One-sided or origin-dependent record

Your record is suspect because it was selected, copied, or defined by the very story it seems to support.

Your trading journal has convinced you that you have a real edge, and the entries back you up: sharp reads, well-timed exits, lessons learned. Then, flipping back through it, you notice something about the entries themselves: you mostly wrote on days you won. The record and the story share an author.

This diagnosis says your record is suspect because it was selected, copied, or defined by the very story it seems to support. Wins got logged with pride; losses became vague "lessons" or silence. The cue is a record that flatters its keeper suspiciously well. Steps: ask how each entry came to exist, and specifically whether a loss had the same chance of being written down as a win; reconstruct the missing side from a source the story couldn't filter, like your brokerage statements, which recorded every trade indifferently; then re-ask the original question against the complete record. What changes is that the edge either survives on unfiltered data, or evaporates, and both outcomes beat trading real money on a curated memoir. The diagnosis doesn't apply when the record was generated blind to outcomes; a log written before results arrive has no favorite.