The correlation between factors decides which stack pays and which one double counts.
Variance shrinks fastest when the two screens disagree. Value names go cheap and ignored. Momentum names go expensive and crowded. The two cohorts rarely overlap. Stacking them earns the Markowitz cross term. Quality and low-vol both pick boring defensive names. Stacking those is double counting one bet.
Pair correlation matrix . illustrative calm-regime rolling 60M cohort returns
Value
Momentum
Quality
Size
Low Vol
Value
1.00
-0.30
+0.15
+0.45
-0.10
Momentum
-0.30
1.00
+0.10
+0.05
-0.40
Quality
+0.15
+0.10
1.00
-0.15
+0.55
Size
+0.45
+0.05
-0.15
1.00
-0.10
Low Vol
-0.10
-0.40
+0.55
-0.10
1.00
Negative . diversifying pair
Positive . overlapping bet
The pair that did most of the lifting
-0.30
Value plus Momentum. Cheap-and-ignored versus expensive-and-trending almost never own the same names. The portfolio variance shrinks where it counts. This is the canonical 50 year old stack from Asness Moskowitz Pedersen 2013.
The pair that double counts
+0.55
Quality plus Low Volatility. Both screens land on staples, FMCG, pharma, large defensives. Stacking them is buying the same boring stocks twice through two different screens. Adding Low-Vol on top of Quality is what dragged the 5-factor Sharpe below the 3-factor.
Portfolio variance is governed by the cross term, two times w1 w2 sigma1 sigma2 rho. When rho is negative the cross term shrinks the variance. When rho is positive the cross term inflates it. Negative-rho factor pairs are the only stacks that earn the variance reduction. Positive-rho factor pairs are paying twice to own one bet.
Correlation values illustrative, calm-regime, rolling 60M cohort returns on Nifty 200 universe. Crisis-regime correlations compress higher across the board (see correlation cliff post 10 May 2026 . equity-gold rho moved 0.10 calm to 0.55 crisis on the same logic). Citations . Fama French 1992 + 1993 + 2015 + Asness Moskowitz Pedersen 2013 Journal of Finance "Value and Momentum Everywhere" + Asness Frazzini Pedersen 2019 Review of Accounting Studies "Quality Minus Junk" + Harvey Liu Zhu 2016 Review of Financial Studies "and the Cross Section of Expected Returns". Past performance . backtest only . not a guarantee.