Rival: turf-to-AW switchers after a short gap carry the same lift, so the layoff is irrelevant

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-ce91b64d (All-weather spring returners carry the calendar lift, not turf returne) passed discovery as a niche: A/E 1.2 on 480 bets (p 0.11), same-race control 0.96. O-0ab1a07a
2. AskedIs it the turf-to-AW switch that is mispriced at any layoff, or does it need a 150+ day absence? answered
3. MechanismIf the market only discounts turf form on the AW, a horse switching surface after a normal 2-8 week gap should be underpriced too. If the cause is trainer-chosen readiness after a break, short-gap switchers should be fairly priced.
4. Why the market misses itUnder the rival, punters distrust turf form on the AW whatever the gap. Under the main idea, the switch alone is well understood and priced.
5. Predictednegative, A/E 0.93 to 1.03 on 1500+ bets
6. Falsified ifA/E of 1.08 or higher on 1500+ bets, which would make the surface switch, not the layoff, the driver
7. Rule sketchrace_type = 'Flat' AND is_handicap = 1 AND surface = 'AW' AND last_surface = 'Turf' AND days_since_last_run BETWEEN 14 AND 60

Mechanism and rationale

If the market only discounts turf form on the AW, a horse switching surface after a normal 2-8 week gap should be underpriced too. If the cause is trainer-chosen readiness after a break, short-gap switchers should be fairly priced.

The rule

race_type = 'Flat' AND is_handicap = 1 AND surface = 'AW' AND last_surface = 'Turf' AND days_since_last_run BETWEEN 14 AND 60

Reads: race_type, is_handicap, surface, last_surface, days_since_last_run

Pre-registration

Spec hash (sha256 of the canonical pre-spec)

50b7926229762855d3eec694bb2a9783bbd218504c014df07e5b249ac914dc19

Frozen

not frozen

The definition can still change.

Confounders

• Short-gap switchers are often late-season (autumn/winter) horses dropping onto softer AW races, so the calendar differs from the spring returner group. The month refinement addresses this.
• Trainer intent differs: a mid-campaign switch is often a placement for an easier mark or ground, while a switch after a break reflects chosen readiness
• Field size, class and overround differ between AW and turf handicaps
• Ireland's AW programme is small and ran below GB in this thread, so pooled results are GB-dominated
• Price-band composition: lifts in this family sat in the short bands while 16/1+ dragged, so a pooled A/E can hide offsetting bands

Control

{
  "description": "Same-race rivals: the other runners in these AW handicaps should score about 0.95-1.00, as usual for the field. The comparison that matters for the question is long-gap turf-to-AW switchers (150+ days) in the same calendar. If the layoff drives the lift, those should score clearly above the short-gap switchers (the earlier March-April AW returner rule gave 1.195 on 480). If the switch alone drives it, the two groups should score about the same.",
  "rule": "race_type = 'Flat' AND is_handicap = 1 AND surface = 'AW' AND last_surface = 'Turf' AND days_since_last_run >= 150"
}

Designer

Discovery result

fail broad rule · 16 Oct, 14:05

Bets

4,045

419 winners

A/E

1.07

95%: 0.97 to 1.18

SP return

-15.7%

level stakes

Betfair SP return

+5.0%

3,717 bets with a BSP

Strike rate 10.4%. p = 0.089. Same-race control: the other runners in these races had A/E 0.97, 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 · 39s · 16 Oct, 14:04

  Output (schema-checked)

  {
    "rule": "race_type = 'Flat' AND is_handicap = 1 AND surface = 'AW' AND last_surface = 'Turf' AND days_since_last_run BETWEEN 14 AND 60",
    "refinements": [
      {
        "label": "Spring calendar match (March-April)",
        "rule": "race_type = 'Flat' AND is_handicap = 1 AND surface = 'AW' AND last_surface = 'Turf' AND days_since_last_run BETWEEN 14 AND 60 AND month IN (3, 4)"
      },
      {
        "label": "14/1 or shorter, where the returner lift sat",
        "rule": "race_type = 'Flat' AND is_handicap = 1 AND surface = 'AW' AND last_surface = 'Turf' AND days_since_last_run BETWEEN 14 AND 60 AND sp_decimal <= 15"
      },
      {
        "label": "GB only, since Ireland ran below GB throughout this thread",
        "rule": "race_type = 'Flat' AND is_handicap = 1 AND surface = 'AW' AND last_surface = 'Turf' AND days_since_last_run BETWEEN 14 AND 60 AND region = 'GB'"
      }
    ],
    "control": {
      "description": "Same-race rivals: the other runners in these AW handicaps should score about 0.95-1.00, as usual for the field. The comparison that matters for the question is long-gap turf-to-AW switchers (150+ days) in the same calendar. If the layoff drives the lift, those should score clearly above the short-gap switchers (the earlier March-April AW returner rule gave 1.195 on 480). If the switch alone drives it, the two groups should score about the same.",
      "rule": "race_type = 'Flat' AND is_handicap = 1 AND surface = 'AW' AND last_surface = 'Turf' AND days_since_last_run >= 150"
    },
    "pass_criteria": {
      "min_bets": 1500,
      "min_ae": 1.05,
      "max_p": 0.05
    },
    "forward_bar": {
      "min_bets": 300,
      "min_ae": 1,
      "max_p": 0.2
    },
    "confounders": [
      "Short-gap switchers are often late-season (autumn/winter) horses dropping onto softer AW races, so the calendar differs from the spring returner group. The month refinement addresses this.",
      "Trainer intent differs: a mid-campaign switch is often a placement for an easier mark or ground, while a switch after a break reflects chosen readiness",
      "Field size, class and overround differ between AW and turf handicaps",
      "Ireland's AW programme is small and ran below GB in this thread, so pooled results are GB-dominated",
      "Price-band composition: lifts in this family sat in the short bands while 16/1+ dragged, so a pooled A/E can hide offsetting bands"
    ],
    "mechanism": "Rival explanation for the AW spring returner lift. If the market just discounts turf form when a horse runs on the AW, a turf-to-AW switcher after a normal 14-60 day gap should be underpriced as well, which would make the layoff irrelevant. If the lift comes from trainers choosing when a horse is ready after a long break, short-gap switchers have no such signal and should be fairly priced.",
    "market_miss": "Under the rival, punters distrust turf form on the AW whatever the gap, so switchers drift. Under the main idea, a routine surface switch is common, well understood and correctly priced. Only the readiness signal after a long absence is missed.",
    "prediction": {
      "direction": "negative",
      "size": "Fairly priced to slightly over-bet: A/E 0.93 to 1.03 on 1500+ bets",
      "falsified_if": "A/E of 1.08 or higher on 1500+ bets. That would make the surface switch, not the layoff, the driver of the AW returner lift."
    },
    "notes": "This is a rival-explanation test that predicts a null. The main hypothesis is supported if the broad rule comes in near 1.0 while the long-gap control holds well above it. The sketch rule was kept as written: every column is pre-off and none reads the result. min_bets was raised to 1500 to match the predicted sample, so that a null is informative. The surface values 'AW' and 'Turf' follow the encoding used in earlier rules in this family. I assumed last_surface uses the same values and did not check it."
  }

Timeline

1. Job 1736 finished

   16 Oct, 14:05 · engine:compute-1

2. H-e2d31d3d Discovery → KILLED fails: A/E 1.07, needs 0.95 or lower; p 0.9191 (lower tail), needs 0.05; periods [1.15, 1.02], each needs 0.99 or lower; A/E 1.07 against same-race control 0.97, needs 0.03 below

   16 Oct, 14:05 · engine:evaluator

3. discovery result for H-e2d31d3d (broad rule): fail

   16 Oct, 14:05 · engine:evaluator

4. H-e2d31d3d Designed → RESEARCHdiscovery measuring on the discovery zone

   16 Oct, 14:05 · engine:orchestrator

5. compute-1 took job 1736 (evaluate, attempt 1)

   16 Oct, 14:05 · compute-1

6. Job 1734 finished

   16 Oct, 14:04 · engine:ai-1

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

   16 Oct, 14:04 · engine:scheduler

8. H-e2d31d3d Proposed → RESEARCHdesigned pre-spec written and hashed

   16 Oct, 14:04 · agent:designer@v1

9. ai-1 took job 1734 (design, attempt 1)

   16 Oct, 14:04 · ai-1

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

    16 Oct, 14:03 · engine:scheduler

11. New idea H-e2d31d3d: Rival: turf-to-AW switchers after a short gap carry the same lift, so the layoff is irrelevant (rival-explanation)

    16 Oct, 14:03 · 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.