2001issue C101-7
A runs test before volatility and expected-value sizing
Money-management-indicators describe the last N closed trades, not the full book. Editorial reading: let a runs-test decide whether that sequence is still independent, and only then use volatility-position-sizing and an expected-value filter to bound how much of the account the next loss may consume.
- Money-management-indicators are computed from the last N closed trades and hold until a new exit, so they describe recent system behavior rather than the full trade history.
- The runs-test is a specialized z-test on win and loss streaks. Fewer runs than a random ordering is labeled positive trade-dependency; more runs is labeled negative trade-dependency.
- Cutting size because equity is below a moving average is treated as justified only when the runs-test shows trade-dependency.
- If no trade-dependency is found, percent-profitable can serve as an expected-value filter, and volatility-position-sizing then keeps the next loss bounded.
Two gates for the next loss
System-response indicators, used here as money-management-indicators, are computed from a moving window of the last N closed trades and hold their value until a new exit, so they describe recent system behavior rather than the full trade history.
TradersWeek editorial: treat position sizing as a two-gate process. A runs-test first decides whether the recent closed-trade sequence is still independent. Only then do volatility-position-sizing and an expected-value filter set how much of the account the next loss is allowed to consume.
The average-trade-window
The average-trade-window is the mean profit or loss of the last N exits. That window can be divided by average true range so an upward price bias is less likely to be mistaken for a change in system quality.
A t-test of the last N trades against a longer-run mean is used to decide whether a change in average trade is large enough to review allocation or the next unit of risk.
Percent-profitable is the share of winners inside the same N-trade window, compared with a longer-run baseline before it is treated as a new payoff estimate. A z-test on win rate is illustrated with 45 percent winners in the last 30 trades versus a 60 percent longer-run rate, producing a z of -1.68, which did not meet the two-sided 1.96 threshold at 95 percent confidence.
The runs-test
The runs-test is presented as a specialized z-test that compares the observed number of win and loss streaks with the number expected if wins and losses arrived in random order. It is used to decide whether outcomes look independent or serially dependent over a defined lookback of trades.
In a worked 15-trade sequence with 7 wins, 8 losses and 8 runs, the computed z of -0.25 did not meet the 95 percent two-sided threshold of plus or minus 1.96, so no trade-dependency was inferred.
When trade-dependency would change size
Trade-dependency is serial structure in wins and losses. Fewer runs than a random ordering is labeled positive trade-dependency and more runs is labeled negative trade-dependency. The first case would shrink or skip after a loss, and the second would shrink after a win and enlarge after a loss.
A statistically significant drop in recent average trade, equity momentum, or percent-profitable is treated as a reason to cut the next position so exposure stays bounded to current conditions.
Volatility and expected value after independence
If no trade-dependency is found, percent-profitable is treated as the better estimate of the chance the next trade is a winner and can serve as an expected-value filter before adding risk.
Volatility-position-sizing scales the next unit of exposure from account equity, current volatility or stop distance, and the latest window of results so a loss stays bounded before entry and while the trade is open. Expected-value is the probability-weighted payoff implied by the recent win rate and average win versus loss, used as a go or no-go filter on the next unit of risk when the trade sequence does not show trade-dependency.
TradersWeek editorial: those two tools are the second gate. They set how much of the account the next loss is allowed to consume only after the runs-test has spoken on independence.
All readings on this track · 15 readings
- 1986Constructing runs and persistence tests from labeled prices
- 1986Evaluating daily price and volume serial independence windows
- 1986Evaluating advance-decline plus-day runs against chance baselines
- 1986Weekly resamples as a diagnostic filter for statistical windows
- 1988Runs test as a critique of price-series memory
- 1989Evaluating weekday close direction with a counted baseline
- 1989Statistical windows for indicator time parameters
- 1992Channel-height ratios for equity trend evaluation
- 2001A runs test before volatility and expected-value sizing
- 2005Constructing runs-test z-scores for signed return persistence
- 2005Evaluating persistence with runs and autocorrelation
- 2005Weekday FX turning points and close run tests
- 2013Constructing a runs-test turn forecast
- 2017Star rating from slope and swing runs
- 2018Regime-dependent odds after directional price runs