Read the board
The engine starts with observable market data: spread, total, moneyline, book count, price freshness, and line movement.
Published framework
Deterministic scoring. Open method. Protected weights.
Galaxy Sports Edge publishes the factors and decision philosophy behind the model. The exact weights, constants, and aggregation formula remain proprietary.
Score the math
A deterministic factor model scores the market and matchup context. The framework is public; weights, constants, and aggregation formula stay proprietary.
Gate the slate
Publish thresholds, freshness checks, and data-quality gates decide whether a pick reaches the board. Thin slates can produce zero picks.
Factor inventory
What the model can read
Factors are visible. Exact scoring weights are not published.
Market consensus
Book depth
Line movement
Volatility
Head-to-head context
Venue form
Schedule stress
Rest differential
Cross-market agreement
Data quality
Methodology / Trust Surface
The audit trail behind every signal
Every published pick ties back to live markets, timestamped data, factor scoring, and the gates that keep weak picks off the board.
data
Live odds ingestion
We ingest live odds from multiple sportsbooks on a regular schedule and score every available matchup.
data
Bookmaker coverage as a transparency signal
Each pick is scored against the bookmakers that had a market for the game at the time of scoring. We surface the bookmaker count as a transparency signal.
data
Data freshness on every pick
Where available, each pick shows the timestamp of the odds data it was scored against so you can judge freshness for yourself.
model
Calibrated confidence presentation
Confidence is expressed as a label or score depending on the platform's current confidence-display mode. Numeric scores are only shown once calibrated against settled outcomes.
Until we have enough settled outcomes to calibrate against, confidence is shown as a label, not a number.
model
Risk level on every pick
Each pick carries a risk level reflecting bookmaker consensus, market depth, and known volatility factors.
model
Factor breakdown for subscribers
Subscribers with the right entitlement can see a factor-by-factor breakdown of how each pick was scored.
gate
Public performance is gated, not advertised
Public performance statistics are only displayed after the platform has accumulated enough settled, canonical picks to compute them honestly.
When you see win-loss numbers on the Performance page, you'll also see the period, sample size, model version, and the exact win-rate definition.
Line freshness
A pick is only as honest as the line behind it.
Every pick is priced against real bookmaker quotes, and every quote carries its own timestamp. Before anything reaches the board, a freshness gate checks the age of each game's lines. A game whose lines are too old gets dropped, no matter how good the math looks: an edge computed on a dead line is not an edge, it's a mirage.
Checked per game
Freshness is decided game by game, not for the feed as a whole, so one fresh game can never smuggle a stale one onto the board. Every game's lines must clear a hard maximum age before its pick can publish, and that ceiling can tighten as kickoff nears, when lineups, injuries, and late market moves change prices fastest.
Shown, not claimed
The board shows the age of the freshest line behind today's picks, computed from the bookmaker's own timestamp. When the data is older, the badge says so in plain sight. No timestamp, no badge: the site never invents an "updated just now."
A quiet board is a position
When lines are stale or coverage is thin, the system publishes nothing rather than something it can't stand behind. Zero picks on a given day means the gate held, and that discipline is the product.
Reading the market
What the market thinks, and whether we beat it.
Three reads, all built from odds we actually capture, never a projection dressed up as a market.
The market's real price
Every book's latest two-sided quote, de-vigged and taken to a median across books: the market's opinion with the margin stripped out. What the books actually think, not what they charge.
See the fair board →The best of it
The line shop finds the best available price for each side across the books we capture: the same line-shopping edge the pros use, shown as plain transparency. Shop the number; it's your money.
Open the line shop →Did we beat the close?
Closing line value grades whether the price we locked beat where the market closed: the leading indicator of a real edge that sharp bettors trust, and the one number tout services never show. Published under the same gate as the win rate.
CLV methodology (gated until sample) →Ranking law
Ranking probability, confidence, and edge are not the same thing.
Calibration and board ranking use a win probability — never a signed edge gap dressed up as a probability. While live floors are still red, we do not claim PROVEN performance or ROI.
Confidence
A 0–100 composite of market and matchup factors. Useful for UX and gates. Alone it often echoes the books, so discrimination can stay near noise until independents price in.
Ranking probability
The probability used for sort, selective publish, and calibration bake-offs. When independent estimators yield a finite true probability, ranking uses that (or a blend with confidence) — including when the edge claim is PASS. SPEAK/LEAN is the glass-box edge claim, not the ranking gate. Without independents, ranking falls back to confidence on the 0–1 scale.
Edge is not a probability
Edge is the gap between an independent true probability and the market's fair price — a signed difference, not P(side wins). Edge can filter and explain; it never enters Brier, resolution, or separation as if it were a win probability.
Version changelog
Model changes are named.
v5.2.2Dixon–Coles soccer independent + Kalshi ESPN/ET match polarity; ranking math from v5.2.1 unchanged.
v5.2.1Ranking uses independent model P whenever finite (incl. PASS); bake-off never treats edge as a win probability; positive separation required.
v5.2Independent estimators priced into ranking path; edge is never used as a win probability.
v5.0Bootstrap-canonical gating and settled-only learning policy.
v4.0Expanded factor snapshot storage for public audit trails.
v3.0Added schedule stress, rest, and cross-market checks.
See the framework on today's board.
The public board shows every published pick and every free Edge Index. Pro unlocks the detailed factor breakdown.
Sports wagering is real risk. Only stake what you can afford to lose without changing your week. If you or someone you know has a gambling problem, call 1-800-GAMBLER.