Old-school intuition → Class relief modelled through domestic rating
A horse drops back from open company to rating 75 and suddenly looks impossible to beat. Every punter scans the form for this — the horse that's been racing in stronger fields and finally finds one it can handle. But class relief isn't binary.
A horse coming back from open class to rating 75 is one thing. A horse coming back from Group company to rating 75 is one thing. A horse dropping from rating 85 to rating 75 is different again.
What we did: We baked the horse's domestic rating directly into the model — the same NZTR rating that determines what grade the horse races in.
But we didn't stop there.
We also computed its rating versus the field it is up against: the difference between this horse's rating and the average rating of every other horse in the race.
This single feature captures class relief in one number. When a rating-85 horse lines up against an average field rating of 70, the model sees a +15 gap and adjusts the win probability accordingly — just like you would, but applied across every runner simultaneously.
The model also tracks how a horse performed against higher-rated fields (did it compete or get beaten off?), and how it performed when dropping back in class previously.
A horse that consistently runs well when stepping down gets a bigger boost than one that simply can't win at any grade.
Old-school intuition → First 400m sectionals + pace analysis
Trentham in autumn rides differently than Trentham in spring, even at the same official track rating. The surface settles, the rail moves, the prevailing wind direction shifts.
A "Dead 5" in September is not the same as a "Dead 5" in March. Your brain knows this — you've been watching it for years.
What we did: We built a track bias engine that analyses every race's first‑400m sectional time, leader win rate, and on‑pace/off‑pace strike rate. It tracks this per track, per condition band, per 90‑day window.
So the model knows that Cambridge Synthetic in June rides like a Firm-Dead track — not the same as Heavy 10 Trentham in winter.
Sectional timing data on 140,000+ NZ race sectionals means the model has seen pace patterns for every track–weather–season combination that exists.
It doesn't guess — it compares today's race to the most similar historical races and adjusts win probabilities accordingly.
Old-school intuition → Recency-weighted jockey performance
Some jockeys just suit certain tracks. They know the camber of the turn, the best place to make a run, how the track rides after rain.
A form guide might tell you a jockey's lifetime strike rate — but what you really want to know is: how have they been going lately at this specific track?
What we did: We compute jockey–track win rates, jockey–trainer combination stats, and jockey–track–trainer triple stats using an Exponentially Weighted Moving Average (EWMA).
This means a jockey who has ridden 8 winners from 20 rides at Te Rapa in the last 90 days gets a much bigger lift than one whose last Te Rapa win was 18 months ago. The model picks up form cycles faster than any static rating.
The same technique powers all our recency-weighted features — trainer form, horse class performance, and even barrier‑condition edges. Old form decays, hot streaks amplify.
Old-school intuition → Continuous distance profiling
You see a horse stepping up to 2000m for the first time. If it has the breeding for a stayer and closed off strongly over 1400m last start, you know it's ready.
If it's been weakening over 1600m, you know it's not. This is one of the oldest skills in punting — reading a horse's distance trajectory.
What we did: We created distance-profiling features that look at a horse's entire career — best winning distance, average distance of recent starts, performance trend as distance increases, and a distance specialist score.
But more importantly, we built win-rate-at-distance and win-rate-at-distance-by-condition features that compare each horse's actual record at today's distance against the rest of the field.
The model doesn't just check "has this horse won at 2000m before?" It computes a continuous distance-aptitude score from every start at every distance.
A horse that won at 1600m, ran second at 1800m, and carries the right pedigree for 2000m scores highly — even if it has never actually run 2000m. Human intuition does the same thing, but the model does it for all active NZ horses simultaneously.
Old-school intuition → 11 independent trainer profile features
A trainer who hasn't had a winner in 60 days is either out of form or out of ammunition. You check the trial results, you read the stable's strike rate, and often you just skip the race.
But a blanket "cold trainer" label misses the nuance — some trainers lose form everywhere, while others just struggle at one specific track.
What we did: Every horse gets 11 different trainer profile features: overall win rate, track win rate, distance win rate, track‑condition win rate, recent 90‑day strike rate, and more.
These are sourced from two independent pipelines — so the data is cross‑checked. The model treats a trainer who is 3-from-8 in the past month very differently from one who is 3-from-60.
It also knows the difference between a trainer who can't win at Ellerslie specifically versus one who can't win anywhere right now.
That granularity is impossible to capture in a static rating — but the model learns it automatically from hundreds of thousands of historical races.
Want to know more about us then click this link.
We didn't build a black box that eats numbers and spits out picks. We built a system that starts with the same intuition a good punter has — then scales it across every horse, every jockey, every trainer, every track, every condition, every race in New Zealand.
The machine does the math. The intuition was always ours.
© 2026 Winningpost.co.nz. All rights reserved.
No part of this website may be reproduced or shared without written permission.