The Break-Even Transaction Cost of a Signal

A backtest shows a signal earning 8% a year gross, trading every stock in the universe every day. That number is close to meaningless on its own, because it says nothing about whether the signal can survive contact with a real order book. The number that matters is the break-even transaction cost: the round-trip cost per dollar traded at which the strategy's net return falls to exactly zero. Above that cost, the signal loses money no matter how good its gross return looks.

The idea is simple because turnover and cost multiply together. A signal that trades a small fraction of the portfolio each day can tolerate high per-trade costs. A signal that flips positions constantly needs costs to be almost free just to break even, even if its raw predictive power is stronger.

The calculation

Turnover measures how much of the portfolio gets rebalanced per period — a strategy with 100% annual turnover trades the equivalent of the whole book once a year. Each unit of turnover pays the round-trip cost once. So:

$$c_{BE} = \frac{R_{gross}}{T}$$

In words: the break-even cost per dollar traded equals the strategy's gross annual return divided by its annual turnover. If actual trading costs are below $c_{BE}$, the strategy is net profitable; if they're above it, the signal is a paper tiger.

Worked example

A mean-reversion signal backtests at 12% gross annual return with 1,500% annual turnover (the book turns over 15 times a year, typical for a fast signal). Break-even cost:

$$c_{BE} = \frac{0.12}{15} = 0.008$$

That's 0.8% per round trip, or 80 basis points. If the desk's real-world cost — spread crossing plus market impact — for the names it trades is 15 basis points round trip, the strategy clears its break-even with plenty of room: 80 bps available, 15 bps actually paid, leaving roughly 65 bps of net edge per unit of turnover. But if the same signal is pushed into less liquid small-caps where round-trip cost runs 60 bps, the margin shrinks to 20 bps — and if the researcher then tries to run it at twice the turnover to capture more of the signal's decay, the break-even cost is cut in half to 40 bps, and the strategy is suddenly unprofitable in the same names.

What this means in practice

Break-even cost is the first filter a researcher applies before a signal ever reaches a live book, because it converts an abstract backtest Sharpe into a concrete, checkable number against a desk's actual execution costs. It also explains why the same signal can be profitable for one fund and worthless for another — a fund with better execution technology or more patience simply has a lower real cost to compare against the same break-even line.