Our October 2026 factor run tests every pre-IPO fundamental we hold against forward returns, across IPOs priced in the last 12 months. Only factors that survive Benjamini-Hochberg correction for multiple testing are reported as robust — with dozens of variables tested at once, a plain significance threshold would surface noise.

3M forward returns

Measured across IPOs priced in the last 12 months. The table ranks factors by the absolute size of their Information Coefficient — the rank correlation between the factor and the return that followed.

VariablenICpQ5−Q1 spreadRobust
Stock-based comp / revenue %20-0.6090.003-67.107✓
Interest coverage (EBIT / interest)300.598<0.001123.683✓
Direct listing (0/1)128-0.465<0.001-64.843✓
SG&A / revenue %35-0.4240.024-48.274
EBITDA margin %730.3970.00672.706✓
Op. cash flow / revenue %450.3760.03443.082
Direct listing < $1000M cap (0/1)127-0.376<0.001-50.013✓
Operating margin %730.3720.00671.282✓
Strongest pre-IPO factors vs 3M return (IPOs priced in the last 12 months)

Multivariate fit (3M)

Cross-validated Elastic Net (n=127, α=14.40, CV R²=0.008), standardized coefficients:

VariableCoefficient
Direct listing (0/1)-1.9326
Direct listing < $1000M cap (0/1)-1.4172
Net margin %1.0430
Deal-structure flag (0-1)0.8301
Return on assets %0.7678
Quick ratio0.5814

6M forward returns

Measured across IPOs priced in the last 12 months. The table ranks factors by the absolute size of their Information Coefficient — the rank correlation between the factor and the return that followed.

VariablenICpQ5−Q1 spreadRobust
Interest coverage (EBIT / interest)280.649<0.001384.470✓
Beneish M-score (manipulation)240.5100.058313.112
SG&A growth % (YoY)270.4620.132271.103
Cash runway (months)350.4250.078205.108
Op. cash flow / revenue %420.3930.045132.594
Free cash flow / revenue %420.3690.102132.594
UW syndicate: mid-market / quality bank (0/1)710.3600.006143.466
Operating cash flow yield %570.3550.029154.203
Strongest pre-IPO factors vs 6M return (IPOs priced in the last 12 months)

Multivariate fit (6M)

Elastic Net shrank every coefficient to zero.

What this does not prove

These are historical associations across a sector-neutralized cross-section, not a trading rule. Sample sizes are small by quantitative-research standards, the IPO universe turns over quickly, and a factor that clears the significance bar in one month can fail it the next.