What we said before the run
Before anything runs here, we write down what we expect it to show and file it. After the run we score ourselves against it and publish the score, the misses with the hits. It is the only way you can tell a test from a story told afterward.
The last two reader runs: 14 predictions, 10 hit, 2 split, 2 missed. Every miss below says why.
READER 03, part 2
A reader's VWAP-band rule on ES and NQ, 2008 to 2026, the default hours. Filed Sep 20, before either run. The page it became →
| What we said before the run | What happened | |
|---|---|---|
| Both markets fail, score under 45. | hit | 9 and 9. |
| The stop takes more trades than the 2R target on both. | hit | 668 stops against 221 targets on NQ; 943 against 299 on ES. |
| The breakeven (the stop moved to entry after VWAP trades) is the largest single exit on at least one market. | miss | It was one of the two smallest on both (84 on NQ, 135 on ES). Most trades are stopped before VWAP is reached. The reasoning was wrong, not the count. |
| More than 1,000 of 5,000 coin flips match or beat it on the years it never saw, on both. | hit | 4,945 and 4,934. |
| NQ loses after costs; ES lands within $30,000 of zero. | split | NQ lost as said. ES lost $71,892, outside the band. |
| On NQ the recent years (2023 on) are better per trade than the years before. | miss | Backwards: the recent years were its worst per trade on both markets. |
| The two VWAP readings the rule uses sit under 3 points apart on NQ and under 1 on ES. | hit | 0.165 and 0.056 points on average. |
READER 03, part 3
The same rule on the trader's own hours, ES and gold, two sessions a day, his close rule. Filed Sep 20, before any of the four runs. The page it became →
| What we said before the run | What happened | |
|---|---|---|
| All four runs fail, score under 45. | hit | 9, 9, 9 and 9. |
| The stop is the largest single exit on all four. | hit | 210, 298, 353 and 298 stops. |
| Losers flattened at the session close outnumber winners held past it and stopped at entry, on all four. | hit | 90 to 64, 130 to 83, 173 to 146, 94 to 81. |
| All four lose after costs. | hit | All four did. |
| More than 1,000 of 5,000 coin flips match or beat it on the years it never saw, on all four. | hit | 3,649, 3,232, 2,668 and 1,370. |
| The trade counts equal the counts measured before the run (498, 686, 884, 666). | hit | They did, with none dropped. |
| On both morning runs the recent years are no better per trade than the years before. | split | True on ES. On gold the recent years were better, and gold's evening run was the only one whose recent years made money. |
The engine work behind them
Each new piece of the engine is declared the same way: what it will do, the tests written first and failing, and what the build will measure, scored after.
- ENGINE 05, a VWAP from 18:00 across midnight: 5 hit, 0 split, 0 missed, 1 still open.
- ENGINE 06, a breakeven set off by a line that moves: 4 hit, 2 split, 0 missed.
- READER 03, part 1, the reader's rule built as a family: 4 hit, 1 split, 1 missed.
- ENGINE 07, the session-close exit: 2 hit, 0 split, 3 missed.
Hit means it came out as written. Split means part of it did. Missed means it did not, and the record says so rather than rounding it into a hit.