Reference

Betting maths, defined

Every number Stratos shows comes from one of these formulas. Prices and market chances are facts about what a sportsbook offered. Model estimates, edges and expected value are our own calculations — useful, but never a promise about a result.

Search 38 terms

Bet types

Money line: A bet on who wins the event outright, with no handicap applied.Win if side score > opponent score; a draw on a two-way market pushes.
Point spread: A handicap added to one side's score so the two sides are priced closer to even.Adjusted margin = side score - opponent score + handicap. Above 0 wins, 0 pushes.e.g. Team A -3.5 wins 27-20: 27 - 20 - 3.5 = +3.5, a win.
Total (over / under): A bet on the combined score of both sides against a posted number.Over wins if home + away > line. Equal to the line pushes.
Push: A tie against the number. The stake is returned. Only whole-number lines can push.
Player prop: A bet on one player's own statistic — points, yards, strikeouts — against a line.Over wins if the official box-score figure > line.
Alternate line: The same market at a different number, priced accordingly.
Same-game market: Several selections from one event combined. Because the legs share a game script they are not independent, so the true chance is not simply the product of the legs.
Futures / outright: A bet settled at the end of a competition. There is no clean two-way consensus behind these prices, so the fair estimate is weaker.
Parlay: Several selections that must all land. The prices multiply, and so does the risk.Combined decimal price = leg 1 x leg 2 x ... x leg n.

Prices & probability

Decimal odds: Your total return per 1 staked, stake included.e.g. 2.50 returns 2.50 for every 1 risked — 1.50 profit.
American odds: Profit on a 100 stake (positive) or the stake needed to win 100 (negative).+150 = 2.50 decimal. -150 = 1.667 decimal.
Implied probability: How often a bet must land for the price to break even. It still contains the book's margin, so a market's implied chances sum to more than 100%.Implied probability = 1 / decimal oddse.g. -150 implies 150 / (150 + 100) = 60%.
Overround (vig, margin): How much more than 100% a market's implied chances add up to. That excess is the book's built-in edge.Overround = sum of implied probabilities - 1e.g. 55% + 50% = 105%, a 5% margin.
No-vig probability: The market's own view once the margin is scaled out. Still market data — it is what the price says, not what we estimate.No-vig probability = implied probability / sum of all implied probabilities
Model probability: Stratos's own estimate of the chance. It is stored alongside the market number and never replaces it.
Probability edge: The gap between the model estimate and the no-vig market chance, in percentage points.Edge = model probability - no-vig market probabilitye.g. 61% - 54% = +7 points.
Fair odds: The price that would exactly match the model estimate, with no margin either way.Fair decimal = 1 / model probability
Expected value: The theoretical average return per unit staked if the model estimate is right. It is not a promise of profit.EV = probability of winning x decimal price - 1e.g. 0.55 x 2.00 - 1 = +0.10 per 1 staked.
Closing line: The last price before the event starts. Beating it consistently is the usual test of whether picks were priced well.
CLV (closing line value): The difference between the price you took and the closing price.
Kelly stake: The bankroll fraction that maximises long-run growth at a given edge. Zero when there is no edge.Kelly = EV / (decimal price - 1)

Market movement

Line movement: A change in the number itself — a handicap or total moving.Line movement = current line - opening linee.g. -3 to -4 is a one-point move toward the favourite.
Odds movement: A change in the price at the same number. Tracked separately from line movement; the two are never merged into one figure.e.g. -110 to -125 at the same -3.
Movement velocity: How fast a market is moving.Velocity = line change / hours elapsed
Market volatility: How much a price has been jumping about, measured as the standard deviation of its changes.Volatility = standard deviation of successive price changes
Market regime: A label for current conditions — normal, trending, volatile, news-driven, thin, live. It changes how much weight the estimate deserves.

Statistics

Mean: The average of a series.
Standard deviation: How spread out a series is around its average.
Z-score: How unusual one observation is compared with its history.Z = (value - mean) / standard deviation
Correlation coefficient: How closely two series move together, between -1 and +1. It never proves that one causes the other.r = covariance(X, Y) / (sd(X) x sd(Y))
Sample size: How many observations a figure rests on. Below the minimum, Stratos shows INSUFFICIENT HISTORICAL DATA rather than a number.
Monte Carlo simulation: Running an event many thousands of times under a model to see the whole distribution of results, not just one predicted score.

Model quality

Confidence: How reliable the data behind an estimate is — book depth, agreement, freshness, sample size. It is not the chance of winning.
Calibration: Whether stated chances match reality over many picks: outcomes given 60% should land roughly 60% of the time.
Brier score: The average squared error of a probability forecast. Lower is better; 0 is perfect.Brier = average of (probability - outcome)^2
Log loss: A calibration measure that punishes confident wrong calls heavily.
Data freshness: How recent the underlying observation is: real-time under a minute, recent under ten, stale beyond that, or unavailable.
Observation vs prediction: An observation is something that happened — a line moved after a lineup was announced. A prediction is a model output. Stratos labels them separately and never presents one as the other.

Put the maths to work: odds calculator · parlay calculator · how our picks have actually done.