forex-scalping.com

Average win, average loss and payoff: where break-even win rate comes from, with 50 trades

Three numbers from a journal give most of what you need. For 50 trades:

Average win: $84
Average loss: $60
Payoff ratio: 84 / 60 = 1.4

The break-even win rate is average loss divided by the sum of average win and average loss, 60 / 144 = 41.7%. The same thing is 1 / (1 + payoff). Anything above that is profit on average, ignoring costs that are already inside the averages.

If the actual win rate is 48%, the expectancy per trade is 0.48 x 84 - 0.52 x 60 = 40.32 - 31.20 = $9.12. Over 50 trades that is $456.

Now see how sensitive the result is. If the average win were $70, the payoff would be 1.17, the break-even would rise to 46.2%, and the expectancy would fall to 0.48 x 70 - 31.20 = $2.40 a trade, or $120 over 50 trades. A $14 change in the average win cuts the result by three quarters.

The break even win rate is the most useful number for judging whether a strategy has room for error. If the actual win rate is only a few points above break even, small changes in costs or execution can remove the edge entirely.

Use the net amounts, after commission and spread. An average win of $84 that is after a $7 commission is really a $91 gross win, and a $60 average loss is a $53 gross loss plus the same $7. If you calculate payoff from gross figures the break-even looks like 53 / (91 + 53) = 36.8%, which is too low by almost 5 points, because it forgets that costs come out on both sides. The journal should keep commission and swap in the trade's result, and the average columns should be built from that. A quick sanity check is that the sum of all trade results equals the change in account balance over the same period.

The averages hide the size of the distribution. An average win of $84 might be a mix of ten $40 wins and one $480 win. Then the payoff is dominated by the single trade, and the break-even calculation assumes a mix you will not necessarily get next month. Add the median win and median loss beside the averages. If the median win is $48 against an average of $84, the figure depends on a few large trades. That is not wrong, but the confidence interval on the expectancy is wider than the 50 trade average suggests, and the next section of the journal should show the top trades.

The same three numbers help you judge a change to exits. With an average loss of $60 and a 48% win rate, the average win needed to break even is 0.52 x 60 / 0.48 = $65. To make $5 a trade it has to be (0.52 x 60 + 5) / 0.48 = $75.42. Suppose trailing the winners could lift the average win by $10 without lowering the win rate. The gain is 0.48 x 10 = $4.80 a trade, about $240 over 50 trades. The catch is the clause about the win rate, since trailing usually turns some winners into smaller wins or scratches. Estimate how many before relying on the $240.

The three numbers also tell you which way a problem lies. A trader with a 62% win rate and a payoff of 0.5 has a break-even of 66.7% and loses money, even though most trades win. A trader with 35% winners and a payoff of 2.2 has a break-even of 31.3% and is profitable. Neither is visible from the win rate alone. In a journal, a column showing break-even win rate next to the actual win rate each month, updated as the averages move, gives you a gap in percentage points to watch. A gap that shrinks over three months is an early warning.

Consider a scalper aiming for three pips per trade. Over a hundred trades, direction is right about sixty percent of the time, which on its face sounds like a solid edge. The account is still slowly bleeding.

The first look says the exits are bad or the losers are too large. Neither turns out to be the main cause. Going through the data, the average spread paid is around one pip, and slippage on entries and stops adds another half pip across the sample. That is one and a half pips of friction on a three pip target, which is half of the gross gain gone before anything else.

The twist is what happens when the target gets raised slightly. Moving to five pips lowers the win rate a little, but friction becomes thirty percent of the target instead of half, and the net result flips positive. The direction was never the issue. The cost structure was.

That is the part worth taking away: being right is measured in price, but getting paid is measured after costs. On very small targets, the break-even win rate rises sharply, and a method can look good on a chart and fail on a statement.

The numbers above are illustrative and will differ by broker, pair, session and account type. Commission models, in particular, change the picture a lot.

Has anyone worked out the break-even win rate for their own target size after real costs, and how far off was the gut feeling?