SmallCap Dislocation: Why This Mean-Reversion Edge Survived 18 Attempts to Break It

---

Most quantitative strategies look good right up until you touch the parameters.

Change a threshold.
Add a constraint.
Remove a filter.

The edge disappears.

That’s the failure mode we were explicitly trying to avoid while building SmallCap Dislocation—a mean-reversion strategy for NSE small caps designed to operate alongside a trend-following portfolio.

What follows isn’t a performance summary.
It’s an explanation of what happened after we ran 18 sensitivity tests, removed logic, tightened constraints, and forced the strategy to defend itself.

---

The Structural Premise

In NSE small caps, price moves are often driven less by information and more by liquidity and capital flows.

When selling is discretionary, prices move slowly.
When selling is forced—risk limits, redemptions, leverage unwinds—prices overshoot.

That overshoot doesn’t require fundamentals to improve in order to reverse.
It only requires sellers to finish selling.

SmallCap Dislocation is designed to detect that state.

Not by predicting bottoms.
By identifying when forced selling has likely exhausted and downside asymmetry improves.

---

How We Detect Dislocations (Briefly)

A stock becomes eligible only if:

• Its 5-day return falls into the worst tail of the universe
• Selling pressure fails to continue after the initial shock
• Liquidity and governance filters pass
• Portfolio-level constraints allow entry

The mechanics are simple by design.
The robustness comes from what survived testing.

---

Sensitivity Test #1: How Extreme Is “Extreme”?

We tested percentile thresholds on 5-day returns:

3%, 5%, 7%, 10%, 15%

Only one region consistently worked.

At ~7%:

• Average P&L peaked at +1.43% per trade
• Sharpe reached 1.14
• Win rate stabilized around 60%
• Drawdowns were lowest (−7.4%)

Below 5%, results diluted quickly.
Above 10%, performance decayed again.

This wasn’t a smooth curve. It was a structural breakpoint.

Around the 7th percentile sits the zone where:

• Liquidity thins
• Non-discretionary sellers dominate
• Price moves are no longer information-efficient

That’s the regime the strategy needs.

The takeaway wasn’t “7% is magic.”
It was that percentile-based detection isolates seller behavior better than absolute thresholds.

---

Sensitivity Test #2: Single-Day Gap Shocks Added Nothing

We expected large one-day gaps with volume to matter.

So we tested:

• Gap sizes from −5% to −12%
• Volume filters from 1.5× to 3×
• 16 combinations in total

Every configuration produced:

• The same 340 trades
• The same 60.6% win rate
• The same +1.43% average P&L

The gap logic never triggered independently.

In retrospect, this was obvious.

In small caps, forced selling rarely resolves in one session.
It unfolds across days as liquidity disappears gradually.

Multi-day percentile shocks already captured the entire process.

So we removed the gap logic.

Performance stayed the same.
The strategy got simpler.
Fragility went down.

That’s the kind of outcome you hope for in sensitivity testing.

---

Sensitivity Test #3: Sector Limits Were the Real Risk Control

Dislocations cluster.

Stress doesn’t hit one NBFC, one chemical name, or one IT stock in isolation.
It hits funding channels, margins, or earnings expectations across a group.

We tested sector exposure caps at:

5%, 7%, 10%, 15%

The results were unambiguous.

• 5%: blocked high-quality opportunities, edge collapsed
• 15%: over-concentration dominated returns, drawdowns rose
• 7–10%: captured dislocation clusters without turning into sector bets

The optimal region allowed two positions per sector, no more.

This was the single most important portfolio-level parameter in the system.

The lesson is subtle but critical:

Dislocation alpha requires diversification—but not forced diversification.

---

What 18 Sensitivity Tests Actually Revealed

After pushing thresholds, removing logic, and tightening constraints, three patterns stood out:

1. Performance degraded gradually, not abruptly
2. Removing filters didn’t kill the edge
3. Portfolio constraints mattered more than signal tweaks

That’s the opposite of curve-fitting.

The behavior was consistent with a strategy anchored in market mechanics, not parameter luck.

---

Why This Strategy Works as an Overlay

SmallCap Dislocation is not designed to run fully invested.

It activates selectively:

• During stress
• When forced selling creates asymmetry
• When other strategies are already cautious

When deployed as an overlay using idle cash alongside a trend-following small-cap portfolio, the combined system:

• Improved CAGR by +4.3 percentage points
• Maintained drawdown
• Improved Calmar by ~25%
• Increased capital efficiency from ~67% to ~90%+

That improvement didn’t come from leverage or prediction.
It came from using capital when others can’t.

---

The Broader Takeaway for Fund Managers

Robust strategies aren’t defined by how well they perform at one setting.

They’re defined by how slowly they break when assumptions change.

SmallCap Dislocation survived:

• Threshold changes
• Logic removal
• Constraint tightening
• Portfolio integration

Because it’s aligned with how small-cap liquidity actually breaks—and recovers.

In markets like these, durability beats cleverness every time.