Rival: second favourites at 5/1+ in big handicaps also beat their price (overround effect, not favourite status)

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
   • H199-favourite-at-a-big-price-in-a-big-handicap (H199: favourite at a big price in a big handicap) was GREEN in the earlier research: A/E 1.14 on 1086 bets (periods 1.16 / 1.08 / 1.23); now PROMISING. O-a934fc73
2. AskedIs the lift specific to the favourite, or do all short-end runners at 5/1+ in big-field handicaps beat their prices (an overround or price-band effect)? answered
3. MechanismIn 14+ runner handicaps the overround is loaded onto outsiders, so the front of the market is relatively generous. The second favourite at 6.0+ sits in the same part of the book as the favourite, so if overround is the cause it gets the same advantage.
4. Why the market misses itBookmakers spread their margin unevenly across the field, and A/E measured against prices normalised for the whole book credits the shortest runners. Nothing in the form book shows this.
5. Predictedpositive, A/E 1.06 to 1.12 on 1000+ bets if the rival holds
6. Falsified ifSecond favourites score A/E 1.00 or below, which would leave favourite status as the active ingredient
7. Rule sketchis_handicap = 1 AND sp_rank = 2 AND sp_decimal >= 6 AND field_size >= 14

Mechanism and rationale

In 14+ runner handicaps the overround is loaded onto outsiders, so the front of the market is relatively generous. The second favourite at 6.0+ sits in the same part of the book as the favourite, so if overround is the cause it gets the same advantage.

The rule

is_handicap = 1 AND sp_rank = 2 AND is_favourite = 0 AND sp_decimal >= 6 AND field_size >= 14

Reads: is_handicap, sp_rank, is_favourite, sp_decimal, field_size

Pre-registration

Spec hash (sha256 of the canonical pre-spec)

ab8fef557233fc11fef3feb63c63f31ab583a989a7061df3a654426d2f0790d0

Frozen

16 Oct, 15:49

Nothing in the definition can change after this.

Confounders

• Joint favourites: a co-favourite can carry sp_rank 2 or tie at rank 1, so is_favourite = 0 is required to keep favourites out of the rival group
• General favourite-longshot bias: any runner at the front of the market beats par against whole-book normalised prices, so the edge has to show against race-mates, not only above 1.0
• Overlap with H235 (rank 2, 8.0+, 16+ runners, 1.19 on 627): most of H235's bets sit inside this rule, so this is not an independent confirmation, and a lift confined to the 16+/8.0+ corner would just repeat H235
• Festival and big-Saturday handicaps: deep, liquid markets with heavy outsider margins could make up a large share of the sample and concentrate the effect
• Price basis: the H199 edge was positive at BSP but negative at SP, so an A/E above 1 against normalised SP may not be a profitable bet and should be judged at BSP
• Period concentration: H199's latest period was thin, so check that the lift is not carried by a single period

Control

{
  "description": "Two comparisons. (1) The lab's standard one: the other runners in the same races. If the margin is loaded onto outsiders, they should score below par as a group (about 0.93 to 0.98), mostly in the 16/1+ tail, and the gap between the selection and its race-mates is the effect. (2) The H199 favourite in the same kind of race (same handicap, price and field thresholds) is the comparison the question needs. If second favourites come in near the favourites' A/E (around 1.10 or higher), the overround explanation holds. If favourites stay near 1.14 and second favourites sit near 1.00, favourite status is what matters.",
  "rule": "is_handicap = 1 AND is_favourite = 1 AND sp_decimal >= 6 AND field_size >= 14"
}

Discovery result

pass broad rule (niche) · 16 Oct, 15:46

Bets

721

94 winners

A/E

1.16

95%: 0.94 to 1.42

SP return

-11.4%

level stakes

Betfair SP return

+9.3%

693 bets with a BSP

Strike rate 13.0%. p = 0.087. 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

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

• perturbation: Looser thresholds: 4/1+ second favourite in 12+ runner handicapspass

  1,998 bets · A/E 1.07 (0.95 to 1.20) · control 0.99 · 16 Oct, 15:49

  Breakdown

  betsA/E0.71.01.3

  2025-10-01 to 2026-04-279311.11

  after 2026-04-271,0671.04

• perturbation: Tighter field: second favourite at 6.0+ in 16+ runner handicaps, price capped at 8.0pass

  333 bets · A/E 1.24 (0.91 to 1.64) · control 0.97 · 16 Oct, 15:49

Sealed holdout (vault)

Bets

4,207

16 Oct, 15:49

A/E

1.04

p = 0.20

BH threshold then

0.0022

over every vault test so far

Outcome

Failed

Forward record

• forward vs bar {'min_bets': 150, 'min_ae': 1.03, 'max_p': 0.2}info

  0 bets · · 16 Oct, 15:49

  Breakdown

  betsA/E0.71.01.3

  forward0–

Other experiments

• neighbour: >= 6 -> 5 (-20%)pass

  1,125 bets · A/E 1.10 (0.93 to 1.28) · control 0.99 · 16 Oct, 15:46

  Breakdown

  betsA/E0.71.01.3

  2025-10-01 to 2026-04-275061.17

  after 2026-04-276191.04

• neighbour: >= 6 -> 7 (+20%)pass

  306 bets · A/E 1.34 (0.96 to 1.82) · control 0.97 · 16 Oct, 15:46

  Breakdown

Lineage

Agent runs

• designer v1ok · 49s · 16 Oct, 15:45

  Output (schema-checked)

  {
    "rule": "is_handicap = 1 AND sp_rank = 2 AND is_favourite = 0 AND sp_decimal >= 6 AND field_size >= 14",
    "refinements": [
      {
        "label": "14/1 or shorter",
        "rule": "is_handicap = 1 AND sp_rank = 2 AND is_favourite = 0 AND sp_decimal >= 6 AND field_size >= 14 AND sp_decimal <= 15"
      },
      {
        "label": "favourite also at 5/1+ (whole front of book long, mirrors H199 races)",
        "rule": "is_handicap = 1 AND sp_rank = 2 AND is_favourite = 0 AND sp_decimal >= 6 AND field_size >= 14 AND favourite_sp >= 6"
      }
    ],
    "control": {
      "description": "Two comparisons. (1) The lab's standard one: the other runners in the same races. If the margin is loaded onto outsiders, they should score below par as a group (about 0.93 to 0.98), mostly in the 16/1+ tail, and the gap between the selection and its race-mates is the effect. (2) The H199 favourite in the same kind of race (same handicap, price and field thresholds) is the comparison the question needs. If second favourites come in near the favourites' A/E (around 1.10 or higher), the overround explanation holds. If favourites stay near 1.14 and second favourites sit near 1.00, favourite status is what matters.",
      "rule": "is_handicap = 1 AND is_favourite = 1 AND sp_decimal >= 6 AND field_size >= 14"
    },
    "pass_criteria": {
      "min_bets": 300,
      "min_ae": 1.05,
      "max_p": 0.05
    },
    "forward_bar": {
      "min_bets": 150,
      "min_ae": 1.03,
      "max_p": 0.2
    },
    "confounders": [
      "Joint favourites: a co-favourite can carry sp_rank 2 or tie at rank 1, so is_favourite = 0 is required to keep favourites out of the rival group",
      "General favourite-longshot bias: any runner at the front of the market beats par against whole-book normalised prices, so the edge has to show against race-mates, not only above 1.0",
      "Overlap with H235 (rank 2, 8.0+, 16+ runners, 1.19 on 627): most of H235's bets sit inside this rule, so this is not an independent confirmation, and a lift confined to the 16+/8.0+ corner would just repeat H235",
      "Festival and big-Saturday handicaps: deep, liquid markets with heavy outsider margins could make up a large share of the sample and concentrate the effect",
      "Price basis: the H199 edge was positive at BSP but negative at SP, so an A/E above 1 against normalised SP may not be a profitable bet and should be judged at BSP",
      "Period concentration: H199's latest period was thin, so check that the lift is not carried by a single period"
    ],
    "mechanism": "In 14+ runner handicaps the bookmakers load most of the overround onto the many outsiders. The front of the market is therefore priced relatively generously against a whole-book normalised expectation. The second favourite at 5/1 or bigger sits in that same generous part of the book as the H199 favourite. If the H199 lift comes from where the price sits in the book and not from favourite status, the second favourite should get a lift of similar size.",
    "market_miss": "The margin is spread unevenly across the field, heaviest on long shots, so normalising by the whole book under-states the true chance of the front two or three in the betting. Nothing in the form or the ratings shows this. It is a property of how big-field handicap books are built, and SP-based punters cannot exploit it because the SP itself carries the margin. It may only be capturable at exchange prices.",
    "prediction": {
      "direction": "positive",
      "size": "A/E 1.06 to 1.12 on roughly 1000+ bets, about the size of the H199 favourite lift",
      "falsified_if": "The broad rule scores A/E 1.00 or below on 300+ bets. Also falsified, for the overround explanation, if it scores clearly below the H199 favourite control (say 0.08+ lower) and does not beat its race-mates, which would point to favourite status as the active ingredient."
    },
    "notes": "The rule sketch was kept, with is_favourite = 0 added to exclude joint favourites. No book_pct refinement was registered, because the catalogue does not say what scale it uses and a threshold could not be pre-set safely. The 14/1 cap tests the usual finding that edges die at long prices. The favourite_sp >= 6 refinement restricts the sample to races where the whole front of the book is long, which is the setting the mechanism describes. Read the result against H235, which is largely contained in this rule: if the lift appears only when field_size >= 16 and sp_decimal >= 8, it is H235 again, not support for the broader overround explanation. Judge profitability at BSP (see the H199 finding)."
  }

Timeline

1. Job 1921 finished

   16 Oct, 15:49 · engine:compute-1

2. forward result for H-601a8566 (forward vs bar {'min_bets': 150, 'min_ae': 1.03, 'max_p': 0.2}): info

   16 Oct, 15:49 · engine:evaluator

3. compute-1 took job 1921 (forward_check, attempt 1)

   16 Oct, 15:49 · compute-1

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

   16 Oct, 15:49 · engine:scheduler

5. Job 1919 finished

   16 Oct, 15:49 · engine:compute-1

6. H-601a8566 Vault → SHADOW vault SUPPORTIVE: A/E 1.04 on 4207 bets, p 0.1988 against BH threshold 0.002222 over 45 vault tests

   16 Oct, 15:49 · engine:orchestrator

7. vault result for H-601a8566 (vault query (supportive)): pass

   16 Oct, 15:49 · engine:orchestrator

8. H-601a8566 Frozen → FROZENvault vault query started

   16 Oct, 15:49 · engine:orchestrator

9. compute-1 took job 1919 (vault, attempt 1)

   16 Oct, 15:49 · compute-1

10. Job 1918 finished

    16 Oct, 15:49 · engine:compute-1

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

    16 Oct, 15:49 · engine:scheduler

12. H-601a8566 Falsification → FROZEN survived falsification

    16 Oct, 15:49 · engine:falsification

13. falsification result for H-601a8566 (partition: Flat handicaps only): pass

    16 Oct, 15:49 · engine:evaluator

14. compute-1 took job 1918 (attack, attempt 1)

    16 Oct, 15:49 · compute-1

15. Job 1917 finished

    16 Oct, 15:49 · engine:compute-1

16. falsification result for H-601a8566 (control: Control: same second favourites in mid-size handicaps (8 to 13 runners)): fail

    16 Oct, 15:49 · engine:evaluator

17. compute-1 took job 1917 (attack, attempt 1)

    16 Oct, 15:49 · compute-1

18. Job 1916 finished

    16 Oct, 15:49 · engine:compute-1

19. falsification result for H-601a8566 (partition: Outside the H235 corner (drop 16+ runners at 8.0+)): pass

    16 Oct, 15:49 · engine:evaluator