Insights

Correlation Between Short Sleeves

Alphanume Team · February 28, 2026

Why event sleeves co-move and how to manage it.

Event-driven short sleeves are not as uncorrelated as their distinct signal definitions would suggest. They share common-factor exposures — small-cap factor, risk-on/risk-off regime, borrow-market liquidity, retail sentiment — that produce meaningful pairwise correlations. Understanding the correlation structure is essential to interpreting per-sleeve performance and to constructing the multi-sleeve portfolio correctly.

The empirical correlation structure

Typical observed pairwise correlations between sleeve return streams:

	Discrete	ATM	De-SPAC	Lock-up	Toxic
Discrete offerings	1.00	0.5	0.4	0.5	0.4
ATM activations	0.5	1.00	0.45	0.45	0.45
De-SPAC cohort	0.4	0.45	1.00	0.65	0.35
Lock-up expirations	0.5	0.45	0.65	1.00	0.35
Structured financing	0.4	0.45	0.35	0.35	1.00

The numbers are approximate and vary by time period. The pattern: substantial but not extreme correlations across sleeves; highest between sleeves with overlapping cohorts (de-SPAC ↔ lock-up); lower with structurally distinct sleeves (e.g., structured-financing vs others).

Where the correlation comes from

Common-factor sources:

1. Small-cap factor. Most short-side targets are small or micro-cap. Small-cap broad-market moves affect all sleeves jointly. Roughly 0.2-0.3 of correlation comes from this single factor.

2. Risk-on / risk-off regime. All short sleeves perform worse in strong rising markets and better in declining markets. Roughly 0.1-0.2 of correlation.

3. Borrow-market liquidity. When borrow becomes scarce across the market, all sleeves face higher costs and constrained capacity. Affects each sleeve similarly.

4. Retail sentiment. Sleeves with significant retail-flow exposure (de-SPAC, lock-up, structured-financing) move together when retail flow surges or recedes.

5. Cohort overlap. Some names appear in multiple sleeves simultaneously (de-SPAC + ATM, lock-up + structured-financing).

What the correlation means for risk

The portfolio-level volatility is higher than equal-weight diversification of uncorrelated sleeves would suggest:

Five fully uncorrelated sleeves would produce portfolio volatility ~45% of individual sleeve volatility.
Five sleeves with ~0.45 average correlation produce portfolio volatility ~65% of individual sleeve volatility.

The diversification works, but less than uncorrelated assumption.

Managing the correlation

Several practical responses:

1. Reduce gross exposure during high-correlation periods. When sleeve correlations rise (typically during sharp risk-on rallies), broader gross exposure compounds the loss. Reducing gross during these periods limits damage.

2. Add explicitly different sleeves. Combining event-driven shorts with strategies that are mechanically different (statistical arbitrage, volatility selling, etc.) produces meaningfully lower correlation.

3. Hedge common factors. If small-cap factor exposure drives much of the correlation, long index positions or factor hedges can reduce it.

4. Diversify by market cap and sector. Within each sleeve, ensure positions span multiple sectors and size brackets. Reduces idiosyncratic correlation.

The reporting implication

Per-sleeve performance attribution loses meaning if sleeves are highly correlated. The marginal contribution of any single sleeve to portfolio risk is bounded by its average correlation. The reporting:

Per-sleeve attribution: useful but bounded.
Marginal-risk attribution: more accurate for risk-budgeting purposes.
Common-factor exposure: meaningful at portfolio level.

The honest qualification

Correlation estimates are unstable. Pairwise correlations measured over 12 months can be 0.3 in calm markets and 0.7 during stress. Stress correlations are typically higher than calm correlations — exactly the period when diversification matters most. Portfolio construction should assume stress-period correlations rather than calm-period averages.