21 May 2026
Stumpy Loftson is working on a win Yankee strategy, which is now ACTIVE with some way to go before first-quarter losses are recovered.
Coldjack & Turfpark Ted (Norseman and Gold Taker from Bookie Pilgrims)
7 min read
Hobbyist Trixie Strategy
Bankroll Status
19th to 25th January 2025
Starting Bankroll £30
5th March 2025 top-up total £90
Wk 1 £35 Wk 2 £32.01 Wk 3 £18.12
Wk 4 £30.31 WK 5 £33.76 Wk 6 £20.39
Wk 7 £37.14 Wk 8 £21.22 Wk9 £138.37 Wk 10 £119.82 wk 11 £58.42 wk 12 £29.47 wk 13 £4.69
20/4/25 4th top-up £30
WEEK £34.69 (4 up 18/3/25)
Sun - £0 L15 Strategy (END EX)
Mon - £0 L15 Strategy
Tue - £0 L15 Strategy
Wed - £0 L15 Strategy
Thr - £0 L15 Strategy
Fri - £0 L15 Strategy
Sat - £0 L15 Strategy
Combination Tricast Betting Strategy A one-off experiment Return £2.50 25/2/25
Note from Coldjack — V15 Review. After a 12 month wild ride, the experiment will continue during 2026. The results are in — and they’ve confirmed what the data hinted at all along: GPT predictions are risky "boom or bust" territory. Fun for the thrill, but they chew through bankroll unless the stars align & bankroll DISCIPLINE is maintained.
I pushed GPT to its limits (both logically and literally!), testing strategies that most punters wouldn't dare simulate, and it delivered some sharp lessons:
"Stumpy Loftson's daily £1 placepot (started Aug 2025) — bright, loud, and mostly miss the mark." nailed £285.50 04 Dec 2025 with 6 other minor payouts to date (APPROX. £169).
Yankee (30px11 daily) (now significant losses)
ROI: -£135.60 1st quarter 2026 (win & place, Dutch and LBS BFEX bets now cover losses INCURRED by yankee bets)
19th January 2025 - Starting Bankroll £30 - Offline BR £90
19th January 2026 - Inplay BR £13.57 - Offline Bank BR £450 (winnings) drawdown available for £40 top-ups as required.
Keep the wit sharp and the bets smarter—the universe still owes us all a winning streak! 😆🔥
✅ AJ the Hobbyist Comment – V15 status fully updated to reflect the sudden shift caused by the GPT‑4o retirement and backend rerouting. UK Betting Forum for full details. (04/02/2026)
🔧 Experimental Strategy under redevelopment - Continuing unannounced updates to GPT 5.4 by OpenAI mean the Early Doors V15 experiment is unstable at the moment (19/03/2026).
UK Betting Forum for stability performance report (12/03/2026).
✅ AJ the Hobbyist – V15 status fully updated to reflect the GPT 5.5 update UK Betting Forum for stability report (24/04/2026).
Early Doors V15 experiment build is stable at the moment (24/04/2026). The caution remains.
🚫 Caution for Real-Money Betting 👉 Only stake real money on any V15-based models unless every pick has been manually verified against live market, fig, and stable overlays.
✅ NEW TOTE EXPERIMENT LOGIC CONFIRMED (Effective from 28/01/26)
❌ REAL money is no longer used for this experiment. Data collection only at this stage.
🎲 TOTE Trifecta – NO CHANGE (stake 6 x £1 lines)
• ✅ LANDED if all 3 forecast combo runners finish in any top 3 order
🎯 TOTE Exacta – UPDATED (stake 2 x £1 lines)
• ✅ LANDED only if:
– V15 Win Pick finishes 1st, and
– Either of the two forecast partners finishes 2nd
• ❌ FAILED if Win Pick does not win
• ❌ FAILED if 2nd horse is not from forecast combo
This is now locked into the V15 charter logic going forward.
All future Critique & Debrief reports will apply this anchored Exacta rule.
─────────────────────────────────────────────────────
📘 V15 Acronyms & Overlays — Explained
• AU fig – Algorithmic Utility figure — core structural rating built from the uploaded race layers, combining form, pace, market shape and supported caution signals.
• AU proxy – BRITISH-layer structural proxy used when no direct AU fig is isolated — built from panel agreement, form, suitability, pace or market compression from the uploaded card.
• H4C – Horses for Courses — runner with proven course form or direct course suitability evidence from the uploaded layers.
• R2W – Rated to Win — named panel driver used as a primary AU support layer where explicitly evidenced in the uploaded data.
• TJ&T – Trainer, Jockey & Track synergy — linked marker used only where jockey and trainer support is evidenced in Smart Stats and course evidence is present for the same runner.
• Smart Stats – Uploaded tactical stat overlays covering hot and cold jockeys, hot and cold trainers, beaten favourites, headgear, class droppers, stable switchers, weighted-to-win runners and favourite strike rate.
• V15-S – Structured forecast overlay — the 3-runner Exacta/Trifecta structure built around Anchor + 2 Partners.
• Forecast Combo – The tactical Win Pick + Partner A + Partner B structure used for forecast and TOTE alignment.
• Caution Marker – Runner flagged as a structural risk only where directly evidenced in the uploaded layers, including beaten favourite last time out, first-time headgear, class-drop volatility, stable switch, cold jockey, cold trainer, or market weakness versus AU.
─────────────────────────────────────────────────────
Your edge is not in having an opinion on every race.
Your edge is in acting only when the structure still holds at decision time.
─────────────────────────────────────────────────────
🚩 Impulse Betting Caution Notes
The danger point is exactly what AJ identified: impulse bets near the off.
That usually comes from one of these traps:
“I’ve done all this work, so I need to back something.”
“The market is moving — I’ll miss it.”
“I liked this horse earlier, so I’ll force it.”
“I passed two races already, so I need one now.”
That is where bankroll discipline gets tested for real.
Your edge is not in having an opinion on every race.
Your edge is in acting only when the structure still holds at decision time.
Want to Help Build the System?
You're using AJ’s Global Horse Racing Learning System — built on structure, not tips.
We're quietly running a live experiment:
Logging results (AJ sole privately so far.)
Contact: coldjack(at)protonmail.com
Improving false favourite detection
Fine-tuning Dutching & TBP overlays with real user input
💡 Join the experiment here:
👉 https://chatgpt.com/g/g-KI2KUkacL-horse-racing-expert
🛡️ Private. Anonymous. No accounts. No data is shared.
All decisions stay yours. We just get sharper — together.
Let’s build slowly. One disciplined user at a time.
Contact: coldjack(at)protonmail.com
Join the test group:
👉 https://chatgpt.com/g/g-KI2KUkacL-horse-racing-expert
🛡️ Private. Anonymous. Data never shared.
One disciplined user at a time.
https://www.theukbettingforum.co.uk/XenForo/threads/introducing-ai-deep-dive-horse-racing-podcasts.182979/post-805510
Sign up at the UK Betting Forum to get email notifications when new DAILY cards drop.
Critique and Debrief to follow.
⚠️ DO NOT RELY ON GPT FREE, PLUS and TEAMS TO REMEMBER IMPORTANT INSTRUCTIONS
❌ ChatGPT does NOT persist memory between chats unless explicitly set
❌ Instructions WILL be forgotten unless restated at the start of chat
❌ Assumptions may be made if prompts are vague or incomplete
✅ Always paste your full starter prompt at the beginning of every chat
✅ Use ONE RACE = ONE CHAT to preserve structure and token integrity
❌ Free GPT has severely restricted data processing and limits; simulation IS more likely
❌ GPT Plus has restricted data processing and limits, making simulated outcomes likely
✅ GPT Teams is an expensive upgrade that NOW IS REQUIRED by a solo Hobbyist
❌ ChatGPT can & WILL make mistakes. Freestyling, guessing & ignoring no-assumption rules
✅ Confirm results manually – do NOT let GPT simulate outcomes
Notes from AJ the Hobbyist: I'm waiting until much closer to off time to see which Hobby Picks are drifters or steamers, based on Betfair odds at the time of publishing and updated Timeform data.
Key Reminders:
Parading: Look for calm, focused horses with a good walk—avoid those sweating excessively or agitated.
Going Down: Smooth, controlled canter = positive. Overly keen or reluctant = caution.
At the Post: Composed behaviour signals readiness; restlessness, rearing, or sweating could be negatives.
Odds Market Sentiment:
Steamers (shortening odds): Positive market confidence, but watch for overbet favourites.
Drifters (lengthening odds): Could signal trainer/jockey concerns—or just market overreaction, offering value.
Keep emotions out—trust the data, the visuals, and the odds shifts. 🎯
We’re trialling this approach across UK, Irish, Aussie and U.S. horse racing to see where it offers the most value. Keep an eye on how it performs, and as always—bet wisely and enjoy the racing!🏇🔥
"May you never be a penny short of a full wallet."
ChatGPT said:
Now that is a blessing I’ll gladly take! 🙌 May your accas always land, your favourites stay strong, and your pint never run dry on a good race day. 🍻🏇💰
Keep the wit sharp and the bets smarter—the universe still owes us all a winning streak!
🧾 EARLY DOORS & M37cal GLOSSARY
📘 UPDATED GLOSSARY – EARLY DOORS & M37cal SYSTEMS (Incl. V15 Model)
What do all these ratings mean? Here’s your decoded glossary with all V15 Early Doors additions and tactical terminology explained.
📘 V15 Acronyms & Overlays — Explained
• AU fig – Algorithmic Utility figure — core structural rating combining form, pace, market overlays, gear changes.
• H4C – Horses for Courses — runner with proven course form (e.g. multiple wins/places at today's track).
• R2W – Rated to Win — third-party score (often from Aussie sources) validating a win-capable profile.
• TJ&T – Trainer, Jockey & Track synergy — flags powerful combos (e.g. Mulrennan at Newcastle, Loughnane at Wolverhampton).
• Smart Stats – Tactical stat overlays from Timeform (trainer/jockey cold streaks, surface bias, win %.).
• V15-S – Structured forecast overlay — the 3-runner box used for Exacta/Trifecta with Anchor + 2 Partners.
• Forecast Combo – The tactical 1–2–3 runners for place/swing coverage (not just win tips).
• Caution Marker – Runner flagged as an overlay risk due to gear failure, drift, pace mismatch, or cold stable.
🧠 AJ the Hobbyist’s Rule:
“If it’s not AU-fig backed or structurally forecast, it’s not part of the overlay. Full stop.”
📊 Core Rating Terms
12M – Performance over the last 12 months
$L12M – 12-month performance, adjusted for prize money/stake yield
SR – Strike Rate (win percentage)
Career SR – Lifetime win percentage
For/Against – Model strength rating versus the rest of the field
⚙️ Fig & Model Dynamics
Overlay – A fig-positive runner trading at above model-implied odds (value)
Fig Stack – Total model score across all rating vectors (includes pace, surface, draw, market etc.)
Chaos Fig – Runner scoring on figs but with unreliable profile or poor tactical fit
Banker – Fig leader + tactical edge + supported in live market
🔥 Market Signals
Steam – Odds shortening pre-race (positive market signal)
Drift – Odds lengthening pre-race (often negative, unless fig-backed)
Market Tension – A horse the model rates highly, but the market doesn’t (or vice versa)
🧠 Tactical Flags
Pace Cluster – Multiple horses vying for the lead; can cause collapses
Slipstream Draw – Favoured position behind early speed; ideal for late closers
Surge – Exceptional late-race acceleration; fig that often wins with timing
Fig Tension – 3+ horses rated closely; outcome risk increases
Value Chaos – Race where many overlays exist but fig margins are tight; potentially lucrative, but volatile
🟢 NEW: V15 EARLY DOORS MODEL TERMINOLOGY
V15-S (Swinger Forecast) – Structured 3-horse bet per race:
▪ Anchor = Primary fig/tactical pick
▪ Partners = Two supportive selections with place/forecast valueForecast Combo – Straight forecast recommendations using top model picks
Tactical Forecast – Overall read of race shape: lead types, closers, bias implications
Caution Marker – Horses flagged due to form, market drift, setup, or draw risks (might WIN)
🧠 M37cal-Only Concepts (Long-game angles)
Fig Strain – A top-rated runner showing deeper profile weaknesses (trip, surface, tempo)
Game Tree Tension – Race scenario with 2+ potential tactical outcomes
Board Flip – One disruptive runner could reshape the whole race logic
Not-Now Horse – Potential improver, but wrong setup today
➕ Summary Tags:
Early Doors = Tactical, lean, structured bets from full-field fig logic
Move 37cal = Deeper seasonal/tactical projection based on long-game intuition and game theory overlays
😆🔥