The runs-test as a contract-sizing gate

The score as a sizing-eligibility gate

A runs-test is a comparison of the number of win-loss switches in an ordered closed-trade sequence with the number of switches expected if each outcome were independent. The archive workflow computes a signed Z-score from that comparison and then maps the score to a confidence figure.

Editorial reading: treat that score as a sizing-eligibility gate, not as a verdict on whether the entry signals are good. The next contract rule, stay fixed, step up after losses, or add after wins, becomes a testable procedure only after the score clears a high confidence bar.

How the Z-score is built

The Z-score is a signed statistic that measures how far the observed run count departs from an independent-outcomes baseline. It is computed from the total closed-trade count, twice the product of the win count and the loss count, and the number of outcome switches in the ordered sequence.

A run is one uninterrupted stretch of wins or of losses. A new run starts, and a switch is counted, each time the ordered sequence changes from a win to a loss or from a loss to a win.

The calculation is described as requiring a sample of at least 30 closed trades. The Z-score may be positive or negative. The matching confidence-limit is the Z-score restated as the percentage of cases expected inside a matching standard-deviation range, and it is reported only as a positive percentage.

The confidence bar before a size rule

A positive Z-score means more runs than an independent baseline, so wins and losses tend to alternate. A negative Z-score means fewer runs, so like outcomes tend to cluster.

Dependence is treated as unsupported below a 95 percent confidence figure. The 90 to 95 percent band is treated as more likely an anomaly than an exploitable pattern. One, two, and three standard deviations are mapped to about 68, 95, and 99 percent.

The illustrated sample had 45 trades, 24 wins and 21 losses, and a positive Z-score of 2.15 mapped to a 96 percent confidence figure.

Positive scores and after-loss step-ups

When the Z-score is positive, alternation is the working hypothesis. The procedure then compares fixed-contract sizing, a constant one-contract count after each successive loss, with martingale-style step-ups that raise contracts after the first, second, and third consecutive loss, including a two-four-eight schedule.

Fixed-contract sizing is the bounded baseline against which those after-loss step-ups are compared. The martingale rule is a contract-count rule that raises size after consecutive losses under that positive-score hypothesis.

Negative scores and clustered outcomes

When the Z-score is negative, wins and losses tend to arrive in clusters. The procedure keeps or raises size during a winning cluster, cuts size after a loss, or stands aside until a losing cluster ends, without changing the entry rule.

Editorial reading: the three later procedures are stay fixed, step up after losses, or add after wins. None of them is in play until the signed score and its confidence-limit have first cleared the gate.