Modeling Future Returns by Distance from the 52-Week High

Summary

The document poses a research question about whether a stock’s price relative to its 52-week high can help explain subsequent returns. The proposed setup uses the price-to-high ratio as a predictor and a forward return, such as the following week’s return, as the response in a linear regression. The author is concerned that the predictor is not normally distributed and asks whether it should be transformed or otherwise normalized.

The question also asks whether a different method would better reveal the relationship between proximity to a yearly high and future performance. No answers, empirical results, or recommended transformations are included, so the text does not establish that the ratio predicts returns or that normality of the predictor is required for the proposed regression. Any investigation would also need to define the return horizon and evaluate the relationship on data not used to fit the model, while accounting for the time-series and cross-sectional structure of stock observations.

Key ideas

• The proposed predictor is a stock’s price divided by its 52-week high.
• The target is a future stock return, with the following week offered as an example horizon.
• The author asks whether a non-normal predictor needs transformation for linear regression.
• The document provides no answer or evidence that proximity to a 52-week high predicts returns.
• A useful study would specify the horizon and assess the relationship out of sample.

Tags

# How To Regress Returns Vs Price as Pct of 52 week high?


# How To Regress Returns Vs Price as Pct of 52 week high?












I would like to do a linear regression of daily stock price returns, vs the price as a percentage of the 52 week high.

i.e. [next week return] = A * [Price / 52 Week High ] + B

where A and B are constants.

[Price / 52 Week High] will not be normally distributed, so the previous regression will not be very valid.

How can I normalise it to make it more valid?

Is there a better way for me to see how [Price / 52 Week High] affects the future return?

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.