Pre-registration
Writing the rule and the pass or fail bar before looking at the result.
To pre-register a test is to write down, with a date, everything that could be bent afterwards: the exact rule, the data it will run on, the costs, and the numbers it must reach to pass. Only then does the test run. If the result misses the bar, the idea fails; nothing is renegotiated.
It sounds like bureaucracy and it is the single most protective habit in trading research. Without it, the natural move after a weak result is to adjust something, run again, and keep the version that worked, which manufactures a strategy fitted to the past. With it, the researcher gives up that freedom in advance, and a pass means something.
Every logbook entry states the gates that were fixed before the run. When an entry says a side of the trade "was not pre-registered", it means: we saw it in the data afterwards, so it does not count.
Explore the diagram (scroll horizontally on small screens)
Where this word comes up
First needed in the log entry H-042 · The post-earnings bounce did not survive our cost assumptions.
From the same entry: Backtest, Basis point, Buy the dip, Correction for looking twice, Gross edge, Index membership, Large caps, Net return, Post-earnings drift, Published anomaly, Quarterly results, Real costs, RSI indicator, Short side, Survivors only, Terciles.