Choosing a Benchmark for Active Strategy Alpha Tests Across Asset Classes

Summary

This discussion asks whether a crypto strategy’s statistically significant intercept from a regression on established equity factors provides a fair measure of skill. It frames the issue as benchmark choice: an alpha is measured relative to the factors included, so a stock-market factor model may not capture the return drivers or distributional features of crypto. The replies suggest that researchers consider a model based on crypto-market characteristics when evaluating crypto strategies.

The answers are brief and do not establish a standard crypto factor set, offer a formal model, or provide empirical comparisons. One response asserts that crypto correlations with financial markets are high, while another cautions that stock factors omit crypto-specific return, variance, and skew characteristics. These are competing considerations rather than demonstrated findings in the document. The practical lesson is to justify the benchmark, examine relevant asset-specific exposures, and present inference as conditional on the chosen model; significance alone does not prove that a strategy earns robust, risk-adjusted alpha.

Key ideas

• Regression alpha is defined relative to the factors used as benchmarks.
• An equity factor model may omit important crypto-market exposures.
• Researchers should consider whether asset-specific factors are needed for a fair test.
• The brief replies provide suggestions rather than empirical validation or a settled benchmark framework.
• A statistically significant intercept is conditional on model specification and does not alone establish strategy skill.

Tags

# Testing the significance of active trading strategies other than stocks


# Testing the significance of active trading strategies other than stocks












In active asset management industry, a common approach to Test whether my Strategy Provides significant alpha is to Regress Portfolio Returns on Fama French 3 (or 5 factors) and check whether the alpha is significant.

I Developed a Strategy for cryptos wirh significant alpha at the 5% level when regressed on Fama French 3 factor loadings, i am just wondering is that test 'fair'? Can i keep the results or am i supposed to test it against some 'cryptro factors'? There arent many which are as established as the Fama French ones, on the other hand i definetely need some statistical validation as it is for some scientific paper. The same problem of course arises for all other asset classes such as bonds too.

## Answer by Trader2B (score 1)

https://quant.stackexchange.com/a/71605

These days the correlation between crypto and the financial market is approximating 1 anyway so you should be fine I think.

## Answer by Vitomir (score 0)

https://quant.stackexchange.com/a/46381

The alpha you got makes absolutely sense but compared to the Fama French market, i.e. stock market. It does not take into consideration the characteristics of the crypto market (return, variance, skew, etc.). Therefore, I would build another fair value model based on crypto market.

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.