A target three times the initial stop distance looks attractive on a chart. To evaluate it, you still need to know how often each exit occurs, what it earns or loses after costs, and whether the assumed prices were executable. This article separates planned asymmetry from measured expectancy and shows why stretching a target cannot establish an edge on its own.

What an asymmetric position actually is

A planned reward-to-risk ratio compares target distance with initial stop distance; it does not determine the future win rate or mathematical convexity. Convexity concerns the curvature of a payoff function, while expectancy also depends on the frequency and net amount of wins and losses. An ordinary stop sets a trigger rather than a guaranteed loss cap. Evaluate the complete execution rule before calling a payoff profile favourable.

The math of win rate versus reward-to-risk

For a two-outcome model, E = p × W − (1 − p) × L, with positive win amount W, positive loss magnitude L and win probability p. CME illustrates the frequency-and-size calculation in a commodity-futures lesson. Assigning p to an illustration is different from estimating future probabilities from a journal.

Retain a separate hypothetical gross-only calculation with fixed €100 losses, exact assumed outcomes and no fees or compounding. For €100 winners, p = 60%, 50%, 40% gives means +€20, €0, −€20. For €300 winners, p = 40% or 30% gives +€60 or +€20; the zero threshold is 100/(300 + 100) = 25%. For €500 winners, p = 30% gives +€80 and the zero threshold is 1/6 ≈ 16.67%. These are arithmetic assignments, not measured strategies.

With €500 wins and €100 losses, p = 20% gives +€20 gross, and p = 25% gives +€50. A separate symmetric model at p = 55% gives +€10. These choices do not compare actual trend followers and scalpers. Win frequency remains a component of the mean; average profit alone also omits tail losses, uncertainty and the size rule.

R describes a trade's result, but does not weight the money for you

Record A: + 5 R with R = 1 USD. Record B: − 1 R with R = 20 USD. The unweighted average (+ 5 − 1)/2 is + 2 R. However, the amounts are + 5 and − 20 USD, totalling − 15 USD. The apparently larger R winner carried much less monetary weight here.

Positive average R, negative total USDTwo hypothetical outcomes. A earns plus five R with one dollar per R, or plus five dollars. B loses one R with twenty dollars per R, or minus twenty dollars. The unweighted average is plus two R, while the money totals minus fifteen dollars. Panels use separate units and scales.Check units before addingResult in RA+ 5 RB− 1 RAverage: + 2 RResult in USDA: + 5 USDB: − 20 USDTotal: − 15 USD
Green and red bars show the sign and size of each result. The panels use different units; lengths cannot be compared between panels.

Keep the normalised result, monetary value of R and cash outcome in your journal. Equal weighting answers a question about the average profile of these two observations. The total change in capital requires their actual monetary weights. Changing position sizes can reverse the sign of the portfolio result even when the number of winners and their recorded R multiples stay the same.

Assumptions: Two entirely hypothetical, completed net outcomes. R is the initial risk amount fixed for each record: 1 and 20 USD respectively. No further costs or compounding are added; nothing here specifies how frequently these events occur.

Costs and worse fills can change the comparison

Use a fictional USD account buying 10,000 EUR at executed Ask 1.1000, with initial SL 1.0900 and target 1.1300. Define initial price-risk R = 100 USD before commission, not maximum loss or margin. Assign commission 6 USD for the complete cycle, funding/conversion zero, and each full sell at its stated actual Bid fill. These are model assumptions, not a broker offer.

Separate exit scenarioBid fillGross USDNet USD / initial R
Target1.1300300294 / 2.94 R
Exact planned stop1.0900−100−106 / −1.06 R
Assigned worse stop fill1.0700−300−306 / −3.06 R

The gross planned 3:1 becomes 294:106 ≈ 2.77:1 net. Ask/Bid fill differences already reflect spread; do not deduct it again. The last row is a stress assignment, not a worst-case bound or predicted frequency. One +294 USD win would not cover one −306 USD loss.

If the only net outcomes are +294 and −106, zero expectancy requires p = 106/(294 + 106) = 26.5%. Assigned p = 30%, 25%, 40% gives +14, −6, +54 USD respectively. Include all partial/flat outcomes and variable costs when the two-outcome assumption no longer applies.

Two invented 100-trade series show sensitivity to losses: 30 wins at +294 and 70 losses at −106 total +1,400 USD, mean +14. Replacing five ordinary losses with −306 gives 30 × 294 − 65 × 106 − 5 × 306 = +400 USD, mean +4. Both win rates remain 30%, but the total falls by 1,000 USD. The assigned loss frequency is not a market statistic.

A distant target does not make a linear position convex

In MetaQuotes’ Forex/CFD calculation modes, gross P/L for fixed units is the signed price difference times those units. Doubling the price move doubles this component before costs. A distant target changes exit conditions, not that linear instrument formula; an ordinary stop is not a prepaid option premium. This distinguishes planned payoff asymmetry from mathematical curvature.

A separate fixed budget of 0.5% of starting capital per trade gives a 5% reduction after ten exact full-budget losses, before extra costs. If the budget is 0.5% of current equity instead, the reduction is 1 − 0.99510 ≈ 4.89%. These assigned exact losses illustrate two sizing rules; ordinary stops do not guarantee them. A later winner covers the streak only if its net monetary gain exceeds the accumulated losses.

The other side of the coin: picking up pennies in front of a steamroller

“Picking up pennies in front of a steamroller” describes frequent small gains accompanied by a risk of a much larger loss. That distribution does not by itself specify curvature of an instrument payoff, timing of a loss or inevitable account failure. Inspect the entire net-loss distribution rather than only a smooth segment of an equity curve.

Adding exposure or repeatedly postponing closure can increase loss and margin demands; the actual contracts, stops and size rules determine the exposure. Do not assign a guaranteed high win rate or an unavoidable future catastrophe from a technique’s name. Test adverse execution and ask how much a large loss would remove from accumulated gains.

Positive average returns need not mean capital growth

Now change the sizing convention in a separate mathematical counterexample, not the earlier fixed-R tables. Assign each win +75% of current equity and each loss −25%, omitting costs and execution constraints. These deliberately extreme percentages are not a proposed risk budget. With three wins and seven losses, the arithmetic average is (3 × 75% − 7 × 25%)/10 = +5%, yet capital multiplies by 1.753 × 0.757 ≈ 0.7154. Starting 1,000 USD becomes about 715.39 USD, a 28.46% decline.

Do not apply the average +5% as ten identical gains. With these fixed multipliers, order does not change final capital, but can change drawdown and whether continuing is feasible. The unchanged diagram adds a different caveat: +5 R at 1 USD and −1 R at 20 USD average +2 R but total −15 USD. Keep monetary weights and compounding separate from an unweighted mean.

What to do tomorrow

Evaluate asymmetry through actual net outcomes and explicit sizing. A few selected winners, a favorable target ratio or a fixed minimum sample count do not disclose the true future distribution.

  1. Calculate the sample net mean, average win/loss, largest losses and contribution of the largest wins. Keep zero outcomes and fees; distinguish an estimate from a known future expectation.
  2. Record planned versus realized R, the monetary amount of each initial R and the exposure rule. A positive mean in R can accompany a negative monetary total.
  3. Stress worse fills and costs without treating chosen frequencies as market facts. Compare rules on data not used to select targets or sizing.
  4. Check capital paths and drawdown under the actual size rule. Neither a universal 2 ATR trail, 0.5% budget nor one hundred trades establishes an edge or a guaranteed loss cap.

Related reading: the expectancy formula walks through the arithmetic asymmetry rests on; the 1 percent versus 2 percent rule shows how to size positions for a string of losses; stop loss versus take profit organises the logic of defining loss and target, and the trade exit strategy ties trailing stops and partial profits into one system. For broader context, see the risk management section in our library.