Glossary

What each figure and label on Troxler means, written from how the app computes or sources it. Where one word is used for two different figures, each gets its own entry and says where it appears. A dotted underline on a term elsewhere in the app opens the short definition and links here.

Filter terms

66 terms

Prices and markets

Moneyline

Also shown as: ML

The price, in American odds, on a team to win the game outright.

American odds: -150 means stake 150 to win 100; +130 means stake 100 to win 130. The Line Movement board shows each book's first tracked and latest moneyline for the home side.

Spread (point spread)

Also shown as: Spread, Run line, RL, Puck line, PL

The handicap a team must cover, in points, with its American price. Baseball calls it the run line and hockey the puck line.

Run lines and puck lines sit at plus or minus 1.5, so on those boards the price carries the movement and the line is shown as context. Not the same thing as the bid-ask spread on a prediction market.

Spread (bid-ask)

Also shown as: Bid-ask spread

On Kalshi or Polymarket, the ask minus the bid. A venue price is shown only when the order book is two-sided and this gap is 5 cents or less.

The venue price must also have been captured in the last 3 hours. Not the same thing as a point spread.

See also: Ask, Bid, Spread (point spread)

Total

Also shown as: O/U, Over/under

The combined score line for both teams, with the American price on the over.

ATS

Also shown as: Against the spread

Record against the closing spread, from game results.

Implied probability

Also shown as: Prob (props), Implied prob

The win chance a price implies, with the margin left in. On the props boards the Prob column is this figure for the displayed price.

Minus odds: 150 / (150 + 100) = 60% for -150. Plus odds: 100 / (130 + 100) = 43.5% for +130. Both sides of one market add up to a little over 100% because the book's margin is still in. On the Predictions page "Prob" means something else: the model probability.

See also: Model probability, Vig, De-vig

Vig

Also shown as: Hold, Margin, Juice

The book's margin: how far the two sides' implied probabilities add up to more than 100%. Line Movement shows it as the book's hold.

De-vig

Also shown as: De-vigged, No-vig

Removing the margin so both sides sum to 100%. Troxler scales each side's implied probability by their total, from one book's two prices taken at the same time.

This is proportional de-vig. The Books figure on a game card averages each book's de-vigged moneyline with equal weight. Line Movement ranks and colours moneyline moves on the de-vigged figure; the raw figure is on the chip hover. Prediction-market asks are not de-vigged.

Best price

Also shown as: Best Price, Price (Team Props)

The highest-paying price among US-regulated sportsbooks for that line. Kalshi and Polymarket never win it. Blank when no US book prices the line.

Player Props calls this column Best Price and Team Props calls it Price; it is the same figure. Your book filter narrows which books can win. A price more than 25 points of probability away from the row's market is treated as a bad quote and skipped. Offshore books, Pinnacle and DFS apps may be listed in a breakdown but never win.

US-regulated book

A sportsbook on Troxler's US-regulated list, such as DraftKings, FanDuel, BetMGM or Caesars. Best price and the published record use these books only.

The list: DraftKings, FanDuel, BetMGM, Caesars, BetRivers, Fanatics, theScore Bet, Hard Rock Bet, betPARX, Bally Bet, bet365, Circa, PointsBet and Unibet. Picks priced only at an offshore book, a DFS app or "Best Available" are listed without a price and left out of the record figures.

Ask

On Kalshi or Polymarket, what a buyer pays for the contract right now. Venue prices on Troxler are asks on fresh, two-sided books within a 5-cent spread.

Shown in cents (45c means a 45% implied chance) with the matching American price beside it.

See also: Bid, Mid, Venue ask, Spread (bid-ask)

Bid

On Kalshi or Polymarket, the highest price a buyer is currently offering for the contract. Selling fills at the bid.

Mid

Also shown as: Midpoint

Halfway between the bid and the ask. A prediction-market midpoint, not a price you can fill, and never eligible for best price.

Venue ask

Also shown as: ask-only

A Kalshi or Polymarket ask shown on a props row because no US sportsbook prices that line. It is for reference and never counts as best price.

Kalshi

Also shown as: K

A US prediction-market exchange. On a game card, K: is the price Kalshi quotes for the named team, as a probability.

It is the qualifying ask (two-sided book, 5 cents wide or less, captured in the last 3 hours) or no price.

Polymarket

Also shown as: P

A prediction-market exchange. On a game card, P: is Polymarket's ask for the named team, as a probability, shown only when its book qualifies.

Books

Also shown as: Books:

Win probability from sportsbook consensus after removing each book's margin, with the number of contributing books in parentheses.

Each book that priced both sides is de-vigged on its own, then the books are averaged with equal weight. Moneyline only. On CFB and NHL game cards it leads the Markets row. On NFL and MLB it is drawn on the win-probability rail from US-regulated books only, with the book count and the time the prices were captured.

% Win (game card)

The large win percentage in a game card's Markets row is a market price, not the model: the books consensus if present, else Kalshi, else Polymarket.

Win probability (rail)

One team's chance to win on a 0 to 100 line. Diamond: Troxler model. Ticks: B books consensus, K Kalshi, P Polymarket. Before the game starts only.

The rail adds a picture to numbers the card already prints in its Model and Markets rows. Ticks within five points share one key, for example K·P. It is removed at kickoff, because the model is a pre-game number and the venues keep trading.

Markets (game card row)

On a game card or game page, the row of market prices: Kalshi and Polymarket, plus the sportsbook consensus where we have it. The Model row above it is Troxler's own.

Markets (page and menu)

The menu of slates and boards, and the title of the Games page. It lists every game with its lines, prices and model output.

CFB Markets, NHL Markets and EPL Markets are each sport's board under the same menu. During setup, a "Markets" tab filters the platform list to Kalshi and Polymarket.

Volume

Also shown as: Vol

Trading volume by venue: Polymarket in US dollars and Kalshi in contracts traded, reported separately because the units differ.

Kalshi volume on a game adds the contracts traded on both teams' markets.

Liquidity

Also shown as: Liq

Polymarket's reported liquidity for the market, in US dollars, as supplied by its feed. Shown in the Polymarket breakdown only; Kalshi has no such field.

Gap

Also shown as: Venue gap

On EPL match cards, the Kalshi ask minus the Polymarket ask for the same outcome, in percentage points.

Blank when only one venue has a qualifying price. The asks are not de-vigged.

Percentage points

Also shown as: pp, pts

The plain difference between two percentages: 52% to 55% is +3 pp. A positive move means that side became more likely.

Game card badges measure the raw move in implied probability since the first tracked price. The Line Movement board measures the de-vigged home win probability. Both highlight a move of 3 pp or more.

Snaps (Line Movement)

Also shown as: Snapshots, Price captures

On the Line Movement board, how many times we captured that book's prices for the game. Not snap counts from football.

For a game that has started, only captures before the start count.

Open (Line Movement)

Also shown as: Open, At close

The first price our tracking recorded for each book, usually about a day before the start. Not the book's own opener.

Games marked "at close" compare against the last capture before the start; in-play prices are excluded.

Model and record

Model probability

Also shown as: Prob (predictions), Model

Troxler's own probability for the outcome. On the Predictions page "Prob" is this figure, as it stood when the prediction was published.

On game cards the Model row shows the model's win, cover and over probabilities. For CFB the cover and total probabilities are the de-vigged market consensus at the most-quoted line; the CFB model publishes the moneyline only.

Sim

Also shown as: Simulation, Sim probability

The share of simulated games in which an outcome happened. The NFL sim plays each game play by play thousands of times; it is a separate model from the game model.

Every Sim figure is a count over the same simulated games, so lines, player stats and combinations agree with each other. A win share counts outright wins only; a tie counts for neither side and is shown on its own. A market price has no tie outcome, so beside a market the gap includes the tie share. The game model's win probability, margin and total come from different models and can disagree with the Sim. Graded 2024 to 2025 sim outputs did not beat closing lines.

Joint probability (Sim)

Also shown as: Together, If independent

The share of simulated games in which every chosen outcome happened together. "If independent" multiplies the separate probabilities instead.

When the two differ, the outcomes are linked in the simulation: a team that covers a big spread also tends to score more, so its cover and the over happen together more often than independence would say. A push counts as not happening.

Edge

The gap between the model probability and the implied probability of the market price it was measured against, at publication. Positive means the model rates the side higher.

Written by the publishing pipeline, not recalculated in the app. Not shown when the pick has no US-regulated price, or when an NFL game carries a QB change.

QB change: edge not served

Also shown as: Edge not served

An NFL starting quarterback changed. The model does not read injuries, so its number describes the old starter; no edge is shown and no prediction is published.

A side is flagged when the announced starter differs from the last game's starter, or the depth-chart QB1 is Out, Doubtful or on a reserve list.

Predicted margin

Also shown as: Pred Margin

The model's projected winning margin in points or runs, shown as the projected winner and the amount, such as "by 3.0". It is not a sportsbook spread.

It can come from a different model than the win probability on the same card, so the two need not agree at a glance.

Predicted total

Also shown as: Pred Total

The model's projected combined score for both teams, for the full game.

Units (stake)

Also shown as: Unit, u

On a prediction card, the stake recorded on that pick. Stake size is set per pick and has not been constant across the record.

Units (profit)

Also shown as: Units Profit, Units won

In the Record and Predictions headers: units won on winning picks at the price published with the pick, less units lost on losing picks. A push adds nothing.

Because stakes vary by pick, this is not a flat-stake figure. The record counts only picks priced at a US-regulated sportsbook at publish.

ROI

Units profit divided by total units staked, with pushes included in the stakes. It is not a win rate.

Win rate

Wins divided by wins plus losses. Pushes are left out.

Push

Also shown as: W-L-P

A graded pick that neither won nor lost, such as a result landing exactly on the line. The stake is returned and nothing is added.

CLV

Also shown as: Closing line value, Median CLV

Published price versus the market's closing price. Positive CLV means the published price was better than the close.

Computed as the implied probability at the close minus the implied probability at the published price, one-sided and not de-vigged, in percentage points. The close is DraftKings, or the market median where DraftKings had no quote. Covers MLB game markets.

Beat the close

Also shown as: Beat close

The share of measured picks whose CLV was above zero: the published price was better than the closing price.

Brier score

Also shown as: Brier

The mean squared error of probability against outcome. Lower is better; 0.25 is a coin flip.

Skill versus market is one minus the ratio of the model's Brier score to the Brier score of the book's implied probability at the published price.

Log-loss

The mean negative log-likelihood of the outcomes under the stated probabilities. Lower is better, and confident misses cost the most.

Calibration error

Also shown as: ECE, Expected calibration error

The sample-weighted average gap between predicted and observed win rates across the bins shown. A calibrated model sits on the diagonal of the reliability chart.

Picks are grouped into 10 probability bins, equal width by default. Pushes are dropped.

Props

Alt markets

Also shown as: Alt lines, Alternate lines, Main line

Off shows main lines only: lines with both an Over and an Under posted. Yes/No markets such as touchdown scorers are always shown.

Off hides alternate lines posted on one side, lines a sportsbook lists only as alternates, and lines only a prediction market quotes. Alt markets start Off for NFL and NBA player props.

Hit rate

The share of games in a window in which the player or team met this line and side, from game results.

Over hits when the result is above the line, Under when it is below. In the stats modal a result exactly on the line usually counts as a miss; a few views leave such games out instead.

L3, L5, L10, L20

Also shown as: L3, L5, L10, L20, Last N

The last 3, 5, 10 or 20 recorded games. On the props boards L3, L5 and L10 are hit rates at the line over that many games.

L20 appears on the stats tab period filter and on MLB Strikeouts, where it is the window the stats are added up over, not a hit rate.

H2H

Also shown as: Head to head

Hit rate at this line against this opponent, over the seasons our data holds.

Tap or hover a cell for the count of games. For golf the same column is course history.

Opponent rank

Also shown as: Opp Rk

Where the opponent ranks for the statistic this prop measures, from season game results. Early in a season the prior season's rank is used and labelled.

Break-even at this price

Also shown as: Break-even

The hit rate a price needs to pay back over time: its implied probability, margin included. On the props boards a tick in each hit-rate cell marks it for the price shown.

-110 needs 52.4%, +150 needs 40.0%, -1450 needs 93.5%. A row with no price, or priced only at a prediction-market midpoint, has no tick. A rate past the tick is a comparison of two numbers, not a forecast: a short window of games says little about the next one.

Percentile (player profile)

On a player's profile shape, where the player ranks among qualified players at the same position this season, 0 to 100. Higher is better on every axis.

Axes where a lower raw number is better are flipped so the outer edge is always good. At least 8 qualified players are needed for a ranking. Park Performance has its own percentile, defined under park percentile.

Baseball

Park factor

Also shown as: PF

The ballpark's park factor from our MLB games data, where 100 is neutral and higher favours hitters. Game pages call 105 and up hitter friendly, 95 and below pitcher friendly.

xwOBA

Expected weighted on-base average from Statcast batted-ball and plate-discipline data. Park Performance uses career regular-season Statcast from 2015 on.

xwOBA delta

Also shown as: xwOBA Δ, xwOBA here / elsew., Here, Elsewhere

A hitter's xwOBA at this park minus the same hitter's xwOBA at every other park. Shown as observed; on a thin sample it is mostly noise.

A within-player comparison on separate samples. It is not a park factor and not a projection of tonight. Rows are ranked only with 25 plate appearances at the park and 100 elsewhere.

Shrunk delta

Also shown as: Shrunk Δ

The xwOBA delta discounted for sample size, as delta x PA / (PA + 250). Park Performance ranks on it by default.

Delta variance falls as 1 over plate appearances, so ranking on the raw number would put the weakest evidence on top. 250 is measured from this data's own variance. Both numbers are shown so the size of the discount is visible.

Park percentile

Also shown as: Park pctile

Where a hitter's raw xwOBA delta falls among every hitter with a qualifying sample at this park, not just tonight's.

Blank when fewer than 50 hitters qualify at the park.

K%

Also shown as: Opp K%

Strikeouts divided by plate appearances over the window shown. Opp K% is the opposing team's rate over its recent games.

K/9

Also shown as: SO9, Season SO9

Strikeouts per nine innings. K/9 on MLB Strikeouts is over the window shown; Season SO9 is the same rate for the full season.

SwStr%

Also shown as: Swinging strike rate

Swinging strikes divided by pitches thrown, over the window shown.

CSW%

Also shown as: Called strikes plus whiffs

Called strikes plus swinging strikes, divided by pitches thrown, over the window shown.

Opp K% vs Hand

The opposing hitters' plate-appearance-weighted season K% against the starter's throwing hand. Before lineups post it uses the roster (25+ PA), marked TEAM.

Football

Snaps played

Also shown as: SNP, SNP/G, Off snaps, Def snaps

In NFL stats, the count of offensive or defensive snaps the player was on the field for. SNP/G is that count per game played.

RYOE

Rush yards over expected, per carry: how many more yards a runner gained than expected on his carries. Read from our NFL player stats; the app does not compute it.

Future value (Survivor)

Also shown as: Future value

The number of later weeks in which the team clears the win-probability threshold, and its single best one, so the cost of using the team now is visible.

The threshold is a 65% win probability, from the consensus de-vigged moneyline. Only weeks that already carry a price count; an empty value means no later week is priced yet, not a verdict on the season.