The Alpha Lives Where the Costs Are
Key takeaways
- Short-term reversal was the only signal in the study above t=2 (t=2.35 gross) — and its net long/short Sharpe is ~0 once weekly turnover costs are charged.
- Small-cap post-earnings drift earns +2.7%/yr market-neutral at 10 bps — but breaks even on borrow cost at ~5–6%/yr, and Q1 names are exactly the “special”, expensive-to-borrow ones.
- Restrict the short leg to easy-to-borrow names and the edge halves, then vanishes — before any borrow is charged.
- Pairs stat-arb: no large-cap gross edge; small-cap gross +0.48 turns to −1.34 net of cost and borrow.
- Same signature three times: real gross alpha in illiquid names, uncapturable net of the frictions that let it survive.
Three of the signals we tested looked genuinely tradeable in gross terms. Short-term reversal printed the highest t-statistic in the entire program. Small-cap post-earnings drift produced a clean market-neutral spread. Pairs stat-arb showed a positive small-cap edge. Each one died the same way — and the way they died is the most important finding in the whole research program.
Short-term reversal: significant, and untradeable
Buy last week’s losers, short last week’s winners, rebalance weekly. Across point-in-time membership from 2000–2018 this was the only signal to clear t=2 — a full-sample t-statistic of 2.35, positive in every regime. It is, statistically, the strongest gross signal we found. Its net long/short Sharpe is approximately zero. The weekly turnover that generates the signal also generates the trading costs that consume it. In small-cap names the gross t-stat is even higher (+2.27) — because illiquid names mean-revert harder — and the net Sharpe is +0.04, because illiquid names cost even more to trade.
Post-earnings drift: the edge is fenced behind the borrow
Post-earnings-announcement drift is the behavioral anomaly with the best out-of-sample pedigree. Measured properly in event time on small caps, the high-minus-low surprise spread widens to +1.12% by 63 days, and a market-neutral long/short book monetizes it: +2.7%/yr at a realistic 10 bps, Sharpe 0.52. The first thing in the program that looked like deployable alpha over the index. Then we charged the costs that the headline number omits.
| Small-cap PEAD long/short | Annual | Sharpe |
|---|---|---|
| Gross, 10 bps trading, 0% borrow | +2.7% | 0.52 |
| + 3% borrow on the short leg | +1.2% | 0.23 |
| + 6% borrow (breakeven) | −0.3% | −0.05 |
| Short restricted to easy-to-borrow names | ~+1.3% → 0 | 0.20 → −0.02 |
The short leg is the catch. The book needs it — the long-only version is actually negative, because small caps as a group trailed; the edge is dispersion, not a long tilt. But the names you most want to short are the low-surprise small caps that just disappointed: exactly the “special”, hard-to-borrow names that cost 5–50%+ per year to borrow. Breakeven is around 5–6%. And when we restricted the short leg to liquid, borrowable names, the edge halved at the top 50% by dollar volume and disappeared at the top 33% — before charging any borrow at all. The alpha was concentrated in precisely the names you cannot cheaply short.
Pairs stat-arb: the same wall
Classic cointegration pairs, with out-of-sample pair formation to avoid snooping. In modern large-cap US equities the gross Sharpe was −0.03 — the relationship found in formation simply does not persist into trading; the edge has been arbitraged away. In small caps there was a real gross edge (+0.48) — and it turned to −1.34 net once 10 bps of cost and 4% borrow were charged. Same place as the other two.
Why is this the same story three times?
Because it is one mechanism: limits to arbitrage. An edge survives only where it is hard to harvest — illiquid names, wide spreads, costly or impossible borrow, high turnover. The very frictions that stop other investors from competing the edge away are the frictions that stop you, too. Three independent signals showing the identical signature is not bad luck; it is a property of efficient-enough markets. The cleaner your cost accounting, the more completely the alpha dissolves.
The hinge
This is why we harvest a risk premium — compensation for bearing risk others want to shed, which IS capturable because it is not a mispricing — rather than chasing an arbitrage, which is not. The next question is which risks actually pay.
About this series: every figure comes from a leak-free research harness on US equities — point-in-time index membership, fundamentals keyed to filing date, expanding-window walk-forward, and transaction costs charged. Statistics are gross and in-sample unless noted, and describe published anomalies, not a Yayati product. Standing caveats: roughly a third of true historical index members are unpriced by the naive data source (survivorship); a 2 bps cost assumption is optimistic; fundamentals are post-2009 XBRL.
Keep reading
This article is for educational and informational purposes only and is not investment, tax, or legal advice. It describes findings from an internal research program about publicly documented market anomalies and research methodology; it is not a description of any Yayati product or its results. Research statistics are gross, in-sample illustrations subject to survivorship, data-coverage, transaction-cost, and modeling limitations described in the text, and do not represent actual trading or any client account. Past performance and backtested results are not indicative of future results. Yayati Asset Management is a Registered Investment Adviser. © Yayati Asset Management. VOLT™ and PLASMA™ are trademarks of Yayati.
See VOLT™ on a real position.
The tax-smart option overlay behind this paper, for concentrated stock.