Do Cheap Stocks Improve Our Strategy? — Eight Low-PBR Combination Experiments
'Cheap stocks win in the end' — low PBR is one of the oldest quant factors. We combined this proven factor into our regime strategy eight different ways. All eight were rejected. This is the data on why, and what it tells us.
"Stocks cheap relative to book value win in the end." Low PBR (price-to-book) is one of the oldest classic factors in quant investing. So would mixing this proven factor into our regime strategy (REGIME_C) make it better? We combined it eight ways, and all eight were rejected. This is that record.
Starting point: both candidates, bare
First, side by side. (2015–2026, 0.35% round-trip cost, survivorship-adjusted.)
| Strategy | CAGR | Max drawdown | Sharpe |
|---|---|---|---|
| Our regime strategy (REGIME_C) | 30.3% | −24.7% | 1.15 |
| Bottom-20% low PBR (standalone) | 13.5–17.4% | −54.5% | 0.85–0.95 |
| KOSPI (benchmark) | 10.8% | −43.9% | 0.65 |
Standalone low PBR beats the KOSPI. The factor itself has positive value. The problem is what happens when you combine it.
Group 1: replace stocks with low-PBR by regime (V-A/B/C)
The most direct idea: in certain regimes, buy low-PBR stocks instead of ours.
| Method | CAGR | Max DD | Sharpe |
|---|---|---|---|
| V-A (regime switch × low PBR) | 18.9% | −32.6% | 1.12 |
| V-B (sideways cash → low PBR) | 29.0% | −46.1% | 1.06 |
| V-C (bull-market stock blend) | 18.3% | −34.1% | 0.83 |
All three were worse than the original — return shaved, drawdown deeper. The cause was clear: our strategy's strength comes from bull-market momentum, and replacing that with low PBR cut off our own upside engine.
Rejected.
Group 2: use low PBR for bear defense (V-D1/D2)
So leave the bull market alone and swap only the bear-market defense to low PBR? The intuition: cheap stocks should fall less in a downturn.
| Method | CAGR | Max DD | Sharpe |
|---|---|---|---|
| V-D1 (bear → low-PBR swap) | 28.6% | −29.0% | 1.11 |
| V-D2 (bear 50:50 blend) | 29.5% | −25.3% | 1.13 |
Close to the original, but still short. Digging into the data gave the reason. Our bear-market defense selects stocks where retail and institutional flows concentrate, and that defense was stronger than low PBR. In pure down-legs, the original beat the value factor.
Rejected.
Group 3: split the capital, not the stocks (W90/80/70)
Rather than swapping stocks, allocate a slice of capital (10/20/30%) to low PBR — a diversification approach.
| Method | CAGR | Max DD | Sharpe |
|---|---|---|---|
| W90 (90:10) | 29.2% | −24.3% | 1.19 |
| W80 (80:20) | 27.7% | −27.3% | 1.20 |
| W70 (70:30) | 26.2% | −30.2% | 1.22 |
Something interesting happened here. Sharpe (risk-adjusted return) rose slightly above the original (1.15). The diversification benefit was real. But it fell short of our pre-registered adoption bar (drawdown must improve by 2+ percentage points). W90's actual drawdown improvement was only 0.4 points.
Decisively, low PBR's crisis fragility showed through. In the March 2020 COVID plunge, the larger the low-PBR weight, the deeper the fall (W70 −21.5% vs original −14.0%). A factor added to help with defense became a liability in the actual crisis.
Rejected.
Conclusion: not because the factor is bad, but because we were already stronger
All eight experiments pointed the same way.
| Group | Method | Decisive failure |
|---|---|---|
| 1 (V-A/B/C) | bull-market swap | cut off the momentum engine |
| 2 (V-D1/D2) | bear-market swap | flow-based defense beats low PBR |
| 3 (W90/80/70) | capital diversification | tiny diversification gain, crisis exposure |
Low PBR is not a bad factor. It has clear positive value and is nearly independent (orthogonal) from our momentum and flow signals. Yet the combination failed for a simple reason: our strategy was already stronger than low PBR on both the bull and bear axes. A factor that is orthogonal but inferior gives no gain when mixed in.
Stepping back, there is a paradoxical comfort here. That even a proven classic factor cannot improve on it is, flipped around, evidence of how solid our strategy is. With this result we fully closed the low-PBR combination research line — and we record all eight failures rather than hide them.
Figures here are from 2015–2026 backtests, reflecting 0.35% round-trip cost and survivorship adjustment. Everything is an observation about our system's past behavior — not a recommendation to buy or sell any security, factor, or asset. Backtest results do not guarantee future returns.