Wide trip at a sharp track last time costs more ground, so it pays more

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
   • IQ140a-wide-trip-with-figures (IQ140a: wide trip with figures) was GREEN in the earlier research: A/E 1.27 on 1080 bets (periods 1.15 / 1.41); now PROMISING. O-943a9f88
2. AskedIs the wide-trip edge bigger when last time's track had tight bends, where going wide costs the most ground? answered
3. MechanismA horse forced wide round a bend covers extra distance. On a tight, sharp track the bends are tighter and more frequent for the trip, so it loses more ground than on a galloping track with long straights. A close finish after more lost ground means more hidden ability.
4. Why the market misses itThe form book records a beaten distance and maybe 'wide' in the comment, but not how much ground the track shape cost. Punters read 'beaten 3 lengths' the same at Chester and Newmarket.
5. Predictedpositive, A/E 1.30 to 1.45 on 300+ bets, above the 1.27 for the whole wide-trip population
6. Falsified ifA/E at or below the galloping-track wide runners (last_track_sharpness = 0), or below 1.15, or the two halves disagree
7. Rule sketchlast_raced_wide = 1 AND last_position > 1 AND last_beaten_lengths <= 5 AND last_slow_start = 0 AND last_track_sharpness >= 2

Mechanism and rationale

A horse forced wide round a bend covers extra distance. On a tight, sharp track the bends are tighter and more frequent for the trip, so it loses more ground than on a galloping track with long straights. A close finish after more lost ground means more hidden ability.

The rule

last_raced_wide = 1 AND last_position > 1 AND last_beaten_lengths <= 5 AND last_slow_start = 0 AND last_track_sharpness >= 2

Reads: last_raced_wide, last_position, last_beaten_lengths, last_slow_start, last_track_sharpness

Pre-registration

Spec hash (sha256 of the canonical pre-spec)

c4c3492039107e5e161122a5d612d10632a539b3fd1668ab72c449c1192af3ae

Frozen

16 Oct, 12:11

Nothing in the definition can change after this.

Confounders

• Field size: sharp tracks run smaller fields, so judge only on the field-size-corrected yardstick and check the same-race other runners
• Code and surface mix: sharp tracks are mostly Flat and all-weather ovals (Chester, Wolverhampton, Chelmsford, Lingfield), so a Flat or AW effect could pass for a track-shape effect; check the Flat-only refinement and the split by last_surface
• Course concentration: a few tracks may supply most bets, so report bets and A/E by last course and require the result to hold with the biggest track removed
• Overlap with the parent: this is a subset of the wide-trip system (A/E 1.27 on 1080). Beating 1.00 proves nothing; the test is whether it beats the galloping-track control and the parent
• Short window: last_raced_wide exists only from November 2025, so there are two halves, not three periods, and both must agree
• Longshot artefact: beaten-last-time runners lift at 16/1+, so the 14/1-or-shorter refinement must hold up too
• The sharpness scale comes from course guides and is coarse; a few mis-graded tracks could move the boundary at 2

Discovery result

pass broad rule (niche) · 16 Oct, 11:16

Bets

612

106 winners

A/E

1.18

95%: 0.96 to 1.43

SP return

+23.7%

level stakes

Betfair SP return

+67.0%

610 bets with a BSP

Strike rate 17.3%. p = 0.054. Same-race control: the other runners in these races had A/E 0.97, so the selection beat its own races. Zone: discovery. Evaluator eval-1.0, data 2887c9abd664f46f

Falsification

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

• control: Control: sharp track, close finish, NOT raced widefail

  11,057 bets · A/E 0.98 (0.93 to 1.03) · control 1.01 · 16 Oct, 12:11

  Breakdown

  betsA/E0.71.01.3

  2025-10-01 to 2026-04-274,1140.92

  after 2026-04-276,9431.01

• price_band: Short-priced only: 8/1 or shorterpass

  378 bets · A/E 1.09 (0.88 to 1.35) · control 0.97 · 16 Oct, 12:11

Sealed holdout (vault)

Not queried. Only a frozen idea can spend its one query against the 2020-25 holdout.

Forward record

• forward vs bar {'min_bets': 100, 'min_ae': 1.05, 'max_p': 0.2}info

  0 bets · · 16 Oct, 12:12

  Breakdown

  betsA/E0.71.01.3

  forward0–

Other experiments

• neighbour: > 1 -> 0 (-20%)pass

  865 bets · A/E 1.09 (0.93 to 1.28) · control 0.98 · 16 Oct, 11:16

  Breakdown

  betsA/E0.71.01.3

  2025-10-01 to 2026-04-274221.12

  after 2026-04-274431.06

• neighbour: > 1 -> 2 (+20%)pass

  425 bets · A/E 1.31 (1.02 to 1.66) · control 0.95 · 16 Oct, 11:16

  Breakdown

Lineage

Agent runs

• designer v1ok · 38s · 16 Oct, 11:14

  Output (schema-checked)

  {
    "rule": "last_raced_wide = 1 AND last_position > 1 AND last_beaten_lengths <= 5 AND last_slow_start = 0 AND last_track_sharpness >= 2",
    "refinements": [
      {
        "label": "14/1 or shorter",
        "rule": "last_raced_wide = 1 AND last_position > 1 AND last_beaten_lengths <= 5 AND last_slow_start = 0 AND last_track_sharpness >= 2 AND sp_decimal <= 15"
      },
      {
        "label": "Flat only",
        "rule": "last_raced_wide = 1 AND last_position > 1 AND last_beaten_lengths <= 5 AND last_slow_start = 0 AND last_track_sharpness >= 2 AND race_type = 'Flat' AND last_race_type = 'Flat'"
      }
    ],
    "control": {
      "description": "The main comparison is the same wide-trip, close-finish population whose last run was at a galloping track (last_track_sharpness = 0). The mechanism says the sharp-track group should beat it clearly; if the two match, track shape adds nothing to the known wide-trip edge. The other runners in the same races should sit near 1.00 after the field-size correction. If they also come out above 1.00, the lift belongs to the races (small sharp-track fields, all-weather cards), not to the trigger.",
      "rule": "last_raced_wide = 1 AND last_position > 1 AND last_beaten_lengths <= 5 AND last_slow_start = 0 AND last_track_sharpness = 0"
    },
    "pass_criteria": {
      "min_bets": 150,
      "min_ae": 1.15,
      "max_p": 0.05
    },
    "forward_bar": {
      "min_bets": 100,
      "min_ae": 1.05,
      "max_p": 0.2
    },
    "confounders": [
      "Field size: sharp tracks run smaller fields, so judge only on the field-size-corrected yardstick and check the same-race other runners",
      "Code and surface mix: sharp tracks are mostly Flat and all-weather ovals (Chester, Wolverhampton, Chelmsford, Lingfield), so a Flat or AW effect could pass for a track-shape effect; check the Flat-only refinement and the split by last_surface",
      "Course concentration: a few tracks may supply most bets, so report bets and A/E by last course and require the result to hold with the biggest track removed",
      "Overlap with the parent: this is a subset of the wide-trip system (A/E 1.27 on 1080). Beating 1.00 proves nothing; the test is whether it beats the galloping-track control and the parent",
      "Short window: last_raced_wide exists only from November 2025, so there are two halves, not three periods, and both must agree",
      "Longshot artefact: beaten-last-time runners lift at 16/1+, so the 14/1-or-shorter refinement must hold up too",
      "The sharpness scale comes from course guides and is coarse; a few mis-graded tracks could move the boundary at 2"
    ],
    "mechanism": "A horse carried wide round a bend covers extra ground. On a sharp, tight track the bends are tighter and take up more of the trip than on a galloping track with long straights, so the same 'raced wide' note costs more distance. A horse that still finished within 5 lengths after losing more ground has more hidden ability, so the wide-trip edge should be biggest when last time's track was sharp.",
    "market_miss": "The form book gives a beaten distance and maybe 'wide' in the comment, but not how much ground the track's shape cost. Punters and the SP read 'beaten 3 lengths, raced wide' the same at Chester and Newmarket, so they under-rate the horse when the bends were tight.",
    "prediction": {
      "direction": "positive",
      "size": "A/E 1.30 to 1.45 on 300+ bets, above the 1.27 for the whole wide-trip population and clearly above the galloping-track control",
      "falsified_if": "A/E at or below the galloping-track wide runners (last_track_sharpness = 0), or below 1.15, or the two halves (Nov 2025 onwards, split at the midpoint) disagree in direction"
    },
    "notes": "I kept the rule sketch as written. Every column is pre-off, and track sharpness is available from 2020, but last_raced_wide limits the test to November 2025 onwards. The rule leaves out the parent's NOT coalesce(held_up/midfield AND tilt) clause, so it can be compared directly with the published wide-trip-beaten-under-five system. Following the hypothesis, I took sharpness 0 as galloping and 2 or more as sharp; the scale itself is not documented in the catalogue. I set the discovery A/E bar at 1.15, which matches the falsification line. Credit for the mechanism requires both: beating the galloping-track control, and holding the edge with the biggest course removed."
  }

Timeline

1. Job 1279 finished

   16 Oct, 12:12 · engine:compute-1

2. forward result for H-17cdbcfa (forward vs bar {'min_bets': 100, 'min_ae': 1.05, 'max_p': 0.2}): info

   16 Oct, 12:12 · engine:evaluator

3. compute-1 took job 1279 (forward_check, attempt 1)

   16 Oct, 12:12 · compute-1

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

   16 Oct, 12:12 · engine:scheduler

5. Job 1262 finished

   16 Oct, 12:11 · engine:compute-1

6. H-17cdbcfa Frozen → SHADOW vault skipped: last_raced_wide, last_beaten_lengths, last_slow_start do not exist before the discovery zone; track it forward instead

   16 Oct, 12:11 · engine:orchestrator

7. compute-1 took job 1262 (vault, attempt 1)

   16 Oct, 12:11 · compute-1

8. Job 1249 finished

   16 Oct, 12:11 · engine:compute-1

9. Queued vault on the compute lane: frozen: one sealed vault query

   16 Oct, 12:11 · engine:scheduler

10. H-17cdbcfa Falsification → FROZEN survived falsification

    16 Oct, 12:11 · engine:falsification

11. falsification result for H-17cdbcfa (partition: Handicaps only today): pass

    16 Oct, 12:11 · engine:evaluator

12. compute-1 took job 1249 (attack, attempt 1)

    16 Oct, 12:11 · compute-1

13. Job 1248 finished

    16 Oct, 12:11 · engine:compute-1

14. falsification result for H-17cdbcfa (partition: Turf only last time (sharp turf bends, not AW ovals)): pass

    16 Oct, 12:11 · engine:evaluator

15. compute-1 took job 1248 (attack, attempt 1)

    16 Oct, 12:11 · compute-1

16. Job 1247 finished

    16 Oct, 12:11 · engine:compute-1

17. falsification result for H-17cdbcfa (perturbation: Beaten-lengths cut tightened to 4): pass

    16 Oct, 12:11 · engine:evaluator

18. compute-1 took job 1247 (attack, attempt 1)

    16 Oct, 12:11 · compute-1

19. Job 1246 finished

    16 Oct, 12:11 · engine:compute-1

20. falsification result for H-17cdbcfa (price_band: Short-priced only: 8/1 or shorter): pass

    16 Oct, 12:11 · engine:evaluator