Skip to content
All library documents

Testing an Event-Driven Strategy with Regression Dummies

Article Quant Q&A · Author: JOHN
Read the original

Summary

The note proposes a regression-based way to evaluate a hypothesis that companies with positive momentum and a refinancing corporate action will increase in value. The suggested setup uses a broad sample of companies and a binary indicator that equals one when a company meets both conditions and zero otherwise. The indicator’s estimated coefficient represents the associated effect in the regression, while its t-statistic gives a measure of statistical significance.

This is a basic research design, not a complete backtest or causal analysis. The document does not specify the dependent variable, sample period, controls, treatment of overlapping events, or how momentum is measured. A significant coefficient alone would not establish tradable profitability: implementation costs, selection effects, and out-of-sample performance would also matter. The proposal is therefore a starting point for testing the event-and-momentum association, with important design choices left open.

Key ideas

  • Use a large company sample to test the association between refinancing events, momentum, and subsequent value changes.
  • Create a binary indicator that marks companies meeting both the momentum and refinancing criteria.
  • Interpret the indicator coefficient as the estimated magnitude of the association in the chosen regression.
  • Use the t-statistic to assess statistical significance, while recognizing it does not establish profitability or causation.

Tags

Event-drivenMomentumStatisticsBacktesting
Full text
# how to measure a event driven strategy?


# how to measure a event driven strategy?












How to measure an event driven investment strategy? Say I have a strategy which I assume that if a firm has positive momentum and it has a refinance corporate action, it's value will increase. For strategies like this? How do I measure it?

## Answer by horseless (score 2)

https://quant.stackexchange.com/a/25637

I would do regression analysis with a dummy variable. Take a large sample of companies, and add a 0 - 1 dummy variable where that variable is equal to 1 if it meets the momentum and refi criteria, and 0 otherwise. The coefficient will indicate the magnitude and the t-statistic the significance.

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.