Fancied lightly raced four-year-old returners: the market flags which ones are ready

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-52d3c28e (Irish lightly raced four-year-old returners in spring handicaps are fu) was predicted negative but came out the other way: A/E 1.35 on 135 bets (p 0.212 for the opposite direction). O-6b28e152
   • H-1a3e129f (Lightly raced four-year-old returners in Flat non-handicaps: progress ) passed discovery but was falsified: IRE A/E 0.861; driven by one price band (up to 3/1): the rest score 1.00. O-d5ec2505
2. AskedWithin the H126 base, is the lift confined to returners priced 14/1 or shorter, with 16/1+ returners at or below their prices in both countries and on both surfaces? answered
3. MechanismA lightly raced four-year-old back from a winter break may have strengthened physically. Only some are fully wound up first time out, and the stable's confidence (gallops reports, money) pushes the ready ones to credible prices. The ones that need the run drift to 16/1 or bigger and rarely win.
4. Why the market misses itPunters treat a 150+ day absence and a thin three-year-old record as uncertainty and price for some regression. They do not credit enough the improvement of a fit, maturing horse whose form figures come from when it was weaker. The market gets the direction right (shorter means readier) but not the size.
5. Predictedpositive, A/E 1.15 to 1.30 on 900+ bets, with 16/1+ returners at or below 0.90
6. Falsified if

Mechanism and rationale

A lightly raced four-year-old back from a winter break may have strengthened physically. Only some are fully wound up first time out, and the stable's confidence (gallops reports, money) pushes the ready ones to credible prices. The ones that need the run drift to 16/1 or bigger and rarely win.

The rule

race_type = 'Flat' AND is_handicap = 1 AND month IN (3, 4, 5) AND age = 4 AND prior_runs <= 10 AND days_since_last_run >= 150 AND sp_decimal <= 15

Reads: race_type, is_handicap, month, age, prior_runs, days_since_last_run, sp_decimal

Pre-registration

Spec hash (sha256 of the canonical pre-spec)

bcff0e43f88c0e679007be13563370b4b158b504b58aad338d41d7396786febc

Frozen

16 Oct, 12:51

Nothing in the definition can change after this.

Confounders

• sp_decimal is the price at the off, so the cap mixes the market's readiness signal with general favourite-longshot bias; the price-matched non-returner control rule is there to separate the two
• Price-band slicing after seeing the numbers: the 14/1 cut came from the Irish and all-weather discovery splits, so discovery A/E is likely inflated (earlier vault lifts shrank to about 1.10)
• A few big yards dominating the shorter-priced returners, so trainer concentration could pass for a horse-level effect
• Country mix: Irish returners differ in their price spread, and Ireland has underperformed in side cuts
• Surface mix: all-weather and turf returners differ in their price distributions
• SP-based rule can only be bet at the off/BSP, so the edge may not survive in early prices

Control

{
  "description": "Two comparisons. (1) Inside the H126 base, the complement priced 16/1 or bigger (same rule but sp_decimal > 15) should score A/E 0.90 or lower; if it scores as high as the capped group, the price cap is not where the lift lives. (2) Favourite-longshot bias check: runners at 14/1 or shorter in the same spring Flat handicaps that are NOT returning from a 150+ day break should score about 0.97 to 1.00. If they score close to the capped returners, the 'lift' is just short-price bias and not readiness. The standard same-race other-runners control should come in at or a little below 1.0.",
  "rule": "race_type = 'Flat' AND is_handicap = 1 AND month IN (3, 4, 5) AND sp_decimal <= 15 AND days_since_last_run < 150"
}

Discovery result

pass broad rule (niche) · 16 Oct, 12:49

Bets

337

59 winners

A/E

1.21

95%: 0.92 to 1.56

SP return

+6.0%

level stakes

Betfair SP return

+28.6%

337 bets with a BSP

Strike rate 17.5%. p = 0.083. 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

Each attack is a deliberate attempt to kill the idea. 4 of 5 survived.

• perturbation: Price cap tightened to 11/1 (sp <= 12)pass

  281 bets · A/E 1.21 (0.91 to 1.57) · control 0.95 · 16 Oct, 12:51

  Breakdown

  betsA/E0.71.01.3

  2025-10-01 to 2026-04-271611.30

  after 2026-04-271201.08

• perturbation: Stricter returner definition: 180+ days offpass

  241 bets · A/E 1.19 (0.86 to 1.60) · control 0.96 · 16 Oct, 12:51

Sealed holdout (vault)

Bets

1,567

16 Oct, 12:51

A/E

1.11

p = 0.048

BH threshold then

0.0031

over every vault test so far

Outcome

Failed

Forward record

• forward vs bar {'min_bets': 100, 'min_ae': 1.03, 'max_p': 0.2}info

  0 bets · · 16 Oct, 12:51

  Breakdown

  betsA/E0.71.01.3

  forward0–

Other experiments

• neighbour: <= 10 -> 8 (-20%)pass

  255 bets · A/E 1.27 (0.94 to 1.68) · control 0.94 · 16 Oct, 12:51

  Breakdown

  betsA/E0.71.01.3

  2025-10-01 to 2026-04-271451.41

  after 2026-04-271101.10

• neighbour: <= 10 -> 12 (+20%)pass

  388 bets · A/E 1.18 (0.91 to 1.50) · control 0.96 · 16 Oct, 12:51

  Breakdown

Lineage

Agent runs

• designer v1ok · 35s · 16 Oct, 12:45

  Output (schema-checked)

  {
    "rule": "race_type = 'Flat' AND is_handicap = 1 AND month IN (3, 4, 5) AND age = 4 AND prior_runs <= 10 AND days_since_last_run >= 150 AND sp_decimal <= 15",
    "refinements": [
      {
        "label": "8/1 or shorter: the strongest readiness signal",
        "rule": "race_type = 'Flat' AND is_handicap = 1 AND month IN (3, 4, 5) AND age = 4 AND prior_runs <= 10 AND days_since_last_run >= 150 AND sp_decimal <= 9"
      }
    ],
    "control": {
      "description": "Two comparisons. (1) Inside the H126 base, the complement priced 16/1 or bigger (same rule but sp_decimal > 15) should score A/E 0.90 or lower; if it scores as high as the capped group, the price cap is not where the lift lives. (2) Favourite-longshot bias check: runners at 14/1 or shorter in the same spring Flat handicaps that are NOT returning from a 150+ day break should score about 0.97 to 1.00. If they score close to the capped returners, the 'lift' is just short-price bias and not readiness. The standard same-race other-runners control should come in at or a little below 1.0.",
      "rule": "race_type = 'Flat' AND is_handicap = 1 AND month IN (3, 4, 5) AND sp_decimal <= 15 AND days_since_last_run < 150"
    },
    "pass_criteria": {
      "min_bets": 200,
      "min_ae": 1.08,
      "max_p": 0.05
    },
    "forward_bar": {
      "min_bets": 100,
      "min_ae": 1.03,
      "max_p": 0.2
    },
    "confounders": [
      "sp_decimal is the price at the off, so the cap mixes the market's readiness signal with general favourite-longshot bias; the price-matched non-returner control rule is there to separate the two",
      "Price-band slicing after seeing the numbers: the 14/1 cut came from the Irish and all-weather discovery splits, so discovery A/E is likely inflated (earlier vault lifts shrank to about 1.10)",
      "A few big yards dominating the shorter-priced returners, so trainer concentration could pass for a horse-level effect",
      "Country mix: Irish returners differ in their price spread, and Ireland has underperformed in side cuts",
      "Surface mix: all-weather and turf returners differ in their price distributions",
      "SP-based rule can only be bet at the off/BSP, so the edge may not survive in early prices"
    ],
    "mechanism": "A lightly raced four-year-old (10 or fewer runs) back from 150+ days off in a spring Flat handicap may have strengthened over the winter, and its handicap mark rests on weaker three-year-old form. Only some are fully fit first time out. Stable confidence (gallops reports, stable money) pushes the ready ones to 14/1 or shorter. The ones that need the run drift to 16/1+ and rarely win: in discovery they went 0 from 48 on the all-weather and 1 from 63 in Ireland.",
    "market_miss": "Punters price a long absence and a thin record as uncertainty and expect some regression. They use the price correctly to rank readiness (shorter means fitter) but underrate how much a fit, maturing horse has improved on form figures earned when it was weaker. So the fancied returners are still too long, while the drifters are priced about right or too short.",
    "prediction": {
      "direction": "positive",
      "size": "A/E 1.15 to 1.30 on the capped group (900+ bets across all data), with the 16/1+ complement at or below 0.90",
      "falsified_if": "Any of these falsifies it: A/E below 1.08 on 900+ bets; either country (GB/IRE) or either surface (Turf/AW) below 1.0; the 16/1+ complement scoring as high as the capped group; or price-matched non-returners scoring as high as the capped returners."
    },
    "notes": "The rule sketch is kept unchanged. sp_decimal <= 15 includes 14/1 and excludes 16/1, which matches the finding's split. min_ae is raised to 1.08 to match the falsification threshold. Discovery n for the capped group is likely about 250 to 300, so min_bets is 200. The forward bar is raised to A/E 1.03 because past vault results shrank to about 1.10. The one refinement checks whether the readiness signal is monotone in price; it is looked at only if the broad rule passes."
  }

Timeline

1. Job 1474 finished

   16 Oct, 12:51 · engine:compute-1

2. hypothesis note H-1a6f83c9

   16 Oct, 12:51 · engine:perturb

3. perturbation result for H-1a6f83c9 (neighbour: <= 15 -> 18 (+20%)): pass

   16 Oct, 12:51 · engine:perturb

4. perturbation result for H-1a6f83c9 (neighbour: <= 15 -> 12 (-20%)): pass

   16 Oct, 12:51 · engine:perturb

5. perturbation result for H-1a6f83c9 (neighbour: >= 150 -> 180 (+20%)): pass

   16 Oct, 12:51 · engine:perturb

6. perturbation result for H-1a6f83c9 (neighbour: >= 150 -> 120 (-20%)): pass

   16 Oct, 12:51 · engine:perturb

7. perturbation result for H-1a6f83c9 (neighbour: <= 10 -> 12 (+20%)): pass

   16 Oct, 12:51 · engine:perturb

8. perturbation result for H-1a6f83c9 (neighbour: <= 10 -> 8 (-20%)): pass

   16 Oct, 12:51 · engine:perturb

9. compute-1 took job 1474 (mutate, attempt 1)

   16 Oct, 12:51 · compute-1

10. Queued mutate on the compute lane: parameter stability: do neighbouring thresholds hold?

    16 Oct, 12:51 · engine:scheduler

11. Job 1469 finished

    16 Oct, 12:51 · engine:compute-1

12. forward result for H-1a6f83c9 (forward vs bar {'min_bets': 100, 'min_ae': 1.03, 'max_p': 0.2}): info

    16 Oct, 12:51 · engine:evaluator

13. compute-1 took job 1469 (forward_check, attempt 1)

    16 Oct, 12:51 · compute-1

14. Queued forward_check on the compute lane: daily forward check of a shadow system

    16 Oct, 12:51 · engine:scheduler

15. Job 1466 finished

    16 Oct, 12:51 · engine:compute-1

16. H-1a6f83c9 Vault → SHADOW vault SUPPORTIVE: A/E 1.113 on 1567 bets, p 0.04842 against BH threshold 0.003125 over 32 vault tests

    16 Oct, 12:51 · engine:orchestrator