How It Works Performance The Difference Sign In JOIN FREE
Results tracked publicly
67,000+ runner entries
Retrained every night

ELITE RACING EDGE.
BUILT ON DATA.
BACKED BY RESULTS.

A machine learning scanner identifies genuine positive edge against the market — not gut feel, not form opinions. Every selection is logged before the race runs. Every result published regardless of outcome. No spin. No hiding.
JOIN DURING EARLY ACCESS SEE HOW IT WORKS
Free during early access · Paid plans coming soon
Jumps model top-1
67.1%
vs ~14% random baseline
Main model top-1
58.8%
all UK flat & jumps races
Runner entries
67,000+
12 months of UK race data
vs Random Baseline
4.8×
above random baseline
Scanner Active
Selections released: 12:00 & 16:00 UK
Join to receive selections →
67.1%
Jumps value model top-1 accuracy
+18.75pts
Live P&L from 49 selections
67,000+ races
Race history powering the model
Every result
Published publicly, win or lose
Nightly retraining
Model adapts to every new race
No backdating
All selections logged before races run
Walk-forward CV
Tested on data the model has never seen
67.1%
Jumps value model top-1 accuracy
+18.75pts
Live P&L from 49 selections
67,000+ races
Race history powering the model
Every result
Published publicly, win or lose
Nightly retraining
Model adapts to every new race
No backdating
All selections logged before races run
Walk-forward CV
Tested on data the model has never seen
The difference

NOT ANOTHER TIPSTER.

The UK tipster market is full of accounts that disappear after a bad run. Here is what makes ERE different — and why we can prove it.

Others
Cherry-picked results, only winners posted
ERE
Every selection published before the race. Every result logged automatically.
Others
Gut feel and form opinions passed off as analysis
ERE
Dual-model ML system drawing on 67,000+ runner entries of history. Accuracy tested on data it has never seen.
Others
20 tips a day to manufacture a winner somewhere
ERE
Only selections with genuine positive edge over the market are flagged. Quality over volume.
Others
Static model that never changes, or no model at all
ERE
Retrains every night on the latest results. Adapts automatically to form cycles and seasonal changes.
Process

HOW THE SCANNER WORKS

Six automated steps run every morning — pulling live data, scoring every runner, and delivering only the selections where the model sees genuine edge over the market.

01
Live Racecards
Each morning, the scanner pulls fresh racecards and live market odds directly from The Racing API — every runner, every race, every course in the UK.
02
ML Scoring
A machine learning model assigns a win probability to every runner, trained on 12 months of race results and retrained automatically each night.
03
Value Engine
Model probability is compared against market-implied probability. Only runners where the model sees genuine positive edge are flagged — no guesswork.
04
Selection Tiers
Selections are ranked into Best Bets and Value picks based on edge strength and model confidence. Outsiders are tracked privately, not released.
05
Telegram & App Delivery
Selections are pushed to your Telegram and the members app at 12pm and 4pm — before racing begins, with full model stats on every pick.
06
Nightly Retraining
Results are backfilled each evening and the model retrains on the updated dataset — continuously adapting with every race run.
Example output

WHAT A SELECTION LOOKS LIKE

Every selection comes with full transparency — model probability, market odds, and edge percentage. You see exactly why the model flagged it, not just the name.

  • —Model win probability vs market implied
  • —Expected value and Kelly fraction
  • —Flat [F] or Jumps [J] model tag
  • —Edge strength visualised as a bar
⭐ BEST BET
Burning Ember
Kempton Park · 16:10 · Handicap Chase
[J] JUMPS
Market Odds
4.50
Model Prob.
31.2%
Fair Odds
3.21
Edge
+9.0%
Exp. Value
+0.40
Kelly %
11.6%
EDGE STRENGTH · 68th PERCENTILE TODAY
Model performance

THE NUMBERS ARE WHAT THEY ARE.

We don't cherry-pick the good days. Every selection is tracked live and published regardless of outcome. This is the full picture.

Model Top-1 Race Accuracy Walk-Fwd CV Mean Training Set Notes
Jumps Model [J] 67.1% 53.6% 1,150 National Hunt, Hurdles, Chases
Flat Model [F] 48.6% 43.3% 26,795 Flat & All-Weather races
Combined Model 58.8% 48.9% 27,945 All UK flat & jump races
Random Baseline ~14% ~14% — 1 ÷ average field size
Live tracked
All P&L logged automatically before races run
Walk-forward
Tested on data the model has never seen
🔄
Nightly updates
Model retrains automatically every evening
Value only
Only positive-edge selections are flagged
Membership

JOIN BEFORE IT GOES PAID.

Early access is open now. Members who join during this period will be notified first when paid plans launch and will receive preferential terms. There is no catch.

Early Access
FREE
free during early access · paid plans coming
  • Daily Best Bet & Value selections
  • Telegram alerts before racing begins
  • App notifications at 12pm and 4pm
  • Full model stats on every pick
  • Live P&L tracking dashboard
JOIN DURING EARLY ACCESS
Why trust us

METHODOLOGY & HONESTY

Most tipsters hide their process. We publish ours. This is what separates data-driven from opinion-driven.

No backdated results
All P&L is tracked live from selections posted before racing. No retroactive picking of winners. Results are logged automatically and published unedited.
Walk-forward validation
Model accuracy is tested on data the model has never seen before — the same rigorous approach used in professional quantitative analysis.
Daily retraining
The model retrains every night with the latest results. It adapts to seasonal changes, track biases, and shifting market conditions automatically.
Value-based, not volume-based
We don't send 20 selections a day to manufacture a winner. Only runners with demonstrable positive edge are flagged — quality over quantity, always. Some races — including the Grand National — are excluded on methodology grounds: field size and market conditions make genuine edge detection unreliable.
Get started

THE SCANNER
DOESN'T GUESS.

Daily ML-powered selections, delivered before the market moves. Every result tracked publicly from the moment it's posted. Early access is free — join now.

JOIN DURING EARLY ACCESS
For educational and entertainment purposes. Gambling involves risk — never bet more than you can afford to lose. Elite Racing Edge does not guarantee profit. 18+ only. Please gamble responsibly. BeGambleAware.org