Reference
Every TBS-specific term, defined plainly. Anchor links at every entry so you can paste a definition into a Slack thread or email without hunting.
Jump to
Our highest-confidence pick of the card. To be a Best Bet, a runner must clear all of the published gating rules — confidence above the model floor, margin above the next runner, price within the $3-$12 band, and edge above the calibrated cutoff. There is no daily cap on Best Bets; if multiple runners gate through, multiple ship. If none do, none ship.
A pick the model rates favourably but which doesn’t clear the Best Bet gates. Value tips ship as flat-stake WIN bets at smaller unit sizes. Lower confidence, wider price band, used to spread risk across the card.
The price below which the bet is pulled. Published on every tip card before the race jumps. If the market shortens past the walk-away before the off, the bet is marked walked, P&L zero, and excluded from ROI math. The rule prevents us from chasing prices into bad value when the market reads the same edge we did.
A two-part bet: half stake on the win, half on the place. We use each-way only at longer prices where the place leg has positive expected value on its own. Settles as two independent legs against the official stewards’ result.
A bet that didn’t run — most commonly because the horse was scratched. Voids don’t count for or against ROI; stake returned, no P&L impact.
Distinct from voided. A walked bet is one we pulled before the jump because the market shortened below the published walk-away price. Logged so the record is honest, excluded from ROI so the math reflects only actual stakes.
The model’s estimate of what a runner should be priced at if the market were efficient. Computed as 1 / model probability. If our fair price is $5.00 and the market price is $6.50, there’s an apparent edge — see edge.
The percentage gap between market price and fair price. (Market / Fair − 1) × 100. Positive means the market is offering longer odds than the model thinks the runner warrants. The edge column is hidden on v2 picks today because the engine’s probability output is not yet calibrated.
Punting Form’s proprietary AI rating. Not used in the TBS rating formula. An earlier version of the engine included PFAI as a 10% additive overlay; that was removed 2026-05-12 in favour of a fully scrape-first composite — all five signals are computed by us against form history we own end-to-end. We still pull PF results for settlement cross-checks, but no black-box vendor rating sits in the picks.
A 0-100 score the model assigns to its top pick in each race, derived from the margin between rank 1 and rank 2 plus the calibration band. A Best Bet requires confidence above the published floor (currently 70).
The model’s spread between its top-rated runner and the next runner, in composite-rating points. A Best Bet requires the top runner to clear the next by at least 3 points. Low-margin races (where two or more runners are close) are not tipped on — better to skip than to coin-flip.
The price we tipped at minus the price the market closed at. Positive CLV means we beat the closing line — which, over a long enough sample, is the only metric that actually proves a tipping service has edge. Surfaced on the dashboard and tracked internally; not yet shown publicly.
A tip whose race has run, the stewards’ result has landed, and P&L has been recorded. Settled tips appear on /results and feed into the running ROI numbers.
A cryptographic proof that a specific document (in our case, the monthly tip CSV) existed at a specific moment in time. Each month we hash the canonical tip set with SHA-256, submit the hash to the OpenTimestamps calendar servers, and store the resulting .ots proof file. Once a Bitcoin block confirms, the proof upgrades to a verifiable anchor on the chain — meaning we can’t quietly edit the record after the fact. See /audit.
The current operating mode. Tips are published before each race, prices and stakes are logged at publish time, and P&L is computed against stewards’ results — but TBS doesn’t place any bets on its own. Subscribers place their own bets through licensed Australian operators. The paper-trade ledger is the public record.
The standard stake measure across every TBS tip. We size in units rather than dollars so the staking rules work for any bankroll. A common convention is $10 = 1u for display purposes, but you set your own $/unit based on your bankroll (typically 0.5-2% of total bank).
Best Bets are sized by market price: ≤ $3.50 gets 2u WIN, $3.50-$8 gets 1u WIN, $8-$12 gets 1u + 1u each-way. Shorter prices get bigger stakes because the edge is more reliable at the front of the market. Cap is 2 units per bet, always.