Young marathon handicap chasers keep the lift outside big-field races

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-b9326edc (Older higher-weighted marathon chasers match the young: weight positio) was predicted positive but the vault years went the other way: A/E 0.9 on 1129 bets. O-7d4a82d7
   • H-8810c1c1 (Rival: short-priced runners of all ages beat their SP in marathon hand) passed discovery as a niche: A/E 1.17 on 157 bets (p 0.205), same-race control 0.72. O-beb20a81
2. AskedDo big-field festival marathon handicaps (Nationals and similar) produce the lift, rather than youth? testing
3. MechanismIf youth is the cause, young horses are underpriced because the market doubts their stamina at extreme trips, and that should hold in ordinary 3m4f+ handicaps with modest fields. If the lift comes from a few big festival races, it should vanish in small fields.
4. Why the market misses itThe form book shows young horses with no proven marathon stamina, and the market prices them against older proven stayers whatever the field size.
5. Predictedpositive, A/E 1.08 to 1.20 on 250+ bets
6. Falsified ifA/E at or below 1.0 on 200+ bets, which would point the lift at big-field races
7. Rule sketchrace_type = 'Chase' AND is_handicap = 1 AND age <= 7 AND distance_yards >= 6160 AND field_size <= 12

Mechanism and rationale

If youth is the cause, young horses are underpriced because the market doubts their stamina at extreme trips, and that should hold in ordinary 3m4f+ handicaps with modest fields. If the lift comes from a few big festival races, it should vanish in small fields.

The rule

race_type = 'Chase' AND is_handicap = 1 AND age <= 7 AND distance_yards >= 6160 AND field_size <= 12

Reads: race_type, is_handicap, age, distance_yards, field_size

Pre-registration

Spec hash (sha256 of the canonical pre-spec)

799a07e51fa50933ea40ffd6afd6aa57098c27af1b25db527601086fcfac4382

Frozen

not frozen

The definition can still change.

Confounders

• race_class: small-field marathon handicaps tend to be lower-grade, so any lift could come from class and not from youth (IRE race_class is only available from 2026-07, so it can't be used as a filter)
• course: a few stamina-test tracks may stage most small-field 3m4f+ chases, so check track_stamina_test and per-course concentration
• book_pct: small fields have a different overround, so an apparent lift could reflect a different market margin
• sp_decimal band: earlier marathon slices drew their lift from a narrow band (4/1 to 7/1 or 16/1+), so check the result is spread across price bands
• Sample thinness: the related young slice at field_size <= 13 had only 63 bets, so discovery here is likely well under 150 and attack survival on thin samples has not predicted vault success
• Trainer concentration: a few yards that place young stayers in marathon chases could drive the result

Control

{
  "description": "Primary control: the other runners (age 8+) in the same small-field 3m4f+ handicap chases. If youth is the cause, they should come in at or slightly below par (about 0.90 to 1.0), so the young runners clearly beat their own race-mates. If race-mates also show a lift, the effect belongs to the race type and not to youth. Secondary control: young runners in big-field (13+) marathon handicap chases, the festival-type races the rival explanation credits. If the lift appears only there and is absent in small fields, the youth mechanism fails.",
  "rule": "race_type = 'Chase' AND is_handicap = 1 AND age <= 7 AND distance_yards >= 6160 AND field_size >= 13"
}

Discovery result

pass broad rule (niche) · 16 Oct, 14:35

Bets

58

12 winners

A/E

1.28

95%: 0.66 to 2.23

SP return

+11.0%

level stakes

Betfair SP return

+22.3%

58 bets with a BSP

Strike rate 20.7%. p = 0.24. Same-race control: the other runners in these races had A/E 0.85, 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. 2 of 5 survived.

• perturbation: Tighter field cut: 10 or fewer runnersinfo

  34 bets · A/E 1.12 (0.45 to 2.31) · control 0.94 · 16 Oct, 14:37

  Breakdown

  betsA/E0.71.01.3

  2025-10-01 to 2026-04-2733–

  after 2026-04-271–

• perturbation: Distance threshold lowered ~10% to 3m2f (5600y)fail

  299 bets · A/E 0.97 (0.73 to 1.26) · control 1.04 · 16 Oct, 14:37

  Breakdown

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.

Other experiments

• neighbour: <= 7 -> 6 (-20%)pass

  13 bets · A/E 1.18 (0.24 to 3.45) · control 0.94 · 16 Oct, 14:35

  Breakdown

  betsA/E0.71.01.3

  2025-10-01 to 2026-04-2713–

  after 2026-04-270–

• neighbour: <= 7 -> 8 (+20%)fail

  103 bets · A/E 0.92 (0.50 to 1.55) · control 1.15 · 16 Oct, 14:35

  Breakdown

  bets

Lineage

Agent runs

• designer v1ok · 32s · 16 Oct, 14:33

  Output (schema-checked)

  {
    "rule": "race_type = 'Chase' AND is_handicap = 1 AND age <= 7 AND distance_yards >= 6160 AND field_size <= 12",
    "refinements": [
      {
        "label": "Small-field young marathon chasers at 16/1 or shorter (edges die at big prices; the lab's prior marathon slices leaned on longshots)",
        "rule": "race_type = 'Chase' AND is_handicap = 1 AND age <= 7 AND distance_yards >= 6160 AND field_size <= 12 AND sp_decimal < 17"
      },
      {
        "label": "Young marathon chasers in genuinely small fields (10 or fewer), furthest from festival-sized races",
        "rule": "race_type = 'Chase' AND is_handicap = 1 AND age <= 7 AND distance_yards >= 6160 AND field_size <= 10"
      }
    ],
    "control": {
      "description": "Primary control: the other runners (age 8+) in the same small-field 3m4f+ handicap chases. If youth is the cause, they should come in at or slightly below par (about 0.90 to 1.0), so the young runners clearly beat their own race-mates. If race-mates also show a lift, the effect belongs to the race type and not to youth. Secondary control: young runners in big-field (13+) marathon handicap chases, the festival-type races the rival explanation credits. If the lift appears only there and is absent in small fields, the youth mechanism fails.",
      "rule": "race_type = 'Chase' AND is_handicap = 1 AND age <= 7 AND distance_yards >= 6160 AND field_size >= 13"
    },
    "pass_criteria": {
      "min_bets": 150,
      "min_ae": 1.05,
      "max_p": 0.05
    },
    "forward_bar": {
      "min_bets": 100,
      "min_ae": 1,
      "max_p": 0.2
    },
    "confounders": [
      "race_class: small-field marathon handicaps tend to be lower-grade, so any lift could come from class and not from youth (IRE race_class is only available from 2026-07, so it can't be used as a filter)",
      "course: a few stamina-test tracks may stage most small-field 3m4f+ chases, so check track_stamina_test and per-course concentration",
      "book_pct: small fields have a different overround, so an apparent lift could reflect a different market margin",
      "sp_decimal band: earlier marathon slices drew their lift from a narrow band (4/1 to 7/1 or 16/1+), so check the result is spread across price bands",
      "Sample thinness: the related young slice at field_size <= 13 had only 63 bets, so discovery here is likely well under 150 and attack survival on thin samples has not predicted vault success",
      "Trainer concentration: a few yards that place young stayers in marathon chases could drive the result"
    ],
    "mechanism": "If youth is the cause, young horses (7 or under) are underpriced at 3m4f+ because the market doubts the unproven stamina of horses with no marathon form. That doubt doesn't depend on field size, so the lift should survive in ordinary small-field (12 or fewer runners) marathon handicap chases. If the lift instead comes from a few big, high-overround festival marathons, it should vanish here.",
    "market_miss": "The form book shows young horses with no proven marathon stamina. The market prices them against older proven stayers and discounts the stamina young chasers usually still have in hand, in any field size. In small, low-profile races there is less public attention and analysis to correct that discount.",
    "prediction": {
      "direction": "positive",
      "size": "A/E 1.08 to 1.20. Discovery sample is likely thin (roughly 60 to 120 bets, given 63 at field_size <= 13), so the full size is judged in the vault and forward record",
      "falsified_if": "A/E at or below 1.0 on 200+ bets (pooled discovery plus vault), especially if young runners in 13+ fields keep a lift above 1.10. That would point the effect at big-field races and not at youth."
    },
    "notes": "The draft rule is sound: every column is pre-off and available from 2020-07, with no post-race reads. I kept field_size <= 12 as sketched rather than tuning it to the earlier 13 cut. Expect discovery to fall short of 150 bets. Under lab policy, a thin result in the right direction goes on to falsification and forward tracking rather than being killed. The secondary control (young runners in 13+ fields) is what actually answers question Q-56f7bf65."
  }

Timeline

1. Job 1811 finished

   16 Oct, 14:37 · engine:compute-1

2. H-d2faaee2 Falsification → KILLED falsified: driven by one price band (up to 7/1): the rest score 0.93; fewer than two periods with enough bets to compare (thin: the forward record will judge); IRE: too few bets to judge

   16 Oct, 14:37 · engine:falsification

3. falsification result for H-d2faaee2 (partition: Restrict to 16/1 or shorter): pass

   16 Oct, 14:37 · engine:evaluator

4. compute-1 took job 1811 (attack, attempt 1)

   16 Oct, 14:37 · compute-1

5. Job 1810 finished

   16 Oct, 14:37 · engine:compute-1

6. falsification result for H-d2faaee2 (control: Older runners in the same 4/1-7/1 small-field marathon band): pass

   16 Oct, 14:37 · engine:evaluator

7. compute-1 took job 1810 (attack, attempt 1)

   16 Oct, 14:37 · compute-1

8. Job 1809 finished

   16 Oct, 14:37 · engine:compute-1

9. falsification result for H-d2faaee2 (price_band: Remove the 4/1-7/1 band that carries the discovery lift): info

   16 Oct, 14:37 · engine:evaluator

10. compute-1 took job 1809 (attack, attempt 1)

    16 Oct, 14:37 · compute-1

11. Job 1808 finished

    16 Oct, 14:37 · engine:compute-1

12. falsification result for H-d2faaee2 (perturbation: Distance threshold lowered ~10% to 3m2f (5600y)): fail

    16 Oct, 14:37 · engine:evaluator

13. compute-1 took job 1808 (attack, attempt 1)

    16 Oct, 14:37 · compute-1

14. Job 1807 finished

    16 Oct, 14:37 · engine:compute-1

15. falsification result for H-d2faaee2 (perturbation: Tighter field cut: 10 or fewer runners): info

    16 Oct, 14:37 · engine:evaluator

16. compute-1 took job 1807 (attack, attempt 1)

    16 Oct, 14:37 · compute-1

17. Job 1802 finished

    16 Oct, 14:37 · engine:ai-1

18. Queued attack on the compute lane: attack: Restrict to 16/1 or shorter

    16 Oct, 14:37 · engine:scheduler

19. Queued attack on the compute lane: attack: Older runners in the same 4/1-7/1 small-field marathon band

    16 Oct, 14:37 · engine:scheduler