Rival: stayers dropped to a sprint pay at non-stiff tracks too

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-9be58fc5 (Rival: dropping in trip after a win over further pays at any track, st) pointed the predicted way but failed the bar: A/E 1.05 on 4671 bets (p 0.141), control 0.99. O-23bbb3b8
   • H-4aa47f2c (Dose-response: the stiff-track edge is weaker at seven furlongs to a m) was predicted positive but came out the other way: A/E 0.9 on 109 bets (p 0.441 for the opposite direction). O-4ab91a2c
2. AskedIn Flat sprints (distance_yards <= 1340), does a winner over at least 330 yards further also beat its price at non-stiff tracks, or is the 1.51 special to stiff finishes? answered
3. MechanismA horse that won over a furlong and a half further may be under-rated in sprints anywhere. Its stamina shows late in a strongly run sprint, whatever the track shape.
4. Why the market misses itPunters mark down horses dropping sharply in trip as 'too short'. Its recent runs over further show slow early sectionals, which look bad but do not predict how it finishes in a sprint.
5. Predictedpositive, A/E 1.10 to 1.30 on 400+ bets at non-stiff tracks
6. Falsified ifA/E at or below 1.02, which would support the stiff-finish mechanism behind the live system
7. Rule sketch

Mechanism and rationale

A horse that won over a furlong and a half further may be under-rated in sprints anywhere. Its stamina shows late in a strongly run sprint, whatever the track shape.

The rule

race_type = 'Flat' AND distance_yards <= 1340 AND track_stiff_finish = 0 AND last_win_distance_yards >= distance_yards + 330

Reads: race_type, distance_yards, track_stiff_finish, last_win_distance_yards

Pre-registration

Spec hash (sha256 of the canonical pre-spec)

4808ac29ded6c801b2686019530a6173adf843ba0e2bdf056630c996458a359d

Frozen

not frozen

The definition can still change.

Confounders

• All-weather sprints dominate non-stiff tracks in winter; the stiff-track system is mostly turf, so compare like with like (turf refinement)
• Price band: lift carried by a few 16/1+ winners; judge on A/E at 14/1 or shorter
• Field size: small fields are rarely strongly run, which the mechanism needs
• Period: the non-stiff trip-drop lift fell entirely after 2026-04-27 (1.14 vs 0.98 before); check the period split and whether it is a Flat-turf season effect
• last_win_distance_yards vs last_win_yards: both columns exist; this rule uses the system's column so it is the exact non-stiff mirror of the live system
• Last win may be old or on another code (e.g. a hurdle win); the trip drop may then be a class or code change rather than a stamina signal

Control

{
  "description": "Same-race other runners should show A/E near 1.00 (about 0.97-1.00); a lift for the selections is only meaningful against that. The decisive comparison is the live stiff-track system (same rule with track_stiff_finish = 1): if this non-stiff rule matches it (within ~0.03 A/E), the hill is not needed and the system's edge belongs to the trip drop; if this rule is at or below 1.02 while the stiff version holds, the stiff-finish mechanism stands. Also check non-stayers in the same non-stiff sprints dropping by less than 330 yards to rule out a generic trip-drop effect.",
  "rule": "race_type = 'Flat' AND distance_yards <= 1340 AND track_stiff_finish = 1 AND last_win_distance_yards >= distance_yards + 330"
}

Discovery result

fail broad rule · 16 Oct, 12:49

Bets

164

15 winners

A/E

1.10

95%: 0.62 to 1.81

SP return

-27.7%

level stakes

Betfair SP return

+1.7%

156 bets with a BSP

Strike rate 9.1%. p = 0.39. Same-race control: the other runners in these races had A/E 0.99, 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 · 37s · 16 Oct, 12:47

  Output (schema-checked)

  {
    "rule": "race_type = 'Flat' AND distance_yards <= 1340 AND track_stiff_finish = 0 AND last_win_distance_yards >= distance_yards + 330",
    "refinements": [
      {
        "label": "14/1 or shorter",
        "rule": "race_type = 'Flat' AND distance_yards <= 1340 AND track_stiff_finish = 0 AND last_win_distance_yards >= distance_yards + 330 AND sp_decimal <= 15.0"
      },
      {
        "label": "turf only (like-for-like with the stiff-track system)",
        "rule": "race_type = 'Flat' AND distance_yards <= 1340 AND track_stiff_finish = 0 AND last_win_distance_yards >= distance_yards + 330 AND surface = 'turf'"
      },
      {
        "label": "strongly run sprints (bigger fields)",
        "rule": "race_type = 'Flat' AND distance_yards <= 1340 AND track_stiff_finish = 0 AND last_win_distance_yards >= distance_yards + 330 AND field_size >= 10"
      }
    ],
    "control": {
      "description": "Same-race other runners should show A/E near 1.00 (about 0.97-1.00); a lift for the selections is only meaningful against that. The decisive comparison is the live stiff-track system (same rule with track_stiff_finish = 1): if this non-stiff rule matches it (within ~0.03 A/E), the hill is not needed and the system's edge belongs to the trip drop; if this rule is at or below 1.02 while the stiff version holds, the stiff-finish mechanism stands. Also check non-stayers in the same non-stiff sprints dropping by less than 330 yards to rule out a generic trip-drop effect.",
      "rule": "race_type = 'Flat' AND distance_yards <= 1340 AND track_stiff_finish = 1 AND last_win_distance_yards >= distance_yards + 330"
    },
    "pass_criteria": {
      "min_bets": 300,
      "min_ae": 1.05,
      "max_p": 0.05
    },
    "forward_bar": {
      "min_bets": 100,
      "min_ae": 1,
      "max_p": 0.2
    },
    "confounders": [
      "All-weather sprints dominate non-stiff tracks in winter; the stiff-track system is mostly turf, so compare like with like (turf refinement)",
      "Price band: lift carried by a few 16/1+ winners; judge on A/E at 14/1 or shorter",
      "Field size: small fields are rarely strongly run, which the mechanism needs",
      "Period: the non-stiff trip-drop lift fell entirely after 2026-04-27 (1.14 vs 0.98 before); check the period split and whether it is a Flat-turf season effect",
      "last_win_distance_yards vs last_win_yards: both columns exist; this rule uses the system's column so it is the exact non-stiff mirror of the live system",
      "Last win may be old or on another code (e.g. a hurdle win); the trip drop may then be a class or code change rather than a stamina signal"
    ],
    "mechanism": "A Flat horse that last won over at least 330 yards (a furlong and a half) further than today's sprint trip has proven stamina. In a strongly run sprint the leaders tire late, and a stayer keeps going at the finish, so it can beat sprinters that fade, whatever the track shape. If true, the stiff finish adds little and the edge comes from the trip drop itself.",
    "market_miss": "Punters mark down horses dropping sharply in trip as 'too short' for a sprint. Their recent runs over further show slow early sectionals and off-the-pace positions, which look like a lack of speed but do not predict how they finish in a sprint. On non-stiff tracks punters lean even harder on the 'speed track, no chance' story, so any underpricing should be at least as large there.",
    "prediction": {
      "direction": "positive",
      "size": "A/E 1.10 to 1.30 on 400+ bets at non-stiff tracks, similar to the stiff-track system",
      "falsified_if": "A/E at or below 1.02 on the broad rule (or the lift exists only at 16/1+ and only after 2026-04-27), which would support the stiff-finish mechanism behind the live system"
    },
    "notes": "Rival-explanation test for SYS-stayer-dropped-to-a-stiff-sprint. The rule sketch was kept unchanged: all columns are pre-off and it is the exact mirror of the live system with track_stiff_finish flipped. last_win_distance_yards is available from 2020-09-01. Refinements come from the reasoning and confounders: price cap (edges die at big prices), turf (all-weather confound), and field size as a proxy for a strongly run sprint. min_bets is raised to 300 because the prediction expects 400+ bets."
  }

Timeline

1. Job 1444 finished

   16 Oct, 12:49 · engine:compute-1

2. H-eb945dd6 Discovery → KILLED fails: 164 bets, needs 300; p 0.391, needs 0.05 or lower; periods [0.77, 1.41], each needs 1.0 (not a niche pass: a period points the other way [0.77, 1.41])

   16 Oct, 12:49 · engine:evaluator

3. discovery result for H-eb945dd6 (broad rule): fail

   16 Oct, 12:49 · engine:evaluator

4. H-eb945dd6 Designed → RESEARCHdiscovery measuring on the discovery zone

   16 Oct, 12:49 · engine:orchestrator

5. compute-1 took job 1444 (evaluate, attempt 1)

   16 Oct, 12:49 · compute-1

6. Job 1442 finished

   16 Oct, 12:47 · engine:ai-1

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

   16 Oct, 12:47 · engine:scheduler

8. H-eb945dd6 Proposed → RESEARCHdesigned pre-spec written and hashed

   16 Oct, 12:47 · agent:designer@v1

9. ai-1 took job 1442 (design, attempt 1)

   16 Oct, 12:47 · ai-1

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

    16 Oct, 12:47 · engine:scheduler

11. New idea H-eb945dd6: Rival: stayers dropped to a sprint pay at non-stiff tracks too (rival-explanation)

    16 Oct, 12:47 · 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.