Turnover-Adjusted Alpha

Alpha — the return a strategy generates beyond what a benchmark or risk model explains — is usually reported per unit of time: percent per year. That hides an important variable: how much trading it took to earn it. A strategy that generates 5% alpha per year by rebalancing once a quarter and a strategy that generates the same 5% alpha by rebalancing daily are not equally attractive, because the second one is spending far more of its gross edge on transaction costs to arrive at the same net number, and it's far more exposed to any degradation in execution quality. Turnover-adjusted alpha normalizes alpha by turnover, answering "how much skill per unit of trading activity" rather than just "how much skill per year."

The idea, stated simply

Turnover, $T$, measures how much of the portfolio is bought and sold over a period — commonly expressed as the fraction of portfolio value traded per year. Turnover-adjusted alpha is roughly:

$$\alpha_{adj} = \frac{\alpha}{T} .$$

In plain English: divide the alpha earned by how much trading was required to earn it — a strategy that achieves its alpha with less turnover scores higher on this measure, because it's converting each unit of trading activity into more return, leaving more edge left over after realistic per-trade costs and less exposure to a bad execution day derailing the whole result.

Why raw alpha alone misleads on comparisons

Two managers can both report "5% alpha" and be running fundamentally different businesses. High-turnover alpha is fragile: it depends on capturing a thin edge many times, so it's more sensitive to a rise in trading costs, more sensitive to capacity constraints as assets under management grow, and more sensitive to execution slippage eating into the realized number. Low-turnover alpha earned from the same headline percentage is generally more robust on all three fronts — turnover-adjusted alpha is one concrete way to surface which situation you're actually looking at, instead of comparing two identical-looking headline numbers that mean very different things.

Worked example: same alpha, different turnover

Strategy P: 6% annual alpha, turnover 200% per year (roughly, sells and rebuys the whole book twice a year). Turnover-adjusted alpha: $\alpha_{adj} = 6\% / 2.0 = 3\%$.

Strategy Q: 6% annual alpha, turnover 50% per year (a quarter of the book turned over). Turnover-adjusted alpha: $\alpha_{adj} = 6\% / 0.5 = 12\%$.

Both strategies show identical headline alpha of 6%, but Strategy Q generates four times as much alpha per unit of trading as Strategy P. If both strategies face similar per-trade cost structures, Q's alpha is far more likely to survive realistic costs intact, and Q has much more room to scale in size before capacity constraints erode its edge — none of which the headline 6% figure alone reveals.

What this means in practice

When comparing strategies, or evaluating a manager's pitch, ask for turnover alongside alpha, not just alpha on its own. A high headline alpha achieved through very high turnover deserves more scrutiny of cost assumptions and capacity limits than the same headline alpha achieved with modest trading activity — turnover-adjusted alpha is a fast way to flag which situation applies before doing the full cost and capacity analysis.