SYS1: pulled up poor jumper as a rule

The notebook

In order: what the lab saw, what it asked, what it proposed and why, what the tests said, and what an investigator concluded. Numbers are aggregates on the discovery data unless a step says otherwise.

1. Hypothesised15 Oct, 23:10

   SYS1: pulled up poor jumper as a rule KILLED

   because Live rule without the run style: parent + earlier Jump Index standing in the worse 40%.

   race_type IN ('Chase', 'Hurdle') AND last_position IS NULL AND l_same_code = 1 AND sp_decimal <= 15 AND s_jump_index_pct_prev IS NOT NULL AND s_jump_index_pct_prev <= 0.4

   made by import

2. Tested15 Oct, 23:10

   HOLDOUT 2023-25, all: A/E 0.95 on 715 bets, by period 0.95. info

   Every period below 1.00. Zone: all. On SYS1: pulled up poor jumper as a rule

   By period

   betsA/E0.71.01.3

   2020-220–

   2023 to a year ago7150.95

   Last year0–

3. Tested15 Oct, 23:10

   Britain: A/E 0.97 on 570 bets, by period 0.97. info

   Every period below 1.00. Zone: all. On SYS1: pulled up poor jumper as a rule

   By period

   betsA/E0.71.01.3

   2020-220–

   2023 to a year ago5700.97

   Last year0–

4. Tested15 Oct, 23:10

   Ireland: A/E 0.86 on 145 bets, by period 0.86. info

   Every period below 1.00. Zone: all. On SYS1: pulled up poor jumper as a rule

   By period

   betsA/E0.71.01.3

   2020-220–

   2023 to a year ago1450.86

   Last year0–

5. Tested15 Oct, 23:10

   discovery year (reference): A/E 1.20 on 420 bets, by period 1.20. info

   Betfair SP return +15.1%. Every period above 1.00 (spread 0.00). Zone: all. On SYS1: pulled up poor jumper as a rule

   By period

   betsA/E0.71.01.3

   2020-220–

   2023 to a year ago0–

   Last year4201.20

6. Promoted15 Oct, 23:10

   SYS1: pulled up poor jumper as a rule Falsification → FROZENimport: frozen

7. Killed15 Oct, 23:10

   SYS1: pulled up poor jumper as a rule Vault → KILLEDfailed the vaultimport: holdout CONTRADICTED: A/E 0.948 on 715 (GB 0.97, IRE 0.857), p 0.6981 against BH 0.003448 over 29 tests