Journal every setup, not just every trade
Log the setups you skipped as well as the ones you took: date, session, level, state reached, why it qualified or failed, planned entry, stop, target and R, and the result. Skipped setups tell you whether your filters help. The Tradeskill journal records simulated trades automatically; add the setup notes yourself.
The practice loop
Replay before you backtest
Chart replay hides the future. Stop at the London open, mark the levels, then step forward candle by candle and make each decision as if it were live. This trains the checklist and exposes vague rules, such as a "strong" candle you cannot define. Fix those before any automated test.
Setup states you can log
A backtest you can trust
Write the rules exactly, so two people would mark the same trades. Use historical OHLCV with realistic spread, fees and slippage. Only use confirmed candles and pivots, so nothing repaints. Develop the rules on one period (in-sample), then test them untouched on a later period (out-of-sample), and repeat that walk forward through time.
- Report the number of trades, win rate, average win and loss, expectancy, maximum drawdown, profit factor and the date range.
- Show the sample size and how uncertain the numbers are: 30 trades say very little.
- Label every statistic as hypothetical, simulated or live.
- Keep each entry method (close, retest, limit) in its own test.
What this playbook does not claim
The Tradeskill has not published a backtest of this playbook, so no win rate or probability is shown anywhere in it. Handwritten estimates such as "about 80% when the 4H agrees" are not evidence unless they come from a documented dataset and method.
Bullish example
- 1Replay a London session and log a bullish sweep setup that reached entry candidate.
- 2Record the plan: entry, stop below the sweep low, target at the Asia high, planned R.
- 3Record the result and whether you followed every rule.
Bearish example
- 1Log a bearish sweep that was blocked because the 4H was bullish.
- 2Record what price did afterwards, without changing the rule.
- 3Review whether the bias filter helped across all logged setups, not just this one.
What confirms it?
- Rules are written precisely enough to repeat.
- Costs are included and labelled.
- Out-of-sample results are reported separately.
What invalidates it?
- Changing rules after seeing the out-of-sample results.
- Using unconfirmed pivots or forming candles in the test.
- Reporting results without sample size or date range.
Common mistakes
- Logging only winning trades.
- Backtesting on the same data used to design the rules.
- Ignoring spread on small stops.
- Quoting a win rate without expectancy or drawdown.
- Mixing entry methods in one result.
Replay exercise
- Run twenty replay sessions on gold with the checklist and log every setup, taken or skipped.
- Calculate expectancy (average R per trade) and the largest losing streak.
- Then pick the next twenty sessions and repeat without changing any rule.
Quick quiz
Pick an answer to see the explanation. Nothing is saved.