Thirteen trades won't tell me if it works. Following the rules will.
How I plan to climb the evaluations one week at a time: trade, log, let the journal get read, change one thing or nothing.
The last time I drew a loop for this newsletter it was day one of the $2,400 test. Six hand-offs, a grader reading 15-minute bars, a weekend test. Three days later I swapped the indicator out from under it.
The loop survived. That's the point of a loop. The thing it improves can change; the way it improves doesn't.
So here is the new one, and the problem it has to solve.
The problem with two months
The zone rule fires about twice a week. Some weeks, nothing. Two months is forty trading days, and at the historical rate that is somewhere between 6 and 19 trades. Call it 13.
Thirteen trades can't tell me whether a change helped.
If the backtest's +0.22R a trade is exactly the truth, the average of 13 trades still lands anywhere from -0.43R to +0.94R. One time in three it reads below zero. It takes about 160 trades, two years at this rate, before luck alone can't put the average under zero.
If I change the indicator because week two was red, I'm steering by noise. Most of what I'd "learn" from the P&L in two months would be the coin, not the rule.
In two months I can't climb on profit. I can climb on how well I follow the rules, and that is checkable on every single trade.
Why that's enough
The backtest already told me where the money leaks. Over five years, the level setups that fired inside the gate lost 0.12R a trade and won one time in five. The ones that fired outside made 0.22R and won one in three. Same setups, same stops, same targets. The only difference was where price sat against yesterday's close.
Every inside trade I take anyway drags the evaluation odds downhill.
Replayed from the five-year record at $200 risk a trade. Follow the rule and about 89 of 100 evaluations pass. Take one in four of the inside signals and it's 75. Take them all and it's a coin flip. The rule never changes along this line. Only how well I follow it.
That slope is the hill. Rule adherence is the climb. And unlike profit, it doesn't need 160 trades to measure. One trade with a tag on it is one data point, and it's exact.
The loop
Five stations, one lap a week. The blue ones run the same every week. Orange is where the week gets judged.
1. Before the open. TradingView draws yesterday's close, the zone and the gate from the VIX. Red background means no entries. Green means my setups are allowed.
2. The trade. My rules, unchanged: a level, a 15-minute trigger, a stop past the bar, a target at least twice the stop, 5 MNQ at most, flat by 4:10.
3. Within five minutes. The trade goes into TradeZella on the "$20k test" strategy. Nineteen rules ticked honestly, including the ones I broke. Three tags: where price was against the zone when I got in, whether I traded with the day's move or against it, and any mistake. The distance from yesterday's close goes in the note.
4. Sunday, 7pm. Claude reads the week through TradeZella's connection and answers three questions. Did I follow the rules, trade by trade? Did the zone still sort winners from losers, or did the inside trades win this week? Is the week's result inside the band the backtest expects for that many trades? The first question has an exact answer. The other two mostly say "too early to tell", and that's a fine answer.
5. Monday. One change, or none. Written down with its test before the week starts. The order never changes: rule adherence first, because that's free money. Then exits, if I'm cutting trades the plan would have held. The gate width is the only real knob, and it doesn't get touched inside this test. Size and a second account come last, and only after the first account has done its job. Default answer: no change.
Then the scoreboard, every Sunday, with the same numbers the review used.
What I do now
- Tag every trade within five minutes. An untagged trade can't be checked, so it counts as a mistake.
- Read the Sunday review as a report card on me, not on the indicator.
- Change one thing a week at most, on paper first, and only in the order above.
- Judge the indicator at the end, on all the trades, not week by week.
Next: The zone: how I draw it before the open.
New posts land on Substack first. Free, one a week. The $20k test has a scoreboard, updated every Sunday.
Subscribe on Substack