What I was looking for
What sits underneath published trading records that look too good.
What I found
Win rate, median, drawdown so far and agreement are the numbers people screen on, and all of them can look best just before the tail event they never measured.
Reviewing published trading records, one showed 28 positive months out of 28 with a win rate near 74%: those losing trades have to land somewhere, as unclosed losses or a martingale. Another had a 2% max drawdown from a disclosed martingale whose losses simply weren't realised yet.
The move
When a record looks too clean, I look for where its losses must be going.