An ES scalp can print green for a fortnight and still bleed the account. The only thing that settles it is expectancy per contract — computed in ticks, paid down to the last fee, and weighed against how many trades you’ve actually taken.
Expectancy in ticks per contract
Expectancy is the average result one trade of a given setup returns when you repeat it many times. In its plainest form:
Expectancy = (Win% × average win) − (Loss% × average loss)
For a scalp, put both the average win and the average loss in ticks per contract rather than dollars. Ticks are the natural unit: they compare cleanly whether you traded one contract or five, and they strip out position sizing so you’re measuring the setup, not the size.
Say your ES scalp wins 60% of the time for an average of 4 ticks and loses 40% of the time for an average of 5 ticks. Gross expectancy is (0.60 × 4) − (0.40 × 5) = 2.4 − 2.0 = +0.4 ticks per contract per trade. Positive — but hold that thought, because we haven’t paid a single cost yet. Drop your own win rate and tick figures into an expectancy calculator to get the gross number before the subtraction starts.
Subtracting commissions and fees per round-turn
Every trade you close pays a round-turn cost: broker commission plus exchange and clearing fees, charged going in and coming out. On index futures that round-turn is a fixed dollar drag per contract — and for a scalp clipping a few ticks, it’s enormous relative to the edge.
Convert the round-turn into ticks so it lands in the same unit as your expectancy. On ES, one tick is worth a fixed dollar amount per contract; divide your total round-turn by that tick value and you have the cost in ticks. Suppose it works out to roughly 1 tick per round-turn. Redo the math:
Net expectancy = +0.4 gross − 1.0 cost = −0.6 ticks per contract
The same setup that looked like a +0.4 winner is a −0.6 loser once it pays its way. That reversal is the single most common reason a “profitable” scalp drains an account, and it’s why costs belong inside the expectancy calculation, never bolted on afterward.
Why a positive scalp can still lose money
Three traps make a scalp look better than it is:
- Costs scale with frequency. The tighter your target, the more trades you take, and the more round-turns you pay. Clipping 3 ticks a hundred times a week means a hundred round-turns. Frequency is never free.
- Skewed averages hide the tail. Scalpers often run a high win rate with small wins and rare large losses. A 70% win rate feels safe, but if the occasional loss dwarfs the average win, expectancy can still be negative. The win rate is not the edge.
- Slippage eats the target. In a fast market the fill on entry and the fill on your stop are both worse than the price you clicked. A couple of ticks of average slippage can flip a marginal scalp negative on its own.
The discipline is to compute net expectancy per contract honestly, with real fills — and to distrust any positive result that ignores frequency and slippage.
Sample size and a confidence interval
Here’s the uncomfortable part: even a correctly computed positive expectancy can be luck when it rests on a handful of trades.
A win rate measured over 15 scalps tells you almost nothing — the plausible range around it is enormous. The same rate over several hundred trades is a far more trustworthy signal. The honest way to express that is a confidence interval: rather than “my win rate is 60%,” you say “given my sample, the true rate sits somewhere in this range.” When the range is wide and straddles the break-even win rate your costs demand, you don’t have evidence of an edge yet — you have a small sample.
This is precisely the read Shibiki puts on every strategy. It wraps your live win rate in a Wilson confidence interval that tightens as the trade count grows, so you can see at a glance whether a positive-looking scalp has enough trades behind it to act on, or whether you’re staring at a lucky run. The trading expectancy explainer is a good companion for the reasoning underneath.
When to keep, tune or kill the strategy
Put the two questions together — is net expectancy positive? and is the sample big enough to believe it? — and the decision nearly makes itself:
| Net expectancy | Sample / interval | Decision |
|---|---|---|
| Clearly positive | Tight, stays positive | Keep — consider scaling within your risk limits |
| Slightly positive | Wide, straddles break-even | Keep small — gather more trades before sizing up |
| Negative | Any | Tune or kill — the costs are winning |
Tuning comes down to one of three levers: widen the target so wins clear the round-turn with room to spare, cut frequency to pay fewer costs, or tighten entries to lift the win rate. Change one lever at a time and re-measure — otherwise you can’t tell which move helped.
Shibiki’s auto-journaling captures each fill and its costs without you retyping anything, so the net-of-fees expectancy and its confidence interval stay current rather than frozen in a month-old spreadsheet. If you’re moving over from a manual tool, the Shibiki vs. TraderVue comparison covers where the automation changes the workflow. The rule is simple: never scale a scalp until the math, after costs and with enough trades behind it, says the edge is real.
Related: Expectancy calculator · Trading expectancy · Shibiki vs. TraderVue