Rival direction: dual-purpose runners away from the festival courses are over-bet in Irish summer staying handicaps

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-4e463106 (Test-mechanism: straight from hurdles into an Irish summer staying Fla) was predicted positive but came out the other way: A/E 0.86 on 130 bets (p 0.437 for the opposite direction). O-470480d3
   • H-7f585029 (Rival: the festival lift is a large-field effect, not a course effect) was predicted positive but came out the other way: A/E 0.79 on 107 bets (p 0.396 for the opposite direction). O-c18c7bad
2. AskedAway from Galway, Listowel, Killarney and Ballinrobe, are dual-purpose runners (3+ hurdle runs) in Irish summer staying Flat handicaps over-bet compared with their race-mates? testing
3. MechanismGood hurdle form is in plain view, and the big jumps yards are well known for running dual-purpose horses. Punters and the yards' own followers back that form on the Flat, but hurdle ability does not fully carry over to Flat pace and speed at ordinary summer meetings.
4. Why the market misses itThe market over-reads a trait everyone can see (jumps form, a big yard) rather than missing one. A string of 0.79 to 0.86 cuts against race-mates near 1.1 fits money chasing a visible label.
5. Predictednegative, A/E 0.80 to 0.92 on 200+ bets, below a same-race control near 1.0 or higher
6. Falsified ifA/E at or above 0.98 on 200+ bets, or no gap below the same-race control

Mechanism and rationale

Good hurdle form is in plain view, and the big jumps yards are well known for running dual-purpose horses. Punters and the yards' own followers back that form on the Flat, but hurdle ability does not fully carry over to Flat pace and speed at ordinary summer meetings.

The rule

region = 'IRE' AND race_type = 'Flat' AND is_handicap = 1 AND month BETWEEN 6 AND 9 AND distance_yards >= 2640 AND prior_hurdle_runs >= 3 AND course NOT IN ('GALWAY', 'LISTOWEL', 'KILLARNEY', 'BALLINROBE')

Reads: region, race_type, is_handicap, month, distance_yards, prior_hurdle_runs, course

Pre-registration

Spec hash (sha256 of the canonical pre-spec)

740625fc773bdb6931ef193a9a9b1465bbbd3bd135d79ea49119044449e9f628

Frozen

not frozen

The definition can still change.

Confounders

• Long-price share differs by course: if the qualifiers are mostly big prices at ordinary tracks, a low A/E can be the usual longshot drag, not mispricing of the label. Check A/E by price band against the control.
• Yard mix: a few big jumps yards supply most dual-purpose runners, so a yard's own strike rate or habit of having runners 'out for a run' could cause the result. Check that no single trainer accounts for most of the shortfall.
• Column definitions: prior_hurdle_runs (record before today) and hurdle_runs may count differently. This rule uses prior_hurdle_runs, matching the festival system and its rivals.
• Field size: dual-purpose runners cluster in big staying fields. Refinement 3 removes 14+ runner fields.
• Overlap with the already-tested big-field complement (A/E 0.79 on 107): part of this sample was already seen, so the broad rule is not fully independent of the observation that motivated it.
• Season-of-year fitness: horses kept for winter jumping may be short of a run in summer, which is a fitness effect rather than over-reading of hurdle form.

Discovery result

fail broad rule · 17 Oct, 03:42

Bets

256

19 winners

A/E

0.98

95%: 0.59 to 1.53

SP return

-11.9%

level stakes

Betfair SP return

+35.2%

256 bets with a BSP

Strike rate 7.4%. p = 0.56. Same-race control: the other runners in these races had A/E 1.01, so the selection did not 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 · 17 Oct, 03:36

  Output (schema-checked)

  {
    "rule": "region = 'IRE' AND race_type = 'Flat' AND is_handicap = 1 AND month BETWEEN 6 AND 9 AND distance_yards >= 2640 AND prior_hurdle_runs >= 3 AND course NOT IN ('GALWAY', 'LISTOWEL', 'KILLARNEY', 'BALLINROBE')",
    "refinements": [
      {
        "label": "10/1 or shorter: where label-chasing money concentrates",
        "rule": "region = 'IRE' AND race_type = 'Flat' AND is_handicap = 1 AND month BETWEEN 6 AND 9 AND distance_yards >= 2640 AND prior_hurdle_runs >= 3 AND course NOT IN ('GALWAY', 'LISTOWEL', 'KILLARNEY', 'BALLINROBE') AND sp_decimal <= 11"
      },
      {
        "label": "Visible hurdle form: at least one hurdle win",
        "rule": "region = 'IRE' AND race_type = 'Flat' AND is_handicap = 1 AND month BETWEEN 6 AND 9 AND distance_yards >= 2640 AND prior_hurdle_runs >= 3 AND course NOT IN ('GALWAY', 'LISTOWEL', 'KILLARNEY', 'BALLINROBE') AND hurdle_wins >= 1"
      },
      {
        "label": "Fields under 14: no overlap with the already-tested big-field cut",
        "rule": "region = 'IRE' AND race_type = 'Flat' AND is_handicap = 1 AND month BETWEEN 6 AND 9 AND distance_yards >= 2640 AND prior_hurdle_runs >= 3 AND course NOT IN ('GALWAY', 'LISTOWEL', 'KILLARNEY', 'BALLINROBE') AND field_size < 14"
      }
    ],
    "control": {
      "description": "The standard control is the other runners in the same races (non-qualifiers in the same Irish summer staying handicaps away from the four festival courses). Earlier cuts put that group near 1.08-1.10, so I expect it at or above 1.0. The effect has to show as a gap below that control, not just a low A/E. Both groups pay the same overround, so a gap cannot come from the race. No separate control rule is needed.",
      "rule": null
    },
    "pass_criteria": {
      "min_bets": 200,
      "min_ae": 1.05,
      "max_p": 0.05
    },
    "forward_bar": {
      "min_bets": 100,
      "min_ae": 1,
      "max_p": 0.2
    },
    "confounders": [
      "Long-price share differs by course: if the qualifiers are mostly big prices at ordinary tracks, a low A/E can be the usual longshot drag, not mispricing of the label. Check A/E by price band against the control.",
      "Yard mix: a few big jumps yards supply most dual-purpose runners, so a yard's own strike rate or habit of having runners 'out for a run' could cause the result. Check that no single trainer accounts for most of the shortfall.",
      "Column definitions: prior_hurdle_runs (record before today) and hurdle_runs may count differently. This rule uses prior_hurdle_runs, matching the festival system and its rivals.",
      "Field size: dual-purpose runners cluster in big staying fields. Refinement 3 removes 14+ runner fields.",
      "Overlap with the already-tested big-field complement (A/E 0.79 on 107): part of this sample was already seen, so the broad rule is not fully independent of the observation that motivated it.",
      "Season-of-year fitness: horses kept for winter jumping may be short of a run in summer, which is a fitness effect rather than over-reading of hurdle form."
    ],
    "mechanism": "Hurdle form and big jumps yards are in plain view. Punters and the yards' followers back that form when these horses run in Flat staying handicaps. Hurdle ability carries over only partly to the pace and finishing speed needed on the Flat at ordinary summer meetings away from the festivals. So horses with 3+ prior hurdle runs are shorter than their Flat chances justify.",
    "market_miss": "This is over-reading of a visible trait, not missing a hidden one. Jumps form and a known dual-purpose yard draw money regardless of how the form translates to the Flat. Earlier cuts outside the festival courses ran 0.79 to 0.86 while race-mates were near 1.1, which fits money chasing a label. Because the label is public, the market should in principle correct for it. The test is whether the gap persists against the same-race control.",
    "prediction": {
      "direction": "negative",
      "size": "A/E 0.80 to 0.92 on 200+ bets, at least 0.10 below the same-race control (which should be near 1.0 or higher)",
      "falsified_if": "A/E at or above 0.98 on 200+ bets, or the gap below the same-race control is under 0.05, or the whole shortfall comes from one price band or one trainer."
    },
    "notes": "Negative-direction hypothesis: pass_criteria are the lab floors applied in the over-bet direction (the vault judges the rule as a lay/avoid, i.e. A/E at or below 1/1.05 ≈ 0.95 with p ≤ 0.05). min_bets is raised to 200 to match the prediction. All columns checked in the catalogue as pre_off: month, prior_hurdle_runs (from 2020-08), hurdle_wins (from 2020-10), sp_decimal, field_size. The rule is kept exactly as sketched so it is the true complement of the festival-course system. Overlap warning: the big-field rival (0.79 on 107) is a subset of this rule, so discovery is partly in-sample. The vault/forward result carries the real weight."
  }

Timeline

1. Job 2718 finished

   17 Oct, 03:42 · engine:compute-1

2. H-030e5400 Discovery → KILLED fails: 256 bets, needs 300; A/E 0.981, needs 0.95 or lower; p 0.5276 (lower tail), needs 0.05; periods [0.98], each needs 0.99 or lower; A/E 0.981 against same-race control 1.011, needs 0.03 below

   17 Oct, 03:42 · engine:evaluator

3. discovery result for H-030e5400 (broad rule): fail

   17 Oct, 03:42 · engine:evaluator

4. H-030e5400 Designed → RESEARCHdiscovery measuring on the discovery zone

   17 Oct, 03:42 · engine:orchestrator

5. compute-1 took job 2718 (evaluate, attempt 1)

   17 Oct, 03:42 · compute-1

6. Job 2716 finished

   17 Oct, 03:36 · engine:ai-1

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

   17 Oct, 03:36 · engine:scheduler

8. H-030e5400 Proposed → RESEARCHdesigned pre-spec written and hashed

   17 Oct, 03:36 · agent:designer@v1

9. ai-1 took job 2716 (design, attempt 1)

   17 Oct, 03:36 · ai-1

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

    17 Oct, 03:36 · engine:scheduler

11. New idea H-030e5400: Rival direction: dual-purpose runners away from the festival courses are over-bet in Irish summer staying handicaps (rival-explanation)

    17 Oct, 03:36 · 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.