Rival: the stiff-sprint system's lift is a 16/1+ longshot effect

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-eb945dd6 (Rival: stayers dropped to a sprint pay at non-stiff tracks too) pointed the predicted way but failed the bar: A/E 1.1 on 164 bets (p 0.391), control 0.99. O-11d05c5a
   • H-cde96a9f (Rival: the trip-drop lift is a Flat turf summer effect) passed discovery: A/E 1.13 on 2045 bets (p 0.038), same-race control 0.97. O-b7f15d9f
2. AskedIs the A/E 1.08 (and the sprint 1.51) carried by runners at 16/1 or longer, given BSP ROI +19% against SP ROI -14.5%? answered
3. MechanismIf the hill really turns a sprint into a stamina test, the edge should hold among contenders the market rates, not only among outsiders. If the lift is only the longshot drift seen in both rivals, it will vanish once the price is capped.
4. Why the market misses itPunters see a horse whose last win came over further and mark it down as too slow for a sprint. On a stiff finish that reading is wrong for horses with a real chance, so they drift beyond their true odds.
5. Predictedpositive, A/E 1.10 to 1.30 on 150+ bets
6. Falsified ifA/E at or below 1.00 at 14/1 or shorter, which would mean the live system's lift is carried by longshots
7. Rule sketch

Mechanism and rationale

If the hill really turns a sprint into a stamina test, the edge should hold among contenders the market rates, not only among outsiders. If the lift is only the longshot drift seen in both rivals, it will vanish once the price is capped.

The rule

race_type = 'Flat' AND distance_yards <= 1340 AND track_stiff_finish = 1 AND last_win_distance_yards >= distance_yards + 330 AND sp_decimal <= 15

Reads: race_type, distance_yards, track_stiff_finish, last_win_distance_yards, sp_decimal

Pre-registration

Spec hash (sha256 of the canonical pre-spec)

77d03278b39755d80b9fe45b77e102b027e330e08d72b24360a542bf3975180c

Frozen

16 Oct, 12:58

Nothing in the definition can change after this.

Confounders

• Small sample after the price cap: the uncapped system is already a niche, so the 14/1-or-shorter subset may fall short of 150 bets in discovery
• Course concentration: a few stiff sprint courses (e.g. Beverley, Carlisle, Galway) may supply most bets, so a course or draw quirk could pass for a stamina effect. Check the per-course split
• Season concentration: earlier trip-drop rivals lifted only after April (1.41 against 0.77). Check the split before and after April
• General trip-drop drift: trip-droppers at non-stiff tracks scored about 1.05 with no edge over control, so the capped rule must clearly beat that baseline as well as 1.00
• Column choice: the live system uses last_win_distance_yards while earlier rules used last_win_yards. Whether the two agree is unresolved, so results may not transfer to rules written on the other column
• Handicap and class mix: stayers dropping to sprints may be mostly low-class handicappers whose marks, not stamina, drive the result

Control

{
  "description": "Primary control: the other runners in the same stiff-track Flat sprints, expected near A/E 1.00 (about 0.97 to 1.02), so any lift in the rule is not a property of these races. Secondary comparison group: the same live-system selections priced above 14/1, which carry the longshot drift seen in both trip-drop rivals (1.31 and 1.41 at 16/1+). If the hypothesis is right, the capped group beats its same-race control by a margin similar to or larger than the 16/1+ group's. If it is wrong, the capped group sits at or below 1.00 and the whole system lift sits in the control rule.",
  "rule": "race_type = 'Flat' AND distance_yards <= 1340 AND track_stiff_finish = 1 AND last_win_distance_yards >= distance_yards + 330 AND sp_decimal > 15"
}

Discovery result

pass broad rule (niche) · 16 Oct, 12:56

Bets

31

6 winners

A/E

1.34

95%: 0.49 to 2.93

SP return

+15.3%

level stakes

Betfair SP return

+43.2%

30 bets with a BSP

Strike rate 19.4%. p = 0.29. 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. 2 of 6 survived.

• perturbation: Loosen trip margin and sprint cap (~+20%)pass

  59 bets · A/E 1.34 (0.67 to 2.40) · control 0.94 · 16 Oct, 12:58

  Breakdown

  betsA/E0.71.01.3

  2025-10-01 to 2026-04-2718–

  after 2026-04-27411.31

• perturbation: Tighten trip margin to 400 yards (+~20%)info

  27 bets · A/E 1.53 (0.56 to 3.32) · control 0.92 · 16 Oct, 12:58

  Breakdown

Sealed holdout (vault)

Bets

141

16 Oct, 12:58

A/E

1.64

p = 0.0068

BH threshold then

0.0029

over every vault test so far

Outcome

Failed

Forward record

• forward vs bar {'min_bets': 100, 'min_ae': 1.0, 'max_p': 0.2}info

  0 bets · · 16 Oct, 12:58

  Breakdown

  betsA/E0.71.01.3

  forward0–

Other experiments

• neighbour: <= 1340 -> 1072 (-20%)fail

  0 bets · · 16 Oct, 12:58

  Breakdown

  betsA/E0.71.01.3

  2025-10-01 to 2026-04-270–

  after 2026-04-270–

• neighbour: <= 1340 -> 1608 (+20%)pass

  56 bets · A/E 1.15 (0.53 to 2.18) · control 0.98 · 16 Oct, 12:58

  Breakdown

  betsA/E

Lineage

Agent runs

• designer v1ok · 39s · 16 Oct, 12:55

  Output (schema-checked)

  {
    "rule": "race_type = 'Flat' AND distance_yards <= 1340 AND track_stiff_finish = 1 AND last_win_distance_yards >= distance_yards + 330 AND sp_decimal <= 15",
    "refinements": [
      {
        "label": "contenders only: 8/1 or shorter",
        "rule": "race_type = 'Flat' AND distance_yards <= 1340 AND track_stiff_finish = 1 AND last_win_distance_yards >= distance_yards + 330 AND sp_decimal <= 9"
      }
    ],
    "control": {
      "description": "Primary control: the other runners in the same stiff-track Flat sprints, expected near A/E 1.00 (about 0.97 to 1.02), so any lift in the rule is not a property of these races. Secondary comparison group: the same live-system selections priced above 14/1, which carry the longshot drift seen in both trip-drop rivals (1.31 and 1.41 at 16/1+). If the hypothesis is right, the capped group beats its same-race control by a margin similar to or larger than the 16/1+ group's. If it is wrong, the capped group sits at or below 1.00 and the whole system lift sits in the control rule.",
      "rule": "race_type = 'Flat' AND distance_yards <= 1340 AND track_stiff_finish = 1 AND last_win_distance_yards >= distance_yards + 330 AND sp_decimal > 15"
    },
    "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": [
      "Small sample after the price cap: the uncapped system is already a niche, so the 14/1-or-shorter subset may fall short of 150 bets in discovery",
      "Course concentration: a few stiff sprint courses (e.g. Beverley, Carlisle, Galway) may supply most bets, so a course or draw quirk could pass for a stamina effect. Check the per-course split",
      "Season concentration: earlier trip-drop rivals lifted only after April (1.41 against 0.77). Check the split before and after April",
      "General trip-drop drift: trip-droppers at non-stiff tracks scored about 1.05 with no edge over control, so the capped rule must clearly beat that baseline as well as 1.00",
      "Column choice: the live system uses last_win_distance_yards while earlier rules used last_win_yards. Whether the two agree is unresolved, so results may not transfer to rules written on the other column",
      "Handicap and class mix: stayers dropping to sprints may be mostly low-class handicappers whose marks, not stamina, drive the result"
    ],
    "mechanism": "A stiff uphill finish turns the last furlong of a Flat sprint into a test of stamina. A horse whose most recent win came at least 330 yards (1.5f) further keeps going while the pure sprinters stop. If that is real, it should lift every horse with a genuine chance, not only outsiders. If the system's lift is just the longshot drift seen in both trip-drop rivals, capping the price at 14/1 (sp_decimal <= 15) should remove it.",
    "market_miss": "Punters read a last win over further as a sign the horse lacks the speed for a sprint and mark it down. At a stiff track that reading is wrong. Horses the market still rates as contenders (14/1 or shorter) drift beyond their true odds, because the price comes from speed-oriented form and ignores how the finish shapes the race.",
    "prediction": {
      "direction": "positive",
      "size": "A/E 1.10 to 1.30 on 150+ bets at 14/1 or shorter, clearly above the same-race control (~1.00) and the ~1.05 non-stiff trip-drop baseline",
      "falsified_if": "A/E at or below 1.00 at 14/1 or shorter, or no better than the same-race control. Either would mean the live system's lift comes from 16/1+ longshots."
    },
    "notes": "The rule keeps the sketch unchanged. last_win_distance_yards is a pre_off column (from GB/IRE 2020-09-01), and the rule uses it to match the live system SYS-stayer-dropped-to-a-stiff-sprint exactly, so this is a clean price-capped slice of that system. sp_decimal <= 15 includes 14/1 (decimal 15.0) and excludes 16/1+, which matches the band boundary in the earlier findings. The single refinement (8/1 or shorter) follows the reasoning's 'contenders the market rates'. No turf or season refinement is pre-registered; those are confounder checks only. If the capped rule comes in thin but points the right way, it goes to falsification and forward tracking as lab policy allows, rather than being loosened."
  }

Timeline

1. Job 1514 finished

   16 Oct, 12:58 · engine:compute-1

2. hypothesis note H-4e7be756

   16 Oct, 12:58 · engine:perturb

3. perturbation result for H-4e7be756 (neighbour: drop: distance_yards <= 1340): pass

   16 Oct, 12:58 · engine:perturb

4. perturbation result for H-4e7be756 (neighbour: drop: race_type = 'Flat'): pass

   16 Oct, 12:58 · engine:perturb

5. perturbation result for H-4e7be756 (neighbour: <= 15 -> 18 (+20%)): pass

   16 Oct, 12:58 · engine:perturb

6. perturbation result for H-4e7be756 (neighbour: <= 15 -> 12 (-20%)): pass

   16 Oct, 12:58 · engine:perturb

7. perturbation result for H-4e7be756 (neighbour: <= 1340 -> 1608 (+20%)): pass

   16 Oct, 12:58 · engine:perturb

8. perturbation result for H-4e7be756 (neighbour: <= 1340 -> 1072 (-20%)): fail

   16 Oct, 12:58 · engine:perturb

9. compute-1 took job 1514 (mutate, attempt 1)

   16 Oct, 12:58 · compute-1

10. Job 1512 finished

    16 Oct, 12:58 · engine:compute-1

11. forward result for H-4e7be756 (forward vs bar {'min_bets': 100, 'min_ae': 1.0, 'max_p': 0.2}): info

    16 Oct, 12:58 · engine:evaluator

12. compute-1 took job 1512 (forward_check, attempt 1)

    16 Oct, 12:58 · compute-1

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

    16 Oct, 12:58 · engine:scheduler

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

    16 Oct, 12:58 · engine:scheduler

15. Job 1510 finished

    16 Oct, 12:58 · engine:compute-1

16. H-4e7be756 Vault → SHADOW vault SUPPORTIVE: A/E 1.635 on 141 bets, p 0.006799 against BH threshold 0.002941 over 34 vault tests

    16 Oct, 12:58 · engine:orchestrator