How Prop Radar works
Prop Radar reads the player props posted for the current slate, projects each player's production for that market, converts the projection into a probability at the posted line, compares that with the probability implied by the price, and ranks what is left. This page explains the method without publishing the formulas themselves.
How do props enter the evaluation pool?
Props enter the pool when a sportsbook posts a line for a player and market Prop Radar supports, and when there is enough stored performance history to project that player for that market. Props without usable data are excluded rather than estimated from thin air.
Board data is rebuilt on a schedule and read from a stored snapshot, so every visitor sees the same evaluated slate rather than a live query fired on page load.
What information is considered?
Evaluation draws on performance over several time windows, usage and opportunity, matchup context, availability, the posted line and the price, plus sport-specific conditions such as weather for outdoor games and park factors for MLB.
Current-season performance
This season's per-game production is weighted most heavily once a player has enough games logged.
Recent performance
Rolling last-three and last-five game windows, ordered by actual game date across seasons.
Historical performance
Previous-season averages are used as a smaller, fading prior, and as the fallback for players with no current-season games.
Performance against the opponent
Where prior meetings exist in the stored game logs, that head-to-head history is shown alongside the projection, market by market.
Player usage and opportunity
Role and volume signals — snaps, targets, carries, lineup slot — determine how much production a player is in line to see.
Matchup context
Opponent strength against the specific market, defensive profile, pressure and trench indicators where the data supports it.
Availability and injuries
Injury reports are ingested weekly for the NFL; a ruled-out teammate can change the opportunity picture for the players around them.
Sportsbook line
The posted line is part of the evaluation — a higher line is harder to clear and the probability moves with it.
Implied versus projected probability
The price implies a probability. Prop Radar compares that with its own projection-based probability for the same line.
Market and odds context
Odds are compared across the sportsbooks in the feed, because books post different lines and prices for the same prop.
Weather
For outdoor sports, wind and conditions are considered for the markets they actually affect.
Ballpark context
For MLB, park factors and the game environment feed the home run and total-based research.
Volatility and consistency
Spread in a player's game-to-game results affects the confidence attached to a projection and how a prop is categorized.
Correlation and risk
Same-player and same-game relationships are flagged when props are combined into a slip.
What is Spot Score and Prop Radar Rank?
Spot Score is the composite number used to order the board. It blends the projection's distance from the line, the probability at that line, the price on offer and the reliability of the underlying data. Prop Radar Rank is simply a prop's position once every evaluated prop is sorted by that score.
This is why the board is not a sorted odds list. A long-priced prop with a weak projection and a short-priced prop with no room against the line can both rank poorly, for different reasons.
How are projected and implied probability used together?
The sportsbook price implies a probability. Prop Radar independently estimates the probability that the player clears the posted line, based on its projection and the spread of that player's results. The two numbers are compared; the gap between them is what a prop's value is judged on.
Probability is always tied to a specific line. Raising the line lowers the over probability and raises the under probability, and the displayed numbers move accordingly. When required data is missing, the probability is withheld rather than replaced by a default value.
How is recent form balanced against longer-term performance?
Recent windows carry the most weight, the current season carries the next largest share, and the previous season is a small and fading prior that fades further as the current season accumulates games. A player with no current-season games is projected off last season and flagged as such.
Game logs are ordered by actual game date across seasons, so a player who changed teams or roles is not carried by stale production from two years ago.
How is the opponent factored in?
Opponent strength is measured against the specific market being projected, not against a single overall defensive ranking. Where prior meetings exist in the stored game logs, that head-to-head history is shown next to the projection, market by market, with an honest note when there are no past meetings.
Head-to-head history is treated as context for the reader. Small samples, coaching changes and roster turnover make old meetings weak evidence, so they are presented rather than allowed to swing a projection.
When do injuries, weather and park effects matter?
They matter where the data supports it and the market is actually affected. A ruled-out teammate changes the opportunity picture for the players around them. Wind matters for some outdoor markets and not others. Park factors and game environment feed MLB home run research.
Adjustments are applied only for the market they plausibly affect, and each one is capped so that no single contextual factor can dominate a projection.
Why doesn't a high probability mean a guaranteed outcome?
A probability is a long-run frequency, not a promise. A prop estimated at 65% is expected to miss roughly one time in three, and any single game can break against the projection through injury, game script or variance. Prop Radar avoids words such as lock or guarantee for this reason.
Confidence and probability are separate readings. Confidence describes how much data stands behind a projection; probability describes the chance of clearing the line. A high-probability prop with thin data is treated differently from one with a deep sample.
How are volatile props treated?
Props from players with wide game-to-game swings are categorized as higher variance rather than hidden. Their probabilities are shrunk toward the middle to reflect the uncertainty, and they are labelled so a user can decide whether the payout is worth the swing.
Why do sportsbooks post different lines for the same prop?
Books set their own numbers based on their models, their exposure and their customers, so the same player can have different lines and different prices at different books. Prop Radar shows the book alongside the line, and the board can be filtered to the books you actually use.
A half-point of line difference can matter more than the price, which is why the line and book are shown together on every card. See the Prop Board for the filtering controls.