Journal and expectancy

Compute expectancy, then keep the process column separate from the money. · 11 min

Hook · Two columns

The result and the rule are not the same sentence.

You followed the written rule and lost 1R. The process column should say…

Process and result are two columns. A followed loss is one data point. It is not a reason to rewrite the rule that same afternoon, and it is not a grade of character.

Squeeze · Average R, not a feeling

One R is the loss you planned on that trade, the same planned loss you sized from the stop. Lose exactly that and the result is −1R. Make twice that and it is +2R. A gap that loses 1.4 times the plan is −1.4R. Fees make the real number a little worse. Expectancy is win rate × average win, minus loss rate × average loss. A sample that wins 40% of the time, with winners of +2R and losers of −1R, is 0.4 × 2 − 0.6 × 1 = +0.2R per trade. The break-even win rate for that payoff is 1 ÷ 3, about 33%, and only if the losses really are 1R. Positive expectancy can sit next to a low win rate. A small sample can look brilliant by luck. The journal does not create an edge. It records whether you followed the rule in a column that is not the money. The common mistake is to edit the rule because one followed loss felt loud, which is the discipline problem waiting at the end of the lab.

Body · Two samples

Enter the expectancy. The second sample is flat.

Five trades: +2R, −1R, +2R, −1R, −1R. What is the expectancy per trade?

Two wins out of five is a 40% win rate. Average win +2R. Average loss 1R. 0.4 × 2 − 0.6 × 1 = +0.2R. That is the sample average. It is not a claim about the next trade. Fees would nibble at it.

Harder only because the sign is boring. Four trades: +1R, −1R, +1R, −1R. Expectancy per trade, before fees?

Win rate 50%, average win 1, average loss 1. 0.5 × 1 − 0.5 × 1 = 0. Fees would push a flat sample negative. A feeling that you are ahead is not in the formula.

Wicks · A gap, then a broken rule that made money

R is allowed to be worse than −1. Process is allowed to be wrong on a winner.

You planned to lose $100. A gap fills you down $140, before fees. The result in R is…

140 ÷ 100 = 1.4 times the plan, and it is a loss, so −1.4R. The formula does not round a gap back to −1R. Fees would make it a little worse. The size lesson warned you about this and the journal is where you write it down.

You skipped the written exit, held on, and the trade made +3R. Process column:

A broken rule that made money is still a broken process. If you only keep the winners that broke the rule, the journal flatters you the way a backtest flatters a peeked fill.

Challenge · Fresh samples

A 1-in-3 win rate, then a single loss that is not a new system.

Winners +3R, losers −1R, win rate 25%. Expectancy per trade, before fees?

0.25 × 3 − 0.75 × 1 = 0.75 − 0.75 = 0. The break-even win rate for a 3-to-1 payoff is 25%, and only if the losses really are 1R. A gap that makes losses −1.4R moves the break-even up.

Three followed winners, then one followed −1R. The sample of four is not the system. The careful next step is…

One loss does not change the expectancy math by itself. A streak is not a new edge. The journal’s job is the two columns, not a new rule written while the loss is still loud.

Reveal · Five trades, worked

+2R, −1R, +2R, −1R, −1R. Two wins, three losses. Win rate 0.4. Average win 2. Average loss 1. 0.8 − 0.6 = +0.2R per trade in the sample. If the third loss had been a gap to −2R, the average loss would be (1+1+2) ÷ 3 = 1.333, and expectancy would be 0.4 × 2 − 0.6 × 1.333 = 0.8 − 0.8 = 0 before fees. The gap ate the edge of this tiny sample. That is why the journal records the real R, not the R you meant. Process on all five said followed. The sample is still too small to be evidence. A backtest of five trades would have gotten the same warning.

If that third loss is −2R instead of −1R, this five-trade sample’s expectancy is about…

Average loss becomes 4 ÷ 3. 0.4 × 2 − 0.6 × (4/3) = 0. The definition of R did not break. The loss was simply larger than the plan.

Case · Case file: two columns

The chart is the trade. The journal is the grade.

Check · Fresh journals

New samples.

You followed the written stop and lost 1R. The two cells for that trade are…

Answer

Process followed, result −1R. Process followed, result −1R. The loss does not rewrite the process column.

You moved the stop, the trade recovered, and the result was +2R. The process cell says…

Answer

Broken, even though the result is +2R. Broken. The flattering result does not repair the rule you left.

The plan has a daily loss limit and you have not hit it. You skip a valid signal because you feel off. Unless the plan wrote that feeling down, you…

Answer

Edited a rule the plan did not contain. Edited the rule. A feeling is not a rule until it is written and then followed as written.

Three winners, written size unchanged, next setup valid. Doubling the size belongs in the journal as…

Answer

A new size the written plan did not earn. A different size from the one the sample earned. The streak is not a new edge.

Key terms

R
One R is the loss you planned on that trade. Lose exactly that and the result is −1R. Make twice that and it is +2R. A gap that loses 1.4 times the plan is −1.4R. Fees make the real number a little worse.