IQ50: costly jumping out of character

Mechanism and rationale

Last run, in today's code: bottom fifth of the field for ground gained jumping, slower half to recover, completed and beaten, from a horse whose earlier runs averaged in the better 40%. A bad day from a good jumper.

The rule

race_type IN ('Chase', 'Hurdle') AND l_same_code = 1 AND l_lgj_pct <= 0.2 AND l_recovery_time_pct >= 0.5 AND last_position > 1 AND s_lgj_pct_prev >= 0.6

Reads: race_type, l_same_code, l_lgj_pct, l_recovery_time_pct, last_position, s_lgj_pct_prev

Pre-registration

Spec hash (sha256 of the canonical pre-spec)

860d1bb1435a5eea84d77877dc282d7203b0b81da41b3c262df5efe98baf8e5b

Frozen

13 Oct, 01:00

Nothing in the definition can change after this.

Direction

positive

Frozen in

research/experiments/raceiq/iq_holdout

Frozen on

2026-10-13

Id

IQ50-costly-jumping-out-of-character

Rule

race_type IN ('Chase', 'Hurdle') AND l_same_code = 1 AND l_lgj_pct <= 0.2 AND l_recovery_time_pct >= 0.5 AND last_position > 1 AND s_lgj_pct_prev >= 0.6

Result (from the archive)

info discovery year (reference) · 15 Oct, 23:10

Bets

815

A/E

1.05

winners vs the prices

SP return

-16.4%

level stakes

Betfair SP return

+12.7%

815 bets with a BSP

Strike rate 9.7%.

Falsification

No attacks yet. Only promising ideas are attacked.

Sealed holdout (vault)

Bets

1,340

14 Oct, 13:00

A/E

1.16

p = 0.040

BH threshold then

0.0034

over every vault test so far

Outcome

Failed

from the archive

Forward record

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

  0 bets · · 15 Oct, 23:21

  Breakdown

  betsA/E0.71.01.3

  forward0–

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

  0 bets · · 16 Oct, 23:21

  Breakdown

  betsA/E0.71.01.3

  forward0–

Lineage

No parent and no descendants yet. When an agent mutates, generalises or splits this idea, the new hypotheses appear here.

Agent runs

No agent has worked on this yet.

Timeline

1. Job 2097 finished

   16 Oct, 23:21 · engine:compute-1

2. forward result for IQ50-costly-jumping-out-of-character (forward vs bar {'min_bets': 100, 'min_ae': 1.0, 'max_p': 0.2}): info

   16 Oct, 23:21 · engine:evaluator

3. compute-1 took job 2097 (forward_check, attempt 1)

   16 Oct, 23:21 · compute-1

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

   16 Oct, 23:21 · engine:scheduler

5. Job 5 finished

   15 Oct, 23:21 · engine:compute-2

6. forward result for IQ50-costly-jumping-out-of-character (forward vs bar {'min_bets': 100, 'min_ae': 1.0, 'max_p': 0.2}): info

   15 Oct, 23:21 · engine:evaluator

7. compute-2 took job 5 (forward_check, attempt 1)

   15 Oct, 23:21 · compute-2

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

   15 Oct, 23:21 · engine:scheduler

9. IQ50-costly-jumping-out-of-character Vault → SHADOW import: holdout SUPPORTIVE: A/E 1.158 on 1340 (GB 1.152, IRE 1.189), p 0.0396 against BH 0.003448 over 29 tests; LEARNINGS read it as holding

   15 Oct, 23:10 · importer

10. IQ50-costly-jumping-out-of-character Frozen → FROZENvault import: vault

    15 Oct, 23:10 · importer

11. IQ50-costly-jumping-out-of-character Falsification → FROZEN import: frozen

    15 Oct, 23:10 · importer

12. import result for IQ50-costly-jumping-out-of-character (discovery year (reference)): info

    15 Oct, 23:10 · importer

13. import result for IQ50-costly-jumping-out-of-character (Ireland): info

    15 Oct, 23:10 · importer

14. import result for IQ50-costly-jumping-out-of-character (Britain): info

    15 Oct, 23:10 · importer

15. import result for IQ50-costly-jumping-out-of-character (HOLDOUT 2023-25, all): info

    15 Oct, 23:10 · importer

16. New idea IQ50-costly-jumping-out-of-character: IQ50: costly jumping out of character (import)

    15 Oct, 23:10 · importer