August 2026 Validation

Source document: reports/Breeding_validation_August_2026.docx Figures taken against the live databases on 5 September 2026.

What was tested and why

The rebuilt breeding table was checked against an independent body of data: the runners that actually raced in Australia on four days in August 2026. Race data and pedigree data were assembled separately and from different origins, so agreement between them is evidence, and disagreement identifies which side is wrong.

Three questions were asked of each runner:

1. Is the horse itself in breeding?
2. Is its sire, at the stated name, country and year of foal?
3. Is its dam?

Each runner and each stated parent was matched on the identity triple. Where a northern-hemisphere parent failed on the exact year, a second attempt was made at year + 1. A match was accepted only where exactly one candidate survived; anything else was recorded as a failure rather than resolved by guesswork.

Results

	1 Aug	2 Aug	8 Aug	9 Aug	Total
Runners checked	1,212	346	1,117	288	2,963
Runners absent from breeding	3	2	0	0	5
Sires not matched	0	0	0	1	1
Dams not matched	5	0	5	1	11
Errors identified in breeding	0	0	0	0	0

Across 2,963 runners and 5,926 stated parents, no error was identified in breeding on any of the four days.

The failures, and what they actually were

Eleven of twelve parent failures were wrong years in the race data

Every one of these mares is present in breeding. They failed because the year the race data states is wrong — and in each case the damsire, which the race data itself supplies, identifies which mare is meant and confirms the breeding record.

Runner	Dam	Race-data year	breeding year	Damsire stated	Dam's sire in breeding
PRAGUE PRIDE	DEVITA (AUS)	0	2016	SNITZEL	SNITZEL
RONNY'S SECRET	MONTANA'S SECRET (AUS)	0	2005	DASH FOR CASH	DASH FOR CASH
ACROPHOBIC	SENSE OF HITE (NZ)	2016	2010	STREET SENSE	STREET SENSE
DEVOTED TO YOU	NOT ONLY FLORINA (IRE)	2011	2013	MAKFI	MAKFI
FEAR NO EVIL	ARBOREAL (AUS)	2011	2015	LONHRO	LONHRO
BLUE ISLES	LADY KITSCH (AUS)	2016	2015	PROFILE IN STYLE	PROFILE IN STYLE
FORTY TIMES	TWENTY ONE TIMES (NZ)	2011	2013	O'REILLY	O'REILLY
LE TROISIR	TEARDROPS (AUS)	2011	2014	FASTNET ROCK	FASTNET ROCK
LILY'S DASH	UNBROKEN (AUS)	0	2013	SHIELD	SHIELD
ROWDIE BLUE	PEACEANDHAPPINESS (AUS)	0	2014	TRUSTING	TRUSTING
IMMORTAL DIVA	PERFECT LOVE (AUS)	(none)	—	SHARKBITE	resolved on damsire

ACROPHOBIC is worth noting on its own. The race data gives its dam as foaled in 2016 — but ACROPHOBIC is itself a 2017 foal, and a mare foaled in 2016 cannot produce a foal in 2017. breeding gives 2010, which yields a normal seven-year gap. This error would not have been found without the pedigree table to check against.

One case remains open

MAJESTIC MORN's sire is given as HIDDEN DRAGON (AUS) foaled 2008. breeding holds five stallions of that name — AUS 1999, TUR 1999, USA 1999, SAF 2010 and IRE 2015 — none an Australian 2008 foal. It is either a wrong year in the race data or a stallion genuinely absent from both source feeds, and a sire has no damsire field to arbitrate it. Recorded as open rather than assigned to either side.

Five runners absent — a source gap, not a matching failure

Runner	Foaled	Sire	Dam
DERELICT (AUS 2020)	13/09/2020	OUTREACH (AUS) 2012 ✓	BEST OF YOU (AUS) 2006 ✓
EMPRESS ROSSA (AUS 2022)	20/08/2022	MAGNUS (AUS) 2002 ✓	SAFFROSSA (AUS) 2004 ✓
THE WILDLING (AUS 2023)	12/08/2023	WILD RULER (AUS) 2017 ✓	TARABAI (AUS) 2014 ✓
TROFEO (AUS 2022)	13/10/2022	MAURICE ✓	COURTYBELLA ✓
THE CAN CAN (AUS 2023)	20/09/2023	BRUTAL ✓	BAILARIN ✓

All are absent from both source feeds, so no matching rule could have found them. Every parent of every missing foal is present and correct, and the dams' other foals are in breeding — Best Of You has eight, Saffrossa seven, Tarabai three. This is dropout of individual horses, not of families.

The gap runs at roughly one horse in six hundred across the four days, and was confined to 1 and 2 August; all 1,405 runners of 8 and 9 August were present.

Why the year differs, and which is right

Seventeen parents across the four days were located only after allowing the year to differ by one. Every one is a northern-hemisphere horse recorded one year lower by the race data. Not one southern-hemisphere parent needed the allowance.

Recurring cases are consistent every time: ELLIPTICAL ORBIT (USA) appears three times, WEDDING VEIL (IRE) and MANDISTANA (FR) twice each.

Worked example — ZIGELLO (IRE)

Checked directly against the Australian Stud Book. Her own record gives her as an Irish mare foaled 30 January 2016 — a date that admits no other reading of her foaling year.

Source	States her year as
Her own studbook record (hid 1173408)	2016
Her own PedigreeQuery page	2016
breeding	2016
Miss Ziggy (AUS 2020), naming her as dam	2015
Terzima (AUS 2021), naming her as dam	2015
Zig Zag Torque (AUS 2022), naming her as dam	2015

The pattern: a horse's own record carries its true foaling year, while an Australian-sourced reference to that horse as a parent quotes the season year, one lower. It applies to northern-hemisphere horses foaled in the first half of the year and never to southern-hemisphere horses — verified across 17,206 northern records holding a real date of birth, with no exceptions in any month.

All three of Zigello's foals linked to the correct mare, and the stated 2015 is preserved on each foal's record so the discrepancy stays visible rather than being silently overwritten.

Recommendations from the report

1. Match parents on name, country and year of foal, permitting year + 1 for northern-hemisphere parents only. Matching on the exact year alone would have failed to link seventeen parents across these four days — silently, with no record that anything was missed.
2. Do not relax the year requirement further. Where a parent cannot be identified on those terms, leave it unlinked and report it. Twelve such cases arose and eleven were resolved by hand within minutes using the damsire. A person reviewing five exceptions is a better outcome than an automatic rule linking to the wrong mare.
3. Treat breeding as the reference for horse identity and parentage. On this evidence it was correct in every contested case, and it identified at least one error in the race data that nothing else was positioned to catch.
4. Consider backfilling the missing foals from the race data, where both parents resolve and the damsire agrees. The race data holds the foaling date, sex, both parents and the breeder, so these records are reconstructable.

Recommendations 1 and 2 are implemented — see Breeding Rules §5. Recommendation 4 is open work.