Measured, including the nulls.
Every claim the Terminal makes was tested on its own recorded data first, and the tests that found nothing are published with the same care as the ones that found something. Methods and scripts follow each note when it is written up.
- 2026-09-13
Cross-venue Bitcoin arbitrage is dead, and zero-fee venues have manufactured tapes
7,199 samples of the top-of-book gap between two USDT venues: 0.79 basis points median against a cheapest possible round trip near 2. Across 258 liquid altcoins on seven venues, walked at $10,000, nine of ten candidates were negative. On every venue advertising zero fees, a real counterparty and a zero fee did not coexist.
- 2026-09-12
Residual lead–lag between venues is real and smaller than one tick
Cross-venue residual reversion on BTC measured at +2.18 basis points: statistically real, under one tick, and half the cost floor of the cheapest execution venue tested. Overlapping signals inflated the apparent edge by 2.5 to 6.6 times until they were de-duplicated.
- 2026-09-08
No level type beats a placebo over 365 days
Hand-placed and automatic levels (volume, time, structure) were scored against random levels over a year of one-minute Bitcoin bars. None separated from the placebo. The engine's level weights are zero as a measured fact.
- 2026-09-12
The divergence detector was broken, and the null that scored it is withdrawn
Pivots two bars wide and no requirement that the oscillator be anywhere in particular. The earlier study that found no divergence edge measured the broken detector and is void; the rebuilt detector is being re-measured.
Full write-ups, with methods, are being prepared from the working notes. Numbers above are as measured on the dates shown.