Rival: the complement's lift is a longshot artefact; test it at 14/1 or shorter

Reasoning

Why this hypothesis exists: what the lab saw, the question it asked, the mechanism it proposed, what it predicted, and the result that would prove it wrong. Written before the test was run.

1. Observed
   • H-46173ef7 (Rival: the broad wide-trip lift is a longshot artefact; test it at 14/) passed discovery: A/E 1.18 on 915 bets (p 0.015), same-race control 0.96. O-2af9c9fd
   • H-3cac18e4 (Rival complement: wide-trip narrow Flat defeats without the early-setb) passed discovery: A/E 1.16 on 786 bets (p 0.05), same-race control 0.97. O-91ddf89c
   • H-3cac18e4 (Rival complement: wide-trip narrow Flat defeats without the early-setb) survived falsification: 5 of 6 attacks left it standing; it is frozen for its one vault query. O-cf94e89a
2. AskedDoes the complement (wide-trip narrow defeats without the early-setback-and-recovery story) still beat its prices at 14/1 or shorter, once the 8 longshot winners are taken out? answered
3. MechanismAny horse that raced wide covered extra ground, so a narrow defeat understates its ability even when it was not caught wide early and did not have to fight back into the race.
4. Why the market misses itThe form line reads as a plain close defeat. Comment-level trip detail gets little weight in the price unless the comment tells the dramatic story of being caught wide early and fighting back.
5. Predictedpositive, A/E 1.05 to 1.15 on about 600 to 700 bets
6. Falsified ifA/E at or below 1.02 on 500+ bets, or no better than its same-race control. Either result would mean the trip edge needs the early-setback story and the complement's lift came from a few longshots.

Mechanism and rationale

Any horse that raced wide covered extra ground, so a narrow defeat understates its ability even when it was not caught wide early and did not have to fight back into the race.

The rule

race_type = 'Flat' AND last_raced_wide = 1 AND last_position > 1 AND last_beaten_lengths <= 5 AND sp_decimal <= 15 AND (last_early_setback IS NULL OR last_early_setback = 0 OR last_recovered IS NULL OR last_recovered = 0 OR last_slow_start IS NULL OR last_slow_start = 1)

Reads: race_type, last_raced_wide, last_position, last_beaten_lengths, sp_decimal, last_early_setback, last_recovered, last_slow_start

Pre-registration

Spec hash (sha256 of the canonical pre-spec)

e07a4741b5e4a2fb0eec298472bbbb8d487f4ec2731c023142f96f57e2513dff

Frozen

17 Oct, 04:41

Nothing in the definition can change after this.

Confounders

• The 14/1 cap may remove most of the complement's ROI. Earlier work found the 16/1+ band carried much of the broad rule's lift (1.95 on 204 from 11 wins).
• The rule overlaps the frozen broad ≤14/1 rule. It is a subset of that rule's runners, so a pass is not independent confirmation.
• A share of IRE runners could inflate the pooled figure (GB alone was 1.086 for the bare wide trip). The GB-only refinement checks this.
• A general close-defeat or runner-up effect could stand in for the wide trip. The not-wide control checks this.
• The raced-wide flag depends on how comments are written. Its coverage and wording may vary by period or by comment writer.

Control

{
  "description": "The same-race runners should score about 0.96 to 1.00, as they did for the broad wide-trip rule, so the lift is not a property of the races. A second control is needed to separate the wide-trip effect from a plain close-defeat effect: Flat runners beaten 5 lengths or less last time that did NOT race wide, at 14/1 or shorter. If the wide-trip complement does not clearly beat this group, the edge belongs to close defeats in general, not to ground lost.",
  "rule": "race_type = 'Flat' AND (last_raced_wide IS NULL OR last_raced_wide = 0) AND last_position > 1 AND last_beaten_lengths <= 5 AND sp_decimal <= 15"
}

Discovery result

pass broad rule (niche) · 17 Oct, 04:36

Bets

701

132 winners

A/E

1.09

95%: 0.91 to 1.29

SP return

-4.6%

level stakes

Betfair SP return

+13.0%

701 bets with a BSP

Strike rate 18.8%. p = 0.17. Same-race control: the other runners in these races had A/E 0.98, so the selection beat its own races. Zone: discovery. Evaluator eval-1.0, data 6a4e98be18e83933

Falsification

Each attack is a deliberate attempt to kill the idea. 4 of 6 survived.

• price_band: Strip the 9/1-14/1 band: 8/1 or shorter onlypass

  482 bets · A/E 1.04 (0.86 to 1.26) · control 0.98 · 17 Oct, 04:41

  Breakdown

  betsA/E0.71.01.3

  2025-10-01 to 2026-04-272020.96

  after 2026-04-272801.10

• control: Not-wide close defeats in the same 9/1-14/1 bandfail

  6,129 bets · A/E 0.98 (0.89 to 1.07) · control 1.00 · 17 Oct, 04:41

Sealed holdout (vault)

Not queried. Only a frozen idea can spend its one query against the 2020-25 holdout.

Forward record

• forward vs bar {'min_bets': 150, 'min_ae': 1.0, 'max_p': 0.2}info

  0 bets · · 17 Oct, 04:41

  Breakdown

  betsA/E0.71.01.3

  forward0–

Other experiments

• neighbour: > 1 -> 0 (-20%)pass

  901 bets · A/E 1.10 (0.94 to 1.27) · control 0.98 · 17 Oct, 04:41

  Breakdown

  betsA/E0.71.01.3

  2025-10-01 to 2026-04-273841.10

  after 2026-04-275171.10

• neighbour: > 1 -> 2 (+20%)pass

  516 bets · A/E 1.17 (0.95 to 1.43) · control 0.97 · 17 Oct, 04:41

  Breakdown

Lineage

Agent runs

• designer v1ok · 30s · 17 Oct, 04:32

  Output (schema-checked)

  {
    "rule": "race_type = 'Flat' AND last_raced_wide = 1 AND last_position > 1 AND last_beaten_lengths <= 5 AND sp_decimal <= 15 AND (last_early_setback IS NULL OR last_early_setback = 0 OR last_recovered IS NULL OR last_recovered = 0 OR last_slow_start IS NULL OR last_slow_start = 1)",
    "refinements": [
      {
        "label": "GB only (removes IRE share confounder)",
        "rule": "race_type = 'Flat' AND last_raced_wide = 1 AND last_position > 1 AND last_beaten_lengths <= 5 AND sp_decimal <= 15 AND (last_early_setback IS NULL OR last_early_setback = 0 OR last_recovered IS NULL OR last_recovered = 0 OR last_slow_start IS NULL OR last_slow_start = 1) AND region = 'GB'"
      }
    ],
    "control": {
      "description": "The same-race runners should score about 0.96 to 1.00, as they did for the broad wide-trip rule, so the lift is not a property of the races. A second control is needed to separate the wide-trip effect from a plain close-defeat effect: Flat runners beaten 5 lengths or less last time that did NOT race wide, at 14/1 or shorter. If the wide-trip complement does not clearly beat this group, the edge belongs to close defeats in general, not to ground lost.",
      "rule": "race_type = 'Flat' AND (last_raced_wide IS NULL OR last_raced_wide = 0) AND last_position > 1 AND last_beaten_lengths <= 5 AND sp_decimal <= 15"
    },
    "pass_criteria": {
      "min_bets": 300,
      "min_ae": 1.05,
      "max_p": 0.05
    },
    "forward_bar": {
      "min_bets": 150,
      "min_ae": 1,
      "max_p": 0.2
    },
    "confounders": [
      "The 14/1 cap may remove most of the complement's ROI. Earlier work found the 16/1+ band carried much of the broad rule's lift (1.95 on 204 from 11 wins).",
      "The rule overlaps the frozen broad ≤14/1 rule. It is a subset of that rule's runners, so a pass is not independent confirmation.",
      "A share of IRE runners could inflate the pooled figure (GB alone was 1.086 for the bare wide trip). The GB-only refinement checks this.",
      "A general close-defeat or runner-up effect could stand in for the wide trip. The not-wide control checks this.",
      "The raced-wide flag depends on how comments are written. Its coverage and wording may vary by period or by comment writer."
    ],
    "mechanism": "A horse that raced wide covered extra ground. A defeat by 5 lengths or less therefore understates its ability, even if it was not caught wide early and did not have to fight back into the race. The lost ground is a cost that can be measured and does not repeat, so the horse should run closer to its true level next time.",
    "market_miss": "The form line reads as a plain close defeat. Bettors and odds compilers give comment-level trip detail little weight unless the comment tells the dramatic story of being caught wide early and fighting back. A quiet 'raced wide' note on a close beaten run is underpriced, including at shorter prices where most money goes.",
    "prediction": {
      "direction": "positive",
      "size": "A/E 1.05 to 1.15 on about 600 to 700 bets at 14/1 or shorter",
      "falsified_if": "Falsified if A/E is at or below 1.02 on 500 or more bets, or no better than the same-race control or the not-wide close-defeat control. Either result would mean the trip edge needs the early-setback story and the complement's lift came from longshots."
    },
    "notes": "I made one fix to the sketch. The parent rule required last_slow_start = 0, so a correct complement must also keep runners whose last_slow_start is NULL. I added 'last_slow_start IS NULL' to the OR clause. I left out a tighter price-band refinement: this test is about the ≤14/1 cap itself, and adding a narrower band would be slicing."
  }

Timeline

1. Job 3009 finished

   17 Oct, 04:41 · engine:compute-1

2. forward result for H-8caad71f (forward vs bar {'min_bets': 150, 'min_ae': 1.0, 'max_p': 0.2}): info

   17 Oct, 04:41 · engine:evaluator

3. compute-1 took job 3009 (forward_check, attempt 1)

   17 Oct, 04:41 · compute-1

4. Queued forward_check on the compute lane: daily forward check of a shadow system

   17 Oct, 04:41 · engine:scheduler

5. Job 2976 finished

   17 Oct, 04:41 · engine:compute-1

6. hypothesis note H-8caad71f

   17 Oct, 04:41 · engine:perturb

7. perturbation result for H-8caad71f (neighbour: <= 15 -> 18 (+20%)): pass

   17 Oct, 04:41 · engine:perturb

8. perturbation result for H-8caad71f (neighbour: <= 15 -> 12 (-20%)): pass

   17 Oct, 04:41 · engine:perturb

9. perturbation result for H-8caad71f (neighbour: <= 5 -> 6 (+20%)): pass

   17 Oct, 04:41 · engine:perturb

10. perturbation result for H-8caad71f (neighbour: <= 5 -> 4 (-20%)): pass

    17 Oct, 04:41 · engine:perturb

11. perturbation result for H-8caad71f (neighbour: > 1 -> 2 (+20%)): pass

    17 Oct, 04:41 · engine:perturb

12. perturbation result for H-8caad71f (neighbour: > 1 -> 0 (-20%)): pass

    17 Oct, 04:41 · engine:perturb

13. compute-1 took job 2976 (mutate, attempt 1)

    17 Oct, 04:41 · compute-1

14. Job 3004 finished

    17 Oct, 04:41 · engine:compute-1

15. H-8caad71f Frozen → SHADOW vault skipped: last_raced_wide, last_beaten_lengths, last_early_setback, last_recovered, last_slow_start do not exist before the discovery zone; track it forward instead

    17 Oct, 04:41 · engine:orchestrator

16. compute-1 took job 3004 (vault, attempt 1)

    17 Oct, 04:41 · compute-1