Hawthorn Multi System · The Gate · Calibration Ledger · Est. R12 2026
Rule #1: the gate's job is to kill bets, not approve them. Some weeks the right output is "no bet".
Rule #1: the gate's job is to kill bets, not approve them. Some weeks the right output is "no bet".
Season betting record
Net cash P&L
Weekly cap
Fixed unit. Never raised after a loss. Winnings withdrawn.
🎯 Next game
🚑 Unavailable / watch
Bet history
🚪 The Gate — run every leg through, or it doesn't exist
Blocks: Player (per leg) → Game (once) → Price & Portfolio (once, partly automatic). Untickable = untrue = the leg dies.
Game checks (once per week)
Price & portfolio checks
Add legs and tick the checks — verdict appears here
📈 Calibration — do my claimed numbers survive contact with reality?
⚠ Standing note from R16: multiple "84–92%" anchors failed in one game (correlated team-level failure). Until the ledger proves otherwise, mentally discount anchor claims by ~5–10%.
Leg log
| Rd | Player | Market | Line | Odds | Claimed | Result | Notes |
|---|
Add leg manually
Bet log
| Round | Stake | Odds | Result | Cash +/− | Notes |
|---|
Add bet
🏉 Player Vault — last-5 current through R16 (auto-pulled from Wheelo Ratings)
Edit any dashed cell, then hit save. Floor = worst of last 5 → anchor lines must sit AT or BELOW floor. Med = median.
| Player | Stat | Last 5 (old→new) | Floor | Med | Anchor line | Status | Notes |
|---|
💡 Insight Bank
📋 Weekly data drop — what to send Claude
- ☑ Hawks + opponent stats — Claude self-serves from wheeloratings.com data feed now (incl. CBA%!) — just confirm the round
- ☐ Both injury lists — afl.com.au/matches/injury-list
- ☐ Confirmed lineups (Thu ~6:20pm) — match page → Line-Ups
- ☐ CBA% / TOG screenshot — dfsaustralia.com player tables (NEW — catches role traps)
- ☐ Sportsbet SGM odds — Disposals + Goals tabs minimum
- ☐ Late outs — recheck lineup within 2h of bounce
- ☐ Match reports / presser — optional, only if something unusual happened
🔁 Working with Claude across chats
1. Start of each session: click 📤 Export state (top right) → paste into the chat. Claude gets the full ledger, floors and insights back instantly.
2. After the game: enter leg results + bet outcome here, then export again next week.
3. Gate verdict gets stamped before the Sportsbet app is opened. If it says KILLED — that's the system winning, not losing.
4. Promo-or-pass. No stake-back / boost attached → default is no bet.
5. Deposit limit inside Sportsbet = season budget. Structural, not willpower.
2. After the game: enter leg results + bet outcome here, then export again next week.
3. Gate verdict gets stamped before the Sportsbet app is opened. If it says KILLED — that's the system winning, not losing.
4. Promo-or-pass. No stake-back / boost attached → default is no bet.
5. Deposit limit inside Sportsbet = season budget. Structural, not willpower.
🐎 Spring Racing · 29 Aug - 29 Nov 2026 · Hawthorn Rules, Carnival Edition
SEASON OPENS 29 AUG
Same objective as the footy system: have fun AND genuinely try to win. What the 5-spring backtest (2021-2025, 14,685 races) proved is that no autopilot formula wins here, so the shot at a winning season comes from the same place it does with the Hawks: picking a small number of spots you actually have a view on, pricing them yourself before looking hard at the market, and letting the gate kill the weak ones. Quality over quantity: 15-25 gated bets across the whole carnival beats 5 ungated bets every Saturday. One season budget set before Memsie Day, never topped up: that's the worst case, and finishing ahead of it is the win condition. Every bet logs a claimed win chance, and the calibration line grades your judgement at season's end, exactly like the footy leg ledger.
Season record
Settled fun-ledger bets; pending in brackets. Each-way = two entries.
Season P&L
Betfair cash only. Finishing ahead of $0 is the win condition.
Season budget
$5 e/w standard · $10 A-grade (max 1 a meeting) · target odds $6-$12
📅 Next meeting
🎟️ Fun Ledger · every bet, reason first
| Date | Track | Selection | My % | Odds | Stake | Result | P&L | Why |
|---|
🚪 The Racing Gate · every box ticks or the bet dies
G1 · A real view, not a vibe. You can say WHY this horse in one sentence grounded in something specific: form line, map/tempo, track or weather bias, class drop, jockey/trainer intent. "Nice name" and "everyone's on it" die here.
G2 · Your price first. Write your claimed win % BEFORE settling on the market price. Only back it if the market's odds beat your number with room to spare. If the market agrees with you, there's no bet: that's the SGM value-leg rule in racing clothes.
G3 · Singles on the exchange, best odds. Betfair or verifiably better fixed odds. Allowed shapes: win, place, each-way (log as a win + place pair), and dutching 2 runners in one race (log as two singles, same reason). No racing multis, ever: legs are independent, so a multi only compounds margin. The Backtest Lab is the proof.
G4 · Season budget is structural: $50, locked 31 Jul 2026. Mirrored as the Betfair deposit limit, never topped up. Stake rule, also locked: $5 each-way standard ($2.50 win + $2.50 place), one $10 A-grade allowed per meeting, target odds $6-$12. That's 8-12 quality bets for the season. The signature shape is the racing SGM: place half collects often, win half is the big one. Footy money and racing money never mix.
G5 · No chasing, no Cup-day impulses. A losing Saturday changes nothing about Sunday. Cup-day extras beyond the plan come from a small pre-set sweep budget, and the gate applies from bet zero.
G6 · Day rules: stop-win and stop-loss. Once the day is in front, it's banked: no more bets that meeting. Down 3 stakes: day over. Simulated on 1,921 real meetings this turns "leave positive 44% of days" into 75%, and caps the worst day at 3 stakes. It doesn't create edge, it shapes the season so the fun survives.
📊 Why no autopilot: every blind formula, 5 springs, commission included
| Strategy | Bets | Strike | ROI (5yr) | Years +ve | Verdict |
|---|
🎯 Bet-type menu · every way to play the carnival, graded honestly
| Bet type | Our data says | Verdict |
|---|---|---|
| Win single (gated) | Blind favs -4.9%; selectivity is the whole game | THE HUNT |
| Place single | Fav-to-place: 68% strike, -2.1% (best blind result; -0.6% in 2025) | BREAD & BUTTER |
| Each-way | Win + place averaged: ~-3.5% blind, high collect rate | OK, GATED |
| Dutch 2 runners | Two market-priced singles; strike ~doubles, EV unchanged | OK WHEN SPLIT VIEW |
| Lay favourite | -2.0% big fields, but win-small/lose-big shape | SKIP: UGLY SHAPE |
| Quinella / exacta / trifecta / first 4 | Tote takeout 14-25%, worst-priced products in racing | CUP-DAY FUN $ ONLY |
| Quaddie | ~20%+ takeout; a $20 flexi is a day-long sweat, priced accordingly | FUN $ ONLY |
| Racing multis | Independent legs; margin compounds per leg | NEVER |
| Futures / all-in | Fat margins + you lose the stake on a scratching | NEVER |
| Promo-attached bets | Stake-back / bonus-back = the only structurally +EV money | ALWAYS PREFER |
🗓️ Season plan · 29 Aug - 29 Nov · quality over quantity
Before Sat 29 Aug (Memsie Day): season budget locked in above and in the Betfair app's deposit limit. Structural, not willpower.
Sep: footy finals are still the main game. Racing = the big Saturdays only, 1-3 gated bets max. Early reads on track bias and in-form stables become Insight Bank entries.
Oct: the deep end: Turnbull, Caulfield Guineas, Caulfield Cup, Cox Plate. This is where doing the form on a target race midweek (like Thursday-night footy prep) pays.
Sat 31 Oct: Derby Day, Flemington.
Tue 3 Nov: Melbourne Cup: sweep + one fun bet from the pre-set Cup budget. The gated bets hunt the undercard, where form holds up.
Sun 29 Nov: season closes. Calibration review: did the claimed win %s match reality? That answer decides what spring 2027 looks like.
Sep: footy finals are still the main game. Racing = the big Saturdays only, 1-3 gated bets max. Early reads on track bias and in-form stables become Insight Bank entries.
Oct: the deep end: Turnbull, Caulfield Guineas, Caulfield Cup, Cox Plate. This is where doing the form on a target race midweek (like Thursday-night footy prep) pays.
Sat 31 Oct: Derby Day, Flemington.
Tue 3 Nov: Melbourne Cup: sweep + one fun bet from the pre-set Cup budget. The gated bets hunt the undercard, where form holds up.
Sun 29 Nov: season closes. Calibration review: did the claimed win %s match reality? That answer decides what spring 2027 looks like.
📋 Race-day prep sheet · what Claude pulls before you price a race
- ☑ Fields, weights, gear + past runs (incl. going of each past start) — Claude self-serves from Racing Australia FreeFields
- ☑ Sectionals + replays context — racing.com, self-served (all carnival tracks are VIC)
- ☑ Weather + radar for the track — BOM station feed, self-served (footy weather trick, racing edition)
- ☑ Official track rating + upgrades/downgrades — Racing Australia / Racing Victoria on the morning; a mid-morning downgrade is a re-price moment
- ☐ Day bias read for late races — what's winning races 1-3: on-pace or swoopers, inside or out. Claimed % gets set AFTER this, not before lunch
- ☐ Speed-map / wet-track tables screenshot — punters.com.au or racenet are policy-blocked for Claude; you open them, paste a screenshot (the CBA% move)
- ☐ The price — read Betfair last, after your own % is written down. Gate G2 lives or dies on this order
- ☐ Log what the day taught you — bias reads, stable patterns, track quirks go in the Hawthorn Insight Bank prefixed RACING: so they're all there by Cup week
Theory mid-carnival? The 14,685-race BSP dataset (springs 2021-2025) is on disk: any "does X actually work" question gets a real backtest answer the same day, before it earns a dollar.