Rival: lightly raced Flat staying handicappers beat their prices whether or not they have hurdled

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-cf4b198a (Sharpen: proven hurdlers lightly raced on the Flat in Irish summer sta) passed discovery as a niche: A/E 1.26 on 50 bets (p 0.341), same-race control 0.94. O-d3f516ff
2. AskedIs the lift in lightly raced Flat staying handicappers about hurdle form, or about any horse with few Flat runs whose mark is still settling? answered
3. MechanismHorses with few Flat runs have provisional handicap marks and often improve sharply with experience and a step up in trip, whether or not they have run over hurdles.
4. Why the market misses itHandicappers can set the early marks too high or too low, and the form book shows little to go on, so lightly raced types could be mispriced for reasons that have nothing to do with hurdling.
5. Predictedpositive, A/E 1.05 to 1.12 on 300+ bets
6. Falsified ifA/E at or below 1.00 on 250+ bets. That would mean the lift needs hurdle form, which supports the information-gap mechanism.
7. Rule sketchregion = 'IRE' AND race_type = 'Flat' AND is_handicap = 1 AND month BETWEEN 6 AND 9 AND distance_yards >= 2640 AND flat_runs <= 6 AND hurdle_runs = 0

Mechanism and rationale

Horses with few Flat runs have provisional handicap marks and often improve sharply with experience and a step up in trip, whether or not they have run over hurdles.

The rule

region = 'IRE' AND race_type = 'Flat' AND is_handicap = 1 AND month BETWEEN 6 AND 9 AND distance_yards >= 2640 AND flat_runs <= 6 AND hurdle_runs = 0

Reads: region, race_type, is_handicap, month, distance_yards, flat_runs, hurdle_runs

Pre-registration

Spec hash (sha256 of the canonical pre-spec)

c707e69deb35b35f7f95d6dc8f95791ceaeaeb9a3f63f2475ce889e26aa1c6d3

Frozen

not frozen

The definition can still change.

Confounders

• age: horses that have never hurdled skew young. Three-year-olds improve naturally and are often already handled as improvers, so check refinement age >= 4.
• trainer: Flat yards rather than dual-purpose yards. Yard-level skill or how the market prices those yards could drive the result instead of the provisional mark.
• field_size: these races vary widely in field size, and long-price variance is greater in big fields
• price profile: the sharpen's lift came at 14/1 or shorter. A lift here carried by long-priced winners would be a different effect, so check the price-capped refinement.
• history truncation: flat_runs and hurdle_runs count only from mid-2020 (hurdle_runs from Aug 2020). Early in the window some exposed horses will look lightly raced and some hurdlers will show hurdle_runs = 0, which dilutes the selection.
• overlap with the festival system: some selections run at Galway, Listowel and the other festival courses, where the family saw course-specific lifts

Control

{
  "description": "The standard control is the other runners in the same races. If lightly raced non-hurdlers really carry an edge, their race-mates should come in near 1.0 or lower. Earlier controls in this family ran 0.94 to 1.10. A same-race control near 1.1 alongside a selection near 1.05 would mean no edge relative to the race. A second control isolates the 'few Flat runs' part of the claim: exposed Flat-only horses (10+ Flat runs, never hurdled) in the same kind of races. They share the Flat-yard and no-hurdling profile but have settled marks. The rival mechanism expects them at about 1.0 or below, clearly under the selection. If they match the selection, any lift is a Flat-yard or non-hurdler effect, not the provisional-mark effect.",
  "rule": "region = 'IRE' AND race_type = 'Flat' AND is_handicap = 1 AND month BETWEEN 6 AND 9 AND distance_yards >= 2640 AND flat_runs >= 10 AND hurdle_runs = 0"
}

Discovery result

fail broad rule · 17 Oct, 03:29

Bets

215

19 winners

A/E

1.05

95%: 0.63 to 1.64

SP return

-38.3%

level stakes

Betfair SP return

-13.9%

215 bets with a BSP

Strike rate 8.8%. p = 0.45. 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

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 · 1m 11s · 17 Oct, 03:22

  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 flat_runs <= 6 AND hurdle_runs = 0",
    "refinements": [
      {
        "label": "Price-capped at 14/1 or shorter (the sharpen's winners all came at 14/1 or shorter)",
        "rule": "region = 'IRE' AND race_type = 'Flat' AND is_handicap = 1 AND month BETWEEN 6 AND 9 AND distance_yards >= 2640 AND flat_runs <= 6 AND hurdle_runs = 0 AND sp_decimal <= 15"
      },
      {
        "label": "Age 4+ (removes three-year-olds' natural maturing, the main age confounder, and matches the sharpen's older profile)",
        "rule": "region = 'IRE' AND race_type = 'Flat' AND is_handicap = 1 AND month BETWEEN 6 AND 9 AND distance_yards >= 2640 AND flat_runs <= 6 AND hurdle_runs = 0 AND age >= 4"
      },
      {
        "label": "Very lightly raced, 3 or fewer Flat runs (dose check: the provisional-mark story predicts a stronger lift on the fewest runs)",
        "rule": "region = 'IRE' AND race_type = 'Flat' AND is_handicap = 1 AND month BETWEEN 6 AND 9 AND distance_yards >= 2640 AND flat_runs <= 3 AND hurdle_runs = 0"
      }
    ],
    "control": {
      "description": "The standard control is the other runners in the same races. If lightly raced non-hurdlers really carry an edge, their race-mates should come in near 1.0 or lower. Earlier controls in this family ran 0.94 to 1.10. A same-race control near 1.1 alongside a selection near 1.05 would mean no edge relative to the race. A second control isolates the 'few Flat runs' part of the claim: exposed Flat-only horses (10+ Flat runs, never hurdled) in the same kind of races. They share the Flat-yard and no-hurdling profile but have settled marks. The rival mechanism expects them at about 1.0 or below, clearly under the selection. If they match the selection, any lift is a Flat-yard or non-hurdler effect, not the provisional-mark effect.",
      "rule": "region = 'IRE' AND race_type = 'Flat' AND is_handicap = 1 AND month BETWEEN 6 AND 9 AND distance_yards >= 2640 AND flat_runs >= 10 AND hurdle_runs = 0"
    },
    "pass_criteria": {
      "min_bets": 250,
      "min_ae": 1.05,
      "max_p": 0.05
    },
    "forward_bar": {
      "min_bets": 150,
      "min_ae": 1,
      "max_p": 0.2
    },
    "confounders": [
      "age: horses that have never hurdled skew young. Three-year-olds improve naturally and are often already handled as improvers, so check refinement age >= 4.",
      "trainer: Flat yards rather than dual-purpose yards. Yard-level skill or how the market prices those yards could drive the result instead of the provisional mark.",
      "field_size: these races vary widely in field size, and long-price variance is greater in big fields",
      "price profile: the sharpen's lift came at 14/1 or shorter. A lift here carried by long-priced winners would be a different effect, so check the price-capped refinement.",
      "history truncation: flat_runs and hurdle_runs count only from mid-2020 (hurdle_runs from Aug 2020). Early in the window some exposed horses will look lightly raced and some hurdlers will show hurdle_runs = 0, which dilutes the selection.",
      "overlap with the festival system: some selections run at Galway, Listowel and the other festival courses, where the family saw course-specific lifts"
    ],
    "mechanism": "Horses with few Flat runs carry provisional handicap marks set on little evidence. In Irish summer staying handicaps they often improve sharply with experience and the step up to 1m4f+, whether or not they have run over hurdles. If that is what drives the sharpen (hurdle winner, six or fewer Flat runs: A/E 1.26 on 50), then lightly raced runners that have never hurdled should also beat their prices. If they do not, the sharpen's lift needs proven hurdle form, which is the information-gap explanation.",
    "market_miss": "Early marks from the handicapper can be badly off, and with few runs the form book shows little about stamina or how much the horse will improve. Punters and bookmakers lean on visible form and so underrate unexposed stayers on a first or second try at the trip. This would be a general pricing error about lightly raced horses, nothing to do with jumping form.",
    "prediction": {
      "direction": "positive",
      "size": "A/E 1.05 to 1.12 on 300+ bets, above both the same-race control and the exposed Flat-only control",
      "falsified_if": "A/E at or below 1.00 on 250+ bets, or the selection comes in at or below its same-race control or the exposed Flat-only control. That would mean the sharpen's lift needs hurdle form, which supports the information-gap mechanism over the provisional-mark rival."
    },
    "notes": "I checked the columns with query_catalog: month, flat_runs, hurdle_runs, age and sp_decimal are all pre_off and cover IRE from 2020-07/08. I kept the draft rule unchanged. It contains the sharpen's flat_runs <= 6 condition and swaps hurdle_wins >= 1 for hurdle_runs = 0, so the two groups do not overlap and the comparison is clean. I set min_bets to 250 to match the hypothesis's own falsification threshold. The sp_decimal <= 15 cap (14/1) follows the sharpen's short-price winner profile."
  }

Timeline

1. Job 2664 finished

   17 Oct, 03:29 · engine:compute-1

2. H-6234a360 Discovery → KILLED fails: 215 bets, needs 250; A/E 1.048, needs 1.05; p 0.4491, needs 0.05 or lower; periods [1.05], each needs 1.0 (not a niche pass: A/E 1.048 below 1.05)

   17 Oct, 03:29 · engine:evaluator

3. discovery result for H-6234a360 (broad rule): fail

   17 Oct, 03:29 · engine:evaluator

4. H-6234a360 Designed → RESEARCHdiscovery measuring on the discovery zone

   17 Oct, 03:29 · engine:orchestrator

5. compute-1 took job 2664 (evaluate, attempt 1)

   17 Oct, 03:29 · compute-1

6. Job 2661 finished

   17 Oct, 03:23 · engine:ai-1

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

   17 Oct, 03:23 · engine:scheduler

8. H-6234a360 Proposed → RESEARCHdesigned pre-spec written and hashed

   17 Oct, 03:23 · agent:designer@v1

9. ai-1 took job 2661 (design, attempt 1)

   17 Oct, 03:22 · ai-1

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

    17 Oct, 03:22 · engine:scheduler

11. New idea H-6234a360: Rival: lightly raced Flat staying handicappers beat their prices whether or not they have hurdled (rival-explanation)

    17 Oct, 03:22 · 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.