Twenty candidate pairs: classify valid or invalid, including equal highs, wick breaches and shortened sessions.
Historical replay
A continuous period with at least 100 qualifying signals. Include expired and filtered signals and track the reason for each rejection.
Resolve intraday ambiguity
A daily candle can touch both triggers, the stop and the target. Use reliable intraday bid/ask or tick data; if the sequence is unknown, flag it and report conservative outcomes. Never choose the winning sequence by assumption.
Reserve an unseen period
If you change the buffer, target or filter after seeing results, the old sample is no longer unseen. Label the revision.
Forward demonstration
At least eight weeks.
Honest note
Evidence before confidence: complete records, consistent execution and positive net expectancy across separate samples with tolerable drawdowns. If that is missing, keep investigating or abandon the method. Do not use a few wins as a reason to increase size.
Key takeaway
No sample size or good month proves a durable edge.
Try it
Pick a start date for your 100-signal replay and write it in your ticket.
This chapter is in the full ebook
Chapters 1 and 2 are free. Get all 21 chapters with every calculator and the journal.