Loading
/
Every call the panel makes is saved the night it is made, and the record is built so that edits can be detected. This page shows how those calls did, and the older tests the rules came from. The three kinds of result are never mixed.
Not yet beating random picks
In the pre-registered test the rules did not clear the bar in either period. The live record below is the one that counts.
Calls made since launch, scored on what a trader could have done.
Period run once, after the rules were fixed. Hypothetical.
| Period | 2022 to March 2026Hypothetical |
|---|---|
| Pick test: buy and hold 10 sessions | |
| Trades | 5,482 |
| Mean excess over market (10 sessions) | +0.42% |
| Market over the same days | -1.02% |
| t-score (2 or more is solid) | 1.76 |
| Beat this share of random picks | 100.0% |
| Trade plan test | |
| Calls made | 5,482 |
| Closed trades | 3,810 |
| Orders filled | 69.5% |
| Hit rate | 32.4% (30.9% – 33.9%) |
| Average win | +9.47% |
| Average loss | -5.78% |
| Expectancy (R) | -0.19R |
| t-score (2 or more is solid) | -1.39 |
Its picks beat random picks, but the trade plan lost money after costs and the t-score is under 2.
Periods the rules were built on. Hypothetical, likely better than real life.
| Period | 2012-2021Hypothetical | April 2026 onHypotheticalOne day, 30 calls: too few to mean anything. |
|---|---|---|
| Pick test: buy and hold 10 sessions | ||
| Trades | 5,733 | 30 |
| Mean excess over market (10 sessions) | +0.21% | -0.99% |
| Market over the same days | +0.39% | -5.98% |
| t-score (2 or more is solid) | 1.24 | – |
| Beat this share of random picks | 97.3% | 11.0% |
| Trade plan test | ||
| Calls made | 5,746 | 30 |
| Closed trades | 3,442 | 28 |
| Orders filled | 59.9% | 93.3% |
| Hit rate | 43.4% (41.7% – 45.0%) | 0.0% (0.0% – 12.1%) |
| Average win | +8.80% | – |
| Average loss | -5.23% | -7.33% |
| Expectancy (R) | -0.17R | -0.87R |
| t-score (2 or more is solid) | -0.84 | – |
The tests used stocks still listed today, so results are likely overstated.
The tests used the panel-v1 rules. The live panel runs panel-v3, which fills orders more strictly; it has no backtest and is judged on the live record.
A market baseline for live calls is not published yet.