Rule lab Rule-lab verdicts are gauntlet arithmetic · descriptive, not a recommendation
Bottom lineWrite a rule; we run it through the same gauntlet as our own studies and tell you honestly if it works.
Read-only demo — synthetic numbers that show the verdict FORMAT, not a real run (the demo rule was judged WEAKER-THAN-BENCHMARK on purpose: the lab's job is honest refusal). The owner composes and queues real rules here.
rule SELECT liquid500 WHERE not_extended RANK BY mom12 TAKE 25 HOLD quarterly
WEAKER-THAN-BENCHMARK descriptive-only
| net return/vol | 0.61 |
| gross return/vol | 1.02 |
| flat-cost return/vol | 0.84 |
| max drawdown | -41.0% |
| half 1 (2012-18) | 0.55 |
| half 2 (2019-26) | 0.66 |
| placebo p95?Placebo p95The 95th percentile of the null distribution: the same book rebuilt N times with RANDOM picks (same universe, same size, same costs). A rule must beat this, not merely beat zero — luck has a high bar. source: evlib placebo machinery via rule_lab_executor.Rule lab — the gauntlet vocabulary | 0.40 |
| observed | 0.61 |
| empirical p | 0.02 |
| annual cost % | 6.00 |
| turnover/rebalance | 0.38 |
| benchmark (Nifty-500 net) | 0.89 |
| capacity breakpoint?Capacity breakpointThe AUM (₹ cr) at which participation-cost impact kills the edge — the largest AUM whose net return/vol still beats the benchmark. A rule with no stated capacity is not a result. source: cost_participation grid scan.Rule lab — the gauntlet vocabulary | unstated |
prereg eb64d10999f899b4… · demo — synthetic numbers, not a real run ·
CSV In plain EnglishPlain read: the verdict is the gauntlet's arithmetic on YOUR rule — walk-forward halves, a random-pick placebo, realistic costs, a capacity read — judged against Nifty-500 buy-and-hold. It is a statistical summary of the rule, not advice about any stock.
How the gauntlet judges
- Both halves?Both halvesWalk-forward split 2012-18 vs 2019-26: a rule must beat the benchmark in BOTH halves or it is noise. source: factory.slice_stats.Rule lab — the gauntlet vocabulary: beat the Nifty-500 net benchmark in 2012-18 AND 2019-26, or it is noise.
- Placebo p95?Placebo p95The 95th percentile of the null distribution: the same book rebuilt N times with RANDOM picks (same universe, same size, same costs). A rule must beat this, not merely beat zero — luck has a high bar. source: evlib placebo machinery via rule_lab_executor.Rule lab — the gauntlet vocabulary: beat the 95th percentile of random-selection books on the same tables — not merely beat zero.
- Capacity breakpoint?Capacity breakpointThe AUM (₹ cr) at which participation-cost impact kills the edge — the largest AUM whose net return/vol still beats the benchmark. A rule with no stated capacity is not a result. source: cost_participation grid scan.Rule lab — the gauntlet vocabulary: the AUM where participation impact kills the edge; no stated capacity, no result.
- Prereg gate hash?Prereg gate hashsha256 of the frozen rule text, recorded BEFORE the run; first registration wins. Tamper-evident: re-deriving the hash must match. source: prereg discipline (rule_lab_prereg).Rule lab — the gauntlet vocabulary: the rule text is frozen and hashed before the run; first registration wins.
- Known-dead shapes are cited verbatim from the strategy ledger BEFORE any run — a dead idea can be re-tested, never re-walked silently.