Rival: the 7f+ wide-trip lift is a turf effect; all-weather last runs at par

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-233f3726 (Test-mechanism: more bends, more lift; wide-trip narrow defeats over 1) pointed the predicted way but failed the bar: A/E 1.17 on 206 bets (p 0.173), control 0.96. O-5a42d28d
   • H-233f3726 (Test-mechanism: more bends, more lift; wide-trip narrow defeats over 1): the period 'after 2026-04-27' carries the effect (A/E 1.33) while the rest score 2025-10-01 to 2026-04-27 0.9. O-6fe083c8
2. AskedIs the 7f+ wide-trip lift a turf-season effect? The 10f+ cut is 1.33 after 27 April 2026 and 0.90 before it, which suggests winter all-weather runs score at par. answered
3. MechanismOn big turf round courses, a wide trip means extra ground on long, often uneven bends, and a wide trip on soft ground can cost much more. All-weather ovals are tight and uniform, the extra ground is smaller, and kickback can even favour racing wide.
4. Why the market misses itMost GB all-weather tracks publish sectionals and tracking data, so a wide trip there shows up in widely used figures and gets priced. On turf the wide trip is often only in the comment, and a narrow defeat makes the run look ordinary.
5. Predictedpositive, A/E 1.28 to 1.40 on 350+ turf bets, with all-weather last runs at or below 1.05
6. Falsified ifThe turf cut scores at or below the parent's 1.26, or all-weather last runs score as high as turf ones.

The question's answer so far: No. The turf-only cut scores 1.19 on 294 (p 0.09), below the 7f+ parent's 1.26, so all-weather last runs carry at least as much lift. The only weak winter period is turf's own (0.68 on 50).

Mechanism and rationale

On big turf round courses, a wide trip means extra ground on long, often uneven bends, and a wide trip on soft ground can cost much more. All-weather ovals are tight and uniform, the extra ground is smaller, and kickback can even favour racing wide.

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 LOWER(last_surface) = 'turf'

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

Pre-registration

Spec hash (sha256 of the canonical pre-spec)

d67bc6f98b3661b67045fe1eb1c0f1eb892ddfb5e0a3957981d0ceb404cf0eaa

Frozen

not frozen

The definition can still change.

Confounders

• Season and surface are confounded. last_raced_wide only exists from 2025-11-01, so turf last runs come almost entirely from April to October 2026, the period where the 10f+ cut scored 1.33 against 0.90 before it. A turf lift may just be that seasonal lift.
• Comment-writing may differ between all-weather and turf meetings, so 'raced wide' may mark a more costly trip on turf.
• The IRE share is larger on turf, and IRE drove much of the earlier wide-trip lift. The GB-only refinement checks this.
• Going: soft turf may carry the effect rather than turf itself.
• Surface switch: horses whose last run was on turf may run today on the all-weather, or the reverse. The turf-to-turf refinement checks this.
• Overlap: this is a subset of the frozen 7f+ ≤14/1 rule (1.259 on 643), so a pass partly restates the parent. The real test is the contrast with the all-weather complement.

Control

{
  "description": "The rival test is the all-weather complement of the parent 7f+ ≤14/1 cut: same conditions, last run not on turf. The hypothesis predicts it scores at or below 1.05 while the turf cut scores 1.28 or more. The usual same-race control (the other runners in the selected races) should sit near 1.0 (the parent's control was 0.94). If it is clearly above 1.0, the lift belongs to the races and not to the wide-trip horses.",
  "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_surface IS NOT NULL AND LOWER(last_surface) <> 'turf'"
}

Discovery result

fail broad rule · 17 Oct, 04:47

Bets

294

63 winners

A/E

1.19

95%: 0.92 to 1.53

SP return

-1.3%

level stakes

Betfair SP return

+16.5%

293 bets with a BSP

Strike rate 21.4%. p = 0.092. Same-race control: the other runners in these races had A/E 0.95, so the selection beat its own races. Zone: discovery. Evaluator eval-1.0, data 6a4e98be18e83933

Falsification

No attacks yet. Only promising ideas are attacked.

Sealed holdout (vault)

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

Forward record

No forward record. Shadow tracking starts after the vault and counts only races run after the freeze.

Lineage

Agent runs

• designer v1ok · 30s · 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 LOWER(last_surface) = 'turf'",
    "refinements": [
      {
        "label": "GB only (removes the 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_distance_yards >= 1540 AND LOWER(last_surface) = 'turf' AND region = 'GB'"
      },
      {
        "label": "Turf to turf (today also on turf, so the surface switch does not carry the effect)",
        "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 LOWER(last_surface) = 'turf' AND LOWER(surface) = 'turf'"
      }
    ],
    "control": {
      "description": "The rival test is the all-weather complement of the parent 7f+ ≤14/1 cut: same conditions, last run not on turf. The hypothesis predicts it scores at or below 1.05 while the turf cut scores 1.28 or more. The usual same-race control (the other runners in the selected races) should sit near 1.0 (the parent's control was 0.94). If it is clearly above 1.0, the lift belongs to the races and not to the wide-trip horses.",
      "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_surface IS NOT NULL AND LOWER(last_surface) <> 'turf'"
    },
    "pass_criteria": {
      "min_bets": 150,
      "min_ae": 1.15,
      "max_p": 0.05
    },
    "forward_bar": {
      "min_bets": 100,
      "min_ae": 1.05,
      "max_p": 0.2
    },
    "confounders": [
      "Season and surface are confounded. last_raced_wide only exists from 2025-11-01, so turf last runs come almost entirely from April to October 2026, the period where the 10f+ cut scored 1.33 against 0.90 before it. A turf lift may just be that seasonal lift.",
      "Comment-writing may differ between all-weather and turf meetings, so 'raced wide' may mark a more costly trip on turf.",
      "The IRE share is larger on turf, and IRE drove much of the earlier wide-trip lift. The GB-only refinement checks this.",
      "Going: soft turf may carry the effect rather than turf itself.",
      "Surface switch: horses whose last run was on turf may run today on the all-weather, or the reverse. The turf-to-turf refinement checks this.",
      "Overlap: this is a subset of the frozen 7f+ ≤14/1 rule (1.259 on 643), so a pass partly restates the parent. The real test is the contrast with the all-weather complement."
    ],
    "mechanism": "On big turf round courses, a wide trip at 7f or more means extra ground on long, often uneven bends, and on soft ground it costs more. A narrowly beaten (≤5 lengths) horse that raced wide on turf therefore ran better than the result shows. All-weather ovals are tight and uniform, so the extra ground is smaller, and kickback can even make racing wide an advantage. The wide-trip excuse is worth less there.",
    "market_miss": "Most GB all-weather tracks publish sectionals and tracking data, so a wide trip there shows up in widely used figures and gets priced. On turf the wide trip often appears only in the comment, and a narrow defeat makes the run look ordinary, so the horse is under-bet next time.",
    "prediction": {
      "direction": "positive",
      "size": "A/E 1.28 to 1.40 on 350+ turf bets, with the all-weather complement at or below 1.05",
      "falsified_if": "The turf cut scores at or below the parent's 1.26 or misses p 0.05, or the all-weather complement scores as high as the turf cut (within about 0.10, or itself above 1.15)."
    },
    "notes": "I kept the rule sketch as drafted. It uses only pre-off columns; LOWER() makes the surface match case-proof. The control rule uses last_surface <> 'turf' (and not NULL) because the exact all-weather label is not confirmed. A caveat for reading results: last_raced_wide is only available from 2025-11-01, so the turf/all-weather split is largely a summer/winter split. Pre-register that a turf lift confined to 2026 summer cannot by itself tell the surface apart from the season. Forward tracking through the coming all-weather winter and next turf season is the real arbiter."
  }

Timeline

1. Job 3022 finished

   17 Oct, 04:47 · engine:compute-1

2. H-2c6644e4 Discovery → KILLED fails: p 0.0923, needs 0.05 or lower; periods [0.68, 1.28], each needs 1.0 (not a niche pass: a period points the other way [0.68, 1.28])

   17 Oct, 04:47 · engine:evaluator

3. discovery result for H-2c6644e4 (broad rule): fail

   17 Oct, 04:47 · engine:evaluator

4. H-2c6644e4 Designed → RESEARCHdiscovery measuring on the discovery zone

   17 Oct, 04:47 · engine:orchestrator

5. compute-1 took job 3022 (evaluate, attempt 1)

   17 Oct, 04:47 · compute-1

6. Job 3020 finished

   17 Oct, 04:43 · engine:ai-1

7. Queued evaluate on the compute lane: discovery run of the broad rule

   17 Oct, 04:43 · engine:scheduler

8. H-2c6644e4 Proposed → RESEARCHdesigned pre-spec written and hashed

   17 Oct, 04:43 · agent:designer@v1

9. ai-1 took job 3020 (design, attempt 1)

   17 Oct, 04:43 · ai-1

10. Queued design on the ai lane: write the pre-spec

    17 Oct, 04:43 · engine:scheduler

11. New idea H-2c6644e4: Rival: the 7f+ wide-trip lift is a turf effect; all-weather last runs at par (rival-explanation)

    17 Oct, 04:43 · agent:investigator@v1

Every change is an event with an actor: an agent and its version, an engine, a human, or the importer. Current stage: Killed.