How Playoff Odds Are Calculated

The short version

Every team's odds come from three ingredients: where it is projected to finish, where the playoff cut line is projected to fall, and how much season is left for those two numbers to move. The bigger the gap between a team's projection and the cut line, and the later in the season it is, the more certain the odds become. Everything is recomputed from live standings every five minutes.

NHL playoff odds, step by step

1. Project final points. Points already banked stay banked. The remaining games are projected at the team's points pace regressed toward a preseason estimate of its strength (last season's pace, pulled most of the way back to the league average of about 1.12 points per game) with a 40-game prior, so a 7-3-0 start counts for something but does not project to 115 points, and the odds do not lurch from the preseason number after one game. By midseason the prior has little pull, and by April almost none. The full schedule is 84 games from 2026-27, 82 before that. The "on pace for" number shown on team pages is the raw pace extrapolation; the odds run on the regressed projection.
2. Find the cut lines. An NHL team qualifies by finishing top three in its division or as one of two conference wild cards. We project both thresholds from the current standings using the same regressed projection: the division line is the midpoint between the third- and fourth-place teams' projections, and the wild card line is the midpoint between the second and third wild cards. Each line has a historical floor (90 and 94 points over 82 games, scaled to the schedule) so early-season standings cannot drag a cut line below where NHL cut lines actually land.
3. Convert the gap to a probability. The difference between the team's projection and each cut line runs through a logistic (S-shaped) curve. At the cut line the odds are 50%; a few points either side moves them quickly; far above or below, the curve flattens toward 99% or 1%. The curve's width is tied to the square root of the games remaining, matching how much an NHL team's point total can still move: about 12 points of spread with a full season left, 8 or 9 at the halfway mark, and barely one with a game to go, so a team sitting above the line in the final week is near-certain. Early in the season it is widened further for uncertainty about how good each team really is, an allowance that fades as games are played. The curve is slightly steeper for the division path (fewer competitors) than for the wild card path.
4. Take the better path. The team's playoff probability is the higher of its division and wild card probabilities. Team pages show which path is active.
5. Position bonus. A team currently holding a playoff spot gets a small edge that grows through the season (up to 3 points shaved off the cut line by season's end), reflecting that incumbents are displaced less often than pace alone suggests. This, the prior, and the early-season widening were set by backtesting the model against the 2023-24 through 2025-26 seasons.
6. Clinch and elimination. The NHL's official x/y/z/p clinch indicators set odds to 100%, and an "e" sets them to 0%. Until then, odds are capped at 99% and floored at 1%.

Stanley Cup odds on the bracket page use a separate series model: point percentage and goal differential set each team's strength, a logistic curve turns the strength gap into a single-game win probability adjusted for home ice (a .600 team is about a 55% favorite per game and a 63% favorite in a series against a .500 team), and a best-of-seven calculation over the remaining games turns that into series odds. Championship odds walk the actual bracket: each future round weighs every opponent a team could meet by that opponent's own chance of getting there, so the field's Cup odds always add up to 100%. Before the bracket is set, the first round is projected from the current standings.

MLB playoff odds, step by step

1. Estimate true talent. Win-loss record is noisy, so each team's underlying strength blends three signals: its actual win rate, its Pythagorean win rate from runs scored and allowed (which strips out luck in one-run games), and a regression-to-.500 prior that fades as the sample grows.
2. Project wins and cut lines. Projected wins extend the current pace over 162 games. The division cut line is the projection of the division leader (or runner-up if the team leads); the wild card cut line is the third wild card, or the first team out if the team is already in.
3. Model the rest of the season. Remaining games are treated as a binomial distribution around each team's true-talent rate. That gives the probability of finishing ahead of any given rival. Division odds are the joint probability of finishing ahead of every division rival; wild card odds compare the team to the bubble rival it needs to beat or hold off.
4. Take the better path, apply official status. As in the NHL, the playoff probability is the higher of the division and wild card paths, clamped to 1-99% until MLB's official clinch or elimination status makes it 100% or 0%.

How this compares to FanGraphs, MoneyPuck, and FiveThirtyEight

The best-known playoff odds are simulation models. FanGraphs (MLB) and MoneyPuck (NHL) simulate the remaining schedule thousands of times, game by game, using player-level projections, current depth charts, and home-field effects, then count how often each team makes it. FiveThirtyEight ran similar Elo-based simulations until its sports forecasts ended in 2023; the site itself closed in 2025, which is why searches for "538 playoff odds" no longer lead anywhere current.

Lindy's Five is a standings model. The practical differences:

In practice the models agree on who is in and who is out, and they converge as the season ends. Expect the largest gaps (often 5-15 percentage points) in the middle of the season on bubble teams with unusually hard or easy remaining schedules, or with major injuries a standings model cannot see.

What the odds do not account for

• Injuries, trades, and roster changes (only their effect on results, once it shows up in the standings).
• Remaining strength of schedule and home/away split.
• Tiebreakers. Regulation wins (NHL) and head-to-head records (MLB) are not modeled; a team exactly on the cut line shows 50%.
• Early-season noise. NHL odds need 5 games and MLB odds need 10 before the model engages; before that every team shows 50%.

Where the data comes from

Standings, schedules, scores, and clinch indicators come from the official NHL and MLB stats APIs and are refreshed every five minutes. Preseason "way-too-early" odds on team pages carry last season's final standings forward (regressed toward the league average) and use the same projection the live odds start from, so opening night picks up where the preseason number left off.

Frequently asked questions