Why is the standard t > 1.96 threshold too low for factors?

With 300+ published factors and data mining, the probability of finding t > 1.96 by chance is very high. Harvey et al. estimate 300+ factors implies a threshold of t > 3.0-4.0 for genuine discovery.

What is the cross-sectional replication rate for published factors?

McLean & Pontiff (2016) found that returns of published anomalies decline by about 26% post-publication and 58% out-of-sample, consistent with data mining and investor learning.

Factor Zoo T-Statistic Significance Threshold Calculator

Determine statistically appropriate t-statistic threshold for factor discovery given multiple testing. Harvey, Liu, and Zhu (2016) recommend t > 3.0 for first discovery; replication requires t > 2.0.

Principal ($)

Annual Rate (%)

Years

Compounding / Year

$ Finance Disclaimer: These calculators are for educational and illustrative purposes only. Results do not constitute financial advice. Consult a licensed financial advisor before making investment or borrowing decisions.

How to Use

Enter the values for the scenario you are assessing. Determine statistically appropriate t-statistic threshold for factor discovery given multiple testing. Harvey, Liu, and Zhu (2016) recommend t > 3.0 for first discovery; replication requires t > 2.0. Use the factor zoo t-stat result to inform your calculation.

Related Calculators

Bonferroni Correction

Finance Ultra-depth

False Discovery Rate

Finance Ultra-depth

Probabilistic Sharpe

Finance Ultra-depth

Factor Zoo T-Statistic Significance Threshold Calculator

How to Use

Related Calculators

Frequently Asked Questions

Why is the standard t > 1.96 threshold too low for factors?

What is the cross-sectional replication rate for published factors?