Small Predictive Correlations, Kelly Betting, and Active Management

Summary

The discussion asks how small correlations might support reliable financial gains, while observing that strong correlations alone do not guarantee a dependable strategy. One answer links the idea to the Kelly criterion, which sizes bets according to expected returns and risk. Another distinguishes correlations between two securities, which may motivate relative-value trading, from correlations between a security and its own future return.

The latter is described as an information coefficient and connected to the Fundamental Law of Active Management. This frames small predictive relationships as potentially useful when applied systematically across opportunities, though the document supplies no derivation, trading rules, empirical evidence, or risk estimates. Correlation by itself does not establish causation or profitability; a practical strategy would need to account for forecast accuracy, costs, position sizing, and changing relationships. The answers point to concepts rather than presenting a tested method.

Key ideas

• The discussion associates exploiting small predictive relationships with Kelly-based position sizing.
• Correlation between securities may suggest relative-value trading, but the answers do not show that weak correlations alone are profitable.
• Correlation between a security and its future return is called an information coefficient in the response.
• The Fundamental Law of Active Management is cited as a framework for understanding predictive information.
• The document offers conceptual pointers rather than a tested strategy or empirical evidence.

Tags

# Harnessing small correlations for reliable profit


# Harnessing small correlations for reliable profit












It is said that Edward O. Thorp was able to harness small correlations for reliable financial gain. I've seen some strategies based on strong correlations which did not seem particularly reliable. Does anyone have an idea or a paper to how the small correlation exploitation could work? I can only think of some kind of relative value arbitrage.

## Answer by dkhokhlov (score 2, accepted)

https://quant.stackexchange.com/a/10889

it is based on Kelly criterion. mentioned in one of stackexchange posts here .

## Answer by uday (score 0)

https://quant.stackexchange.com/a/10938

Correlations between what?

Correlations between stock A and another stock B - relative value arbitrage - not sure if small correlations will help here.

Correlations between stock A and its future stock return Ra - its called Information Coefficient. Try Fundamental Law of Active Management and many similar web info on Fundamental Law for more information.

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.