Partial exits and effective reward-to-risk
Follow quantities through each exit instead of crediting the entire position with the furthest target.
Stress a payoff assumption without pretending a favourable average bounds the next loss.
Price-level arithmetic and user-entered samples. A target ratio is neither a measured edge nor a guarantee of the stop fill.
A two-outcome break-even formula assumes every winner earns one amount and every loser loses another. Real trades can finish flat, partially filled, liquidated or far beyond the intended stop. These outcomes need their own categories. A calculator result is conditional on the chosen branches; it does not establish that a planned loss is the largest possible loss. The target ratio alone cannot tell you whether rare losses overwhelm ordinary winners.
Begin by listing what the simple model omits. For a cash-funded stock example that might include a gap through the stop, unavailable liquidity and extra execution cost. Derivatives and leveraged accounts introduce additional rules and liabilities which this worksheet does not model. A scenario must name its instrument and funding assumptions before its loss magnitude means anything.
Imagine ten completed positions: six at +1R, three at -1R and one at -5R. This invented series has six wins out of ten yet sums to -2R before any extra costs. The 60% observed win fraction does not rescue the payoff. Replace the -5R observation with a -1R observation and the series sums to +2R; that comparison identifies sensitivity to a loss assumption, not permission to remove a real loss from a report.
A second exercise changes a cost rather than a tail: deduct 0.1R from each completed position and the first series becomes -3R. That common deduction is intentionally simplified. In a real record, spreads, commissions and financing belong to the actual position and can depend on size and duration. Reconcile them rather than applying a convenient average twice.
The analyser reports the largest winner as a share of total positive R. That answers how concentrated positive contributions are, not how stable the strategy is. A record with one large winner may still be a legitimate outcome of its stated policy. A leave-one-out calculation is useful as a labelled sensitivity view, but the original complete series must remain the primary record. Apply the same removal rule to every observation rather than selecting only the one that supports your argument.
Sequence also matters for a completed-R path. Permuting outcomes preserves the mean and profit factor but can change prior-peak drawdown and consecutive losing runs. Neither arrangement tells you intratrade equity drawdown. Do not choose a favourable permutation as the historical order. Keep actual timestamps and identify simultaneous positions which a single sequential path cannot adequately represent.
The stock stop-order source establishes a mechanism for execution uncertainty, not the frequency or severity of a strategy-specific gap. The statistical source describes interval assumptions, not a calibrated tail-loss model. To estimate a tail you need a defensible population, sufficient observations, instrument-specific execution evidence and a declared method. A short selected record cannot supply these merely by including one alarming example.
A practical review ends with a stress table showing the unchanged original series, an explicitly hypothetical larger-loss scenario, an alternative cost assumption and the reason each scenario was chosen. Keep estimated probabilities out when no source supports them. Write which account constraints each scenario would challenge, without calling the calculated balance a guaranteed floor. This approach produces a useful question for further research even when it cannot produce a reliable future loss forecast.
Make a table containing the original entered sample and a separately labelled hypothetical stress sample. Alter one loss magnitude without changing its date, identity or other observations. Recompute mean R, total R and prior-peak decline. The comparison shows sensitivity to that chosen loss, not its probability or maximum possible severity.
Repeat with an extra cost applied to every row rather than only losers. Record whether original inputs were already net of that cost, because double deduction can create a misleading result. Save the unchanged original beside both counterfactuals. A stress exercise should remain useful even when it cannot tell you how often a market event occurs; its defensible conclusion is which accounting or exposure assumption drives the result, under clearly preserved fictional inputs.
Documentation establishes the stated mechanism, not tested trading performance. Checked 9 October 2026; a later version can differ.
Follow quantities through each exit instead of crediting the entire position with the furthest target.
Separate completed-trade R, compounded equity and the changing cash base.
The invented full-gap reference exits 100 long units at 94 from entry 100 and stop reference 98. Gross loss is 600; extra cost 20 gives negative 3.1R on original gross planned risk 200. The arithmetic demonstrates why a stop reference is not a guaranteed payoff cap. It supplies no frequency for that gap and no maximum possible loss. The unchanged full-target row stays alongside it. The new three-outcome investigation requires explicitly hypothetical weights before an expected result, while the completed-R sequence guide preserves actual order as a separate source question.