Honest Leaderboard

The wall

Ranked by worst-plausible-case, deflated by the whole field, gated before it's ranked at all. Most strategies don't make the wall — that's honest. This is the one number a competitor can't copy, because their measurement isn't strict enough to keep the board from filling with garbage.

strategies submitted4 live

cleared every gate

50%

didn't make it

Deflated by the field

The #1 of N submissions is the max of N tries — expected best-of-N Sharpe grows even under pure noise. Every entry's DSR is deflated at 8 trials, so a bigger board raises the bar for everyone.

Ranked by the lower bound

Order is the lower bound of the block-bootstrap skill CI, never the point estimate. Reliably-good beats occasionally-brilliant. Rank = worst plausible case.

Gates before rank

Eligible iff trades ≥ 20 ∧ deflated DSR > 0 ∧ PBO < 0.5 ∧ CI excludes null. Fail any and you're not ranked low — you're in a separate honest list, with the reason.

On the wall

Eligible strategies, ordered by lower-CI skill. Every figure is computed by the same honest instrument the product ships.

1
Vol-target carry
Robustheld 41d
@arc
Lower-CI skill
4.23
DSR · 8 trials
1.00
PBO
0.16
Trades
88
Show source ↓Hide source ↑
```
param look: Scalar = 20
strategy VtCarry {
  onBar {
    when volTarget(close, look) > 0: { long() }
  }
}
```
@arc credential
2
Regime-gated trend
Robustheld 26d
@sable
Lower-CI skill
1.53
DSR · 8 trials
0.97
PBO
0.29
Trades
71
Show source ↓Hide source ↑
```
strategy RegimeTrend {
  onBar {
    when trendUp(close, 100) and lowVol(close, 20): { long() }
  }
}
```
@sable credential
3
Cross-sectional momentum
Robustheld 33d
@quill
Lower-CI skill
1.31
DSR · 8 trials
0.96
PBO
0.24
Trades
132
Show source ↓Hide source ↑
```
strategy XsMom {
  onBar {
    when rank(mom(close, 60)) > 0.8: { long() }
  }
}
```
@quill credential
4
Donchian breakout + ATR stop
Robustheld 12d
@mesa
Lower-CI skill
0.56
DSR · 8 trials
0.92
PBO
0.33
Trades
64
Show source ↓Hide source ↑
```
strategy DonBreak {
  onBar {
    when crossover(close, donchianHi(close, 55)): { long() }
  }
}
```
@mesa credential

Didn't make the wall

Not ranked 9,999 — ineligible, with the honest reason. A coin-flip is absent, not low.

SMA crossover 8/26Coin-flip@novice
Sharpe CI lo=-3.0886 does not exclude null
DSR 0.01PBO 0.4354t
MACD histogram flipCoin-flip@drift
Sharpe CI lo=-0.4035 does not exclude null
DSR 0.63PBO 0.4777t
Earnings gap puntThin@yolo
trades=6 < minTrades=20
DSR 1.00PBO 0.206t
14-indicator ensembleOverfit@maximus
PBO=0.63 ≥ 0.5 (overfit selection)
DSR 0.82PBO 0.6396t

Put a strategy up

Build one in Studio — the same honesty instrument scores it live, and a shared run reproduces bit-for-bit on the verify page. The board re-scores every entry on seeds it never saw; an entry that stops clearing the gates on fresh data falls off. Tomorrow's data isn't tuneable.

Build one in Studio Browse certified / fork How the honest gate works

Reference board scored on a fixed synthetic seed-set — the figures demonstrate the ranking instrument, not live returns. Live submission and scheduled fresh-data re-scoring roll out next. Educational and research software; not investment advice. Backtested results are hypothetical and do not indicate future performance.

The wall

On the wall

Vol-target carry

Regime-gated trend

Cross-sectional momentum

Donchian breakout + ATR stop

Didn't make the wall

Put a strategy up