posts at the time of posting. If you require quantitative model development, deployment, verification, or validation, do not hesitate and contact us. We will also be glad to help you with your machine learning or artificial intelligence challenges when applied to asset management, trading, or risk evaluations.