MSPT methodology

How MSPT Calculates and Interprets Trade Review Metrics

MSPT uses recorded entries, initial stops, quantities, fills, and reviews to calculate descriptive trade metrics with explicit sample and coverage limits.

Published and reviewed by Millwheel Oy · Updated October 31, 2026

Source register

What each source contributes

MSPT trade lifecycle guide

Role: Defines the records that connect research, planned risk, execution, management, and review.
Timing: Reviewed with the current public MSPT workflow.
Limitations: A complete lifecycle still depends on accurate user and broker records.

MSPT portfolio risk and trade planning methodology

Role: Explains initial risk, R levels, portfolio context, and plan-relative targets.
Timing: Reviewed with the current public MSPT workflow.
Limitations: Plan-relative risk is not a prediction of execution price or future return.

Position-level realized P/L and R

For each recorded long-stock plan, realized P/L is the sum of (exit fill price minus recorded entry price) multiplied by the quantity of each exit fill. Realized R is realized P/L divided by (recorded entry price minus initial stop) multiplied by the original recorded shares.

When a fully closed base plan has add-on legs, MSPT sums the realized P/L and original dollar risk of those legs before calculating the combined position R. A raised current stop does not reduce the original denominator.

Exit-ladder levels are planned instructions. Analytics use recorded fills, including partial exits, rather than assuming that touching a planned level produced an execution.

Sample-level formulas

Win rate = positive realized-R outcomes divided by valid realized-R outcomes, multiplied by 100.
Average win R = the arithmetic mean of outcomes above zero.
Average loss R = the arithmetic mean of outcomes at or below zero.
Expectancy R = the sum of valid realized-R outcomes divided by their count.
Average R uses the same arithmetic-mean calculation for the selected closed-trade cohort.
Plan adherence uses only reviewed records where followed-plan status was answered; entry discipline uses only rated entries.

Worked calculation

For a $50 entry, $46 initial stop, and 100 original shares, the initial dollar risk is ($50 - $46) times 100, or $400. If recorded exits produce $600 of realized P/L, realized R is $600 divided by $400, or +1.5R.

For a cohort of +2.0R, -1.0R, +0.5R, -0.8R, and +1.3R, MSPT reports five valid outcomes, a 60% win rate, about +1.27R average win, -0.90R average loss, and +0.40R expectancy. Display rounding can create small differences when the components are recombined.

Coverage and evidence labels

Open positions are shown separately and do not enter closed-outcome metrics. A missing or non-finite realized R is excluded rather than assigned a zero. MSPT displays sample size so the user can see how much evidence supports a result.

Fewer than 10 valid closed outcomes: building evidence.
10 to 19 valid closed outcomes: directional.
20 to 39 valid closed outcomes: growing descriptive sample.
40 or more valid closed outcomes: larger descriptive sample.

Coaching cohort comparisons

MSPT can compare reviewed trades assigned to a coaching rule with eligible earlier reviewed trades. The displayed cohort measures include average R, win rate, written-plan adherence, and entry discipline. A short target helps build the habit, while MSPT requires at least 10 rule trades for an initial outcome evaluation and treats 20 as a more meaningful descriptive sample.

The comparison is observational. Market regime, setup mix, portfolio constraints, and execution conditions may differ between the baseline and rule cohorts.

Interpretation and limitations

These are descriptive account records, not audited performance returns or forecasts.
The current stock-plan calculation does not automatically deduct commissions, fees, taxes, dividends, or foreign-exchange effects unless reflected in the recorded data.
Incomplete execution history, incorrect corporate-action adjustment, or a changed plan can alter the result.
Samples should be segmented only by consistently recorded definitions; repeated filtering can produce misleading patterns.
A positive outcome does not prove good process, and a negative outcome does not prove a bad decision.