RatingsModel V8
Building a horse racing ratings model.
Using:
• Official Rating (OR)
• Racing Post Rating (RPR)
• Market odds (implied probability)
Model built the night before to capture early prices.
Odds then updated ~ 60 mins before the off to track market movement.
Each race produces:
🟢 Model Pick
🟠 Main Threat
🟡 Value Pick
All races logged.
All results tracked transparently.
75 race test phase.
Quick update.
Didn’t post over the weekend due to personal reasons. I’ll also be quiet for a few more days.
Everything still ticking over behind the scenes, just no posts for now.
Back soon.
📊 RatingsModel – Today’s Races & Model Picks
No bets today ❌
Model ran across 6 races and found no edge
Races & model picks:
• 13:45 Newmarket → Double Rush
• 14:20 Newmarket → Mister Winston
• 14:45 Goodwood → Novelista
• 16:45 Newmarket → Gamrai
• 16:50 Uttoxeter → Pony Soprano
• 18:45 Hexham → Seeyouinmydreams
Summary:
❌ Favourites overbet
❌ No clear gaps
❌ Value in the wrong places
👉 No bets = disciplined day
We go again tomorrow 🔁
Nothing today ❌
Went through the cards properly, a few that looked tempting on the surface but once you dig in there’s no real edge.
These are the days that save your bankroll 💰 Discipline over action.
We go again tomorrow 🔁
No picks today 📊
Ran the usual process, filtered the card, and there’s just no real edge.
Plenty of races that look playable on the surface, but:
weak structure
compressed markets
no genuine value
A few were “value only” setups, but with no qualifying value, they’re automatic passes.
That’s part of the game.
Discipline > forcing bets.
We move 👍
End of Day Summary 📊
Solid session. More importantly, the model behaved as designed.
---
🎯 Model Picks
- Race 1 → 1st ✅
- Race 2 → 3rd
- Race 3 → 1st ✅
- Race 4 → 4th
- Race 5 → 1st ✅
3/5 winners found by the model.
---
⚠️ AVOID races
- Tabby → 1st (race rated no edge)
- Wondrous Light → 4th
- Al Hussar → 3rd
Mixed results, but outcomes were unstable / non-dominant, exactly why they were avoided.
---
💰 Value Picks
- Alessia Fernanda (14/1) → 1st 🟡
- Dicko The Legend (10/1) → 2nd 🟡
Race 4 is a good example:
Model pick didn’t win, but value did in a race flagged as no edge.
---
🧠 Key takeaway
The model can find winners.
The edge comes from race selection, not just picks.
✔ Structured races → clear outcomes
❌ Weak races → unpredictable
That separation held again today.
---
📈 Still early, still building data
No changes, no chasing
Back tomorrow 👊
That’s the card for today.
✔️ 5 races analysed
❌ 4 skipped
✅ 1 bet
No forcing it, just following the model.
Quality > quantity every time.
Good luck if your having a bet.
THREAD 🧵 | RatingsModel
Structured day today, not chasing volume, just backing the right spots.
Plenty of races analysed, but most don’t meet the bar.
✔️ Clear plan
✔️ Disciplined filtering
✔️ No forcing bets
Let’s get into it 👇
Race 5 - 18:15 Southwell
❌ NO BET
🟢 Model Pick: Tabby (7/2)
🟠 Main Threat: Neyva’s Angel (4/1)
🟡 Value Pick: Dicko The Legend (10/1)
Top 3 tightly grouped, no separation at all.
Pure coin flip, stay out.
RatingsModel – Today’s Plan 🧠
Keeping it disciplined today, only targeting the right race types.
🎯 Core Races
15:15 Musselburgh
15:30 Southwell
⚖️ Support (only if model is strong)
16:05 Southwell
18:15 Southwell
👀 Borderline
14:05 Musselburgh (needs a clear signal or it’s a skip)
No forcing bets, no chasing, quality over quantity.
Let’s see what the model says.
No play today.
Ran the full card across Punchestown, Epsom, Yarmouth, Southwell and Lingfield, nothing met the model criteria.
Plenty of races, but no real gaps, weak class, and too much noise.
That’s part of the edge.
Discipline over action. Back tomorrow.
Late one from me… was at United last night so catching up this morning ⚽️
First full day running the refreshed RatingsModel and it gave us a proper mix to analyse.
✔️ **What went well**
* Model picked a winner early (Balgowan)
* Consistent top 3 identification in several races
* AVOID calls were largely justified, especially in the true chaos races
* No reckless chasing, discipline held
⚠️ **What we learned**
* Some “avoid” races still produced winners within the model ranks, just no clear edge
* Value picks hit in the right race type, but only when structure allowed it
* Big fields + weak gaps = exactly the kind of races to stay out of
* Not all low-gap races are chaos, some are just tightly packed
📊 **Big takeaway**
The model is doing its job.
It’s finding the right horses more often than not.
The next step is improving how we classify races and decide how to play them.
That’s where the edge will come from.
Back at it today 👊
Good luck if you’re playing.
Ratings Model – 27 Apr 2026 🎯
Back to standard service today.
Yesterday’s Eliminator didn’t land ❌
Too many chaotic races, not enough edge.
Today we reset.
Plan is simple:
👉 Only play clear model edges
👉 Skip the noise
👉 Discipline over volume
We’ll be using the model as usual:
🟢 Model Pick
🟠 Main Threat
🟡 Value Pick
🔵 Place Chance
Not every race is a betting race.
Today we wait for the right setups.
Picks below 👇
CORRECTION - 14:00 Bath
🟢 Model Pick: Bohemian Breeze
🟠 Main Threat: Moon Over The Sea
🟡 Value Pick: Itsgottabefun
📊 Balanced race, no real edge
❌ Gap not strong enough
Play: NO BET
15:30 Bath
🟢 Model Pick: Bohemian Breeze
🟠 Main Threat: Moon Over The Sea
🟡 Value Pick: Itsgottabefun
📊 Balanced race, no real edge
❌ Gap not strong enough to trust
Play: NO BET
We move on 👇