Test-mechanism: the 7f+ wide-trip lift is larger after a big-field last run

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-fba909ba (Rival: the 7f+ wide-trip lift is the story runners repackaged; test th) passed discovery: A/E 1.2 on 462 bets (p 0.044), same-race control 0.95. O-1403a720
   • H-e6917785 (Test-mechanism: the wide-trip lift needs a last run with bends (7f+)) survived falsification: 5 of 6 attacks left it standing; it is frozen for its one vault query. O-ce53aa8c
2. AskedIs the lift larger when the last run had a big field, where a wide horse must cover more extra ground round the bends? answered
3. MechanismIn a big field, a horse racing wide is often three or four horses off the rail round every bend, so it covers several extra lengths. In a small field, wide usually means one off the rail.
4. Why the market misses itThe comment just says 'wide' whatever the field size, and the beaten distance looks the same either way. Nothing in the form shows how much extra ground the horse covered.
5. Predictedpositive, A/E 1.30 to 1.45 on 250+ bets, with the under-12-runner half below 1.2
6. Falsified ifThe 12+ runner cut scores at or below 1.26, or last runs with fewer than 12 runners score as high.
7. Rule sketch

Mechanism and rationale

In a big field, a horse racing wide is often three or four horses off the rail round every bend, so it covers several extra lengths. In a small field, wide usually means one off the rail.

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_distance_yards >= 1540 AND last_field_size >= 12

Reads: race_type, last_raced_wide, last_position, last_beaten_lengths, sp_decimal, last_distance_yards, last_field_size

Pre-registration

Spec hash (sha256 of the canonical pre-spec)

985e210dd0e2683a7bb7e58b90db07fda733afed54d35a6cf5f9d5b0573ee419

Frozen

17 Oct, 04:48

Nothing in the definition can change after this.

Confounders

• Big fields are mostly handicaps, so the lift may be a handicap or rating effect, not a ground-loss effect. The handicap-only refinement and the small-field control check this.
• Big fields are more common on turf and in summer. The earlier 10f+ cut was 1.33 after 27 April 2026 and 0.90 before it, so seasonal or surface effects may explain the result.
• Big fields mean longer prices within the 14/1 cap. The sample may lean toward 8/1 to 14/1, where the market is less efficient whatever the story.
• IRE dependence: earlier wide-trip cuts were IRE-heavy (IRE 1.53 against GB 1.12). The GB-only refinement checks this.
• The cut overlaps heavily with the frozen 7f+ ≤14/1 rule (1.259 on 643) and the H-244b221b sharpen, so it is not independent evidence.
• Field size at 12+ may stand for bigger, more competitive races and higher class, not the extra lengths covered.

Control

{
  "description": "Other runners in the same races should be at about par (roughly 0.93 to 1.0, as in the earlier 7f+ cuts at 0.94 to 0.95). That shows the lift belongs to the selected horses and not to the races. The key mechanism control is the same rule restricted to small-field last runs (under 12 runners). If extra ground covered drives the effect, that half should score clearly lower than the 12+ cut, below 1.2. If it scores as high or higher, field size is not the mechanism.",
  "rule": "race_type = 'Flat' AND last_raced_wide = 1 AND last_position > 1 AND last_beaten_lengths <= 5 AND sp_decimal <= 15 AND last_distance_yards >= 1540 AND last_field_size < 12"
}

Discovery result

pass broad rule · 17 Oct, 04:47

Bets

227

52 winners

A/E

1.35

95%: 1.01 to 1.77

SP return

+16.3%

level stakes

Betfair SP return

+41.6%

226 bets with a BSP

Strike rate 22.9%. p = 0.021. Same-race control: the other runners in these races had A/E 0.92, 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 5 survived.

• control: Control: big-field 7f+ narrow defeat WITHOUT the wide tripfail

  3,586 bets · A/E 0.99 (0.91 to 1.07) · control 1.00 · 17 Oct, 04:48

  Breakdown

  betsA/E0.71.01.3

  2025-10-01 to 2026-04-271,2330.94

  after 2026-04-272,3531.01

• perturbation: Perturb field-size cut down to 10+ runnerspass

  387 bets · A/E 1.23 (0.97 to 1.52) · control 0.95 · 17 Oct, 04:48

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': 100, 'min_ae': 1.05, 'max_p': 0.2}info

  0 bets · · 17 Oct, 04:49

  Breakdown

  betsA/E0.71.01.3

  forward0–

Other experiments

• neighbour: > 1 -> 0 (-20%)pass

  293 bets · A/E 1.35 (1.06 to 1.71) · control 0.92 · 17 Oct, 04:49

  Breakdown

  betsA/E0.71.01.3

  2025-10-01 to 2026-04-271141.27

  after 2026-04-271791.41

• neighbour: > 1 -> 2 (+20%)pass

  169 bets · A/E 1.32 (0.93 to 1.83) · control 0.94 · 17 Oct, 04:49

  Breakdown

Lineage

Agent runs

• designer v1ok · 25s · 17 Oct, 04:43

  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_distance_yards >= 1540 AND last_field_size >= 12",
    "refinements": [
      {
        "label": "Dose-response: 14+ runners last time",
        "rule": "race_type = 'Flat' AND last_raced_wide = 1 AND last_position > 1 AND last_beaten_lengths <= 5 AND sp_decimal <= 15 AND last_distance_yards >= 1540 AND last_field_size >= 14"
      },
      {
        "label": "Handicaps only (big fields are mostly handicaps)",
        "rule": "race_type = 'Flat' AND last_raced_wide = 1 AND last_position > 1 AND last_beaten_lengths <= 5 AND sp_decimal <= 15 AND last_distance_yards >= 1540 AND last_field_size >= 12 AND is_handicap = 1"
      },
      {
        "label": "GB only (checks IRE dependence)",
        "rule": "race_type = 'Flat' AND last_raced_wide = 1 AND last_position > 1 AND last_beaten_lengths <= 5 AND sp_decimal <= 15 AND last_distance_yards >= 1540 AND last_field_size >= 12 AND region = 'GB'"
      }
    ],
    "control": {
      "description": "Other runners in the same races should be at about par (roughly 0.93 to 1.0, as in the earlier 7f+ cuts at 0.94 to 0.95). That shows the lift belongs to the selected horses and not to the races. The key mechanism control is the same rule restricted to small-field last runs (under 12 runners). If extra ground covered drives the effect, that half should score clearly lower than the 12+ cut, below 1.2. If it scores as high or higher, field size is not the mechanism.",
      "rule": "race_type = 'Flat' AND last_raced_wide = 1 AND last_position > 1 AND last_beaten_lengths <= 5 AND sp_decimal <= 15 AND last_distance_yards >= 1540 AND last_field_size < 12"
    },
    "pass_criteria": {
      "min_bets": 200,
      "min_ae": 1.1,
      "max_p": 0.05
    },
    "forward_bar": {
      "min_bets": 100,
      "min_ae": 1.05,
      "max_p": 0.2
    },
    "confounders": [
      "Big fields are mostly handicaps, so the lift may be a handicap or rating effect, not a ground-loss effect. The handicap-only refinement and the small-field control check this.",
      "Big fields are more common on turf and in summer. The earlier 10f+ cut was 1.33 after 27 April 2026 and 0.90 before it, so seasonal or surface effects may explain the result.",
      "Big fields mean longer prices within the 14/1 cap. The sample may lean toward 8/1 to 14/1, where the market is less efficient whatever the story.",
      "IRE dependence: earlier wide-trip cuts were IRE-heavy (IRE 1.53 against GB 1.12). The GB-only refinement checks this.",
      "The cut overlaps heavily with the frozen 7f+ ≤14/1 rule (1.259 on 643) and the H-244b221b sharpen, so it is not independent evidence.",
      "Field size at 12+ may stand for bigger, more competitive races and higher class, not the extra lengths covered."
    ],
    "mechanism": "On a round-course Flat trip (7f+), a horse 'raced wide' in a big field (12+) is usually three or four off the rail round every bend, so it covers several lengths more than the winner. In a small field, 'wide' usually means one off the rail and costs little. A narrow defeat (≤5 lengths) after a big-field wide trip therefore hides more ability than the same defeat after a small-field wide trip. The lift should grow with the field size of the last run.",
    "market_miss": "The form comment says 'wide' whatever the field size, and the beaten distance looks the same either way. Nothing in the standard form shows how much extra ground was covered. Bettors and odds-setters treat every 'wide' excuse alike, so they underrate the big-field cases where the excuse was worth several lengths.",
    "prediction": {
      "direction": "positive",
      "size": "A/E 1.30 to 1.45 on 250+ bets for the 12+ runner cut, with the under-12 control below 1.2",
      "falsified_if": "The 12+ runner cut scores A/E at or below 1.26 (no better than the parent 7f+ cut), or the under-12-runner control scores as high as the 12+ cut."
    },
    "notes": "The rule is the investigator's sketch unchanged. Every column is pre-off and describes the last run, and sp_decimal <= 15 is the 14/1 cap. I set discovery stricter than the floor (200 bets, A/E 1.10) because the prediction is 250+ bets at 1.30+. The real test of the mechanism is the comparison with the parent 1.259 and the small-field control, not just passing the floor. Read forward results with the seasonal and turf finding in mind."
  }

Timeline

1. Job 3067 finished

   17 Oct, 04:49 · engine:compute-1

2. hypothesis note H-cf793824

   17 Oct, 04:49 · engine:perturb

3. perturbation result for H-cf793824 (neighbour: <= 15 -> 18 (+20%)): pass

   17 Oct, 04:49 · engine:perturb

4. perturbation result for H-cf793824 (neighbour: <= 15 -> 12 (-20%)): pass

   17 Oct, 04:49 · engine:perturb

5. perturbation result for H-cf793824 (neighbour: <= 5 -> 6 (+20%)): pass

   17 Oct, 04:49 · engine:perturb

6. perturbation result for H-cf793824 (neighbour: <= 5 -> 4 (-20%)): pass

   17 Oct, 04:49 · engine:perturb

7. perturbation result for H-cf793824 (neighbour: > 1 -> 2 (+20%)): pass

   17 Oct, 04:49 · engine:perturb

8. perturbation result for H-cf793824 (neighbour: > 1 -> 0 (-20%)): pass

   17 Oct, 04:49 · engine:perturb

9. compute-1 took job 3067 (mutate, attempt 1)

   17 Oct, 04:49 · compute-1

10. Queued mutate on the compute lane: parameter stability: do neighbouring thresholds hold?

    17 Oct, 04:49 · engine:scheduler

11. Job 3060 finished

    17 Oct, 04:49 · engine:compute-1

12. forward result for H-cf793824 (forward vs bar {'min_bets': 100, 'min_ae': 1.05, 'max_p': 0.2}): info

    17 Oct, 04:49 · engine:evaluator

13. compute-1 took job 3060 (forward_check, attempt 1)

    17 Oct, 04:49 · compute-1

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

    17 Oct, 04:49 · engine:scheduler

15. Job 3056 finished

    17 Oct, 04:48 · engine:compute-1

16. H-cf793824 Frozen → SHADOW vault skipped: last_raced_wide, last_beaten_lengths do not exist before the discovery zone; track it forward instead

    17 Oct, 04:48 · engine:orchestrator