Interpreting Near-Zero CAPM Betas in Second-Level Stock Data

Summary

The document describes a research question about CAPM estimates from second-level quote data for US-listed stocks, using SPY returns as the market benchmark. The researchers report that some stocks have beta estimates close to zero, suggesting little measured co-movement with the chosen market return at that sampling frequency. They wonder whether stock price increments and market movements of different sizes could contribute to the result, and ask how to adjust the data.

No answer, diagnostic, or adjustment method is provided, and the referenced screenshot is not included in the text. The proposed explanation is therefore only a hypothesis, not an established finding. The document offers no regression outputs, data-cleaning details, or comparison across sampling intervals, so it cannot establish whether the small estimates reflect genuine exposure or measurement and sampling effects. Its value is as a statement of an empirical issue for researchers examining intraday beta estimates.

Key ideas

• The research estimates CAPM betas from second-level stock and market returns.
• SPY returns serve as the market benchmark in the described analysis.
• Some estimated betas are reported to be close to zero.
• The suggested role of differing price increments is a hypothesis without supporting analysis in the document.

Tags

# CAPM Beta zero-correlation performance issue


# CAPM Beta zero-correlation performance issue












I am working on a research project that requires me to run a CAPM regression on all intra-day stock quotes in NSDAQ, NYSE and all other U.S. exchanges since 1993.

The precision of the quote data is at second-level. We calculated the second by second stock returns, market returns, risk free rate, and ran a CAPM regression on the data. We used SPY return as the market benchmark.

However, an interesting problem that occurred was - there are some stocks with really small beta (very close to 0). That is, the stock barely moves with the market.

While pondering about the reasons behind this, we think it might be caused by different minimum stock movements. Some stocks might be trading at \$1, but a market movement of, say, \$10, will overshoot the price movement of such stocks.

However, we are still not exactly sure about the reasons and how to adjust the data properly.

The follow screenshot is an example of how small the beta can be (estimate is the beta value).

Does anyone have an idea on this?

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.