5927sign in
5927
density: 34.8% · nothing connects
34.80%

Below the threshold, clusters stay local.

Add density. Still nothing crosses.

58.9 percent. Close. Not enough.

One connected path. The system flips.

percolation. a sharp threshold. either it flips or it does not.

Markets have thresholds too. We study scheduled events where someone must act at a known time. Then we measure whether price behavior around these moments is real or noise.

scheduled events · sample

Every hypothesis is registered before we look at data. Population, thresholds and kill criteria are frozen first. If the numbers miss the bar, the thesis dies.

thesis check · one fails, the thesis dies
population
cost model
kill criterion
regime split
significance

We defend against self deception. Fixed grids. Bootstrap on event level. Consistency across regimes. Censored data is reconstructed, never ignored.

drag the threshold
t = 0.35
fail
17 / 120
bar: 15%

Survivors are not trusted either. They trade on paper, then on demo with real order mechanics and measured slippage. Only evidence promotes a thesis.

Most theses die. That is the design.
5927 · a private research project · not investment advice · no offering
contact@5927.ai