Glossary

American odds#

The +/− price format. −150 means you risk $150 to win $100; +150 means you risk $100 to win $150. The bigger the minus, the more the market expects it to hit.

Also shown as: −110 · +120 · price

Bankroll#

The money you’ve set aside for betting. Your unit is a fixed slice of it, which is what keeps bet sizing honest when things are going well and when they aren’t.

Confidence pool#

A free office-pool format: rank every game in the week from 1 to N, and a correct pick scores the number you put on it. Sixteen games means your best pick is worth 16 and your shakiest is worth 1. Highest total that week wins the pool. It isn’t a bet — there’s no price, no payout and no book involved.

Also shown as: pick’em pool · office pool · confidence pick’em

Line#

The over/under number. You’re betting whether the real result lands above (over) or below (under) it.

Market#

The stat being bet — points, rebounds, receiving yards, strikeouts. Each one gets its own line, so the same player can be an over in one market and an under in another. PRA is points + rebounds + assists; 3PM is threes made; TB is total bases.

Also shown as: PTS · REB · AST · PRA · 3PM · TB · Ks

Over / Under#

Bet the player finishes OVER or UNDER the line for that stat. Some apps call the same two sides More and Less.

Also shown as: More / Less · O / U

Paper bet#

A bet logged for tracking only — no money moves. It grades exactly like a real one, so you can find out what an approach actually does before it costs you anything.

Slate#

All the games on a given day and the props that come with them. Some nights the slate is huge; some nights there’s nothing worth showing, and we’d rather say so than fill the page.

Units#

Bet size as a slice of your bankroll, not dollars — so your sizing scales with you and is never “too small.” 1u is one unit; 0.5u is half.

Also shown as: u · 1u · 0.25u

Calibrated#

A probability is calibrated when the things we call 60% actually happen about 60% of the time. We check ours against real results and publish the comparison — being right on average matters more than sounding certain.

Chance to hit#

Our estimate of how often this side would land if the game were played over and over. It’s a rate, not a forecast of tonight — a 70% chance still misses about three times in ten, and that’s the model working, not failing.

Also shown as: our chance · % to hit · hit % · calibrated chance

Confidence#

How sure the model is, based on how stable the player’s minutes and production have been lately. High confidence still loses regularly — it means the inputs are steadier, not that the outcome is safe.

Edge (the gap)#

How far our projection sits from the line. A bigger gap isn’t automatically a better bet — variance and price matter too, and a huge gap more often means the line knows something we don’t.

Also shown as: gap

Expected points#

Add up each pick’s confidence times its win probability. It’s the score you’d average over many identical weeks. Ranking strictly by probability maximises it — which is exactly why it’s not enough on its own: a pool pays the highest score, not the average one, and everyone else is maximising the same number.

Projection#

Our model’s estimate of the stat — roughly minutes × recent per-minute production. An estimate with a spread of outcomes around it, not a prediction of what will happen tonight.

Also shown as: proj

Recent form (L5 / L10)#

The player’s last five or ten games against tonight’s line. It’s context, not confirmation: our projection already accounts for recent form, so it isn’t a second, independent reason to believe us.

Also shown as: L5 · L10

StatIQ Lean#

The side our model favors — over or under the line. It’s a projection, not a guarantee.

Also shown as: our lean · StatIQ play

Variance#

How much a result swings around even when the projection is right. High-variance stats — threes, home runs — miss badly in both directions far more often, which is a reason to bet less on them, not to avoid them.

Variance band#

How much this stat usually bounces game to game. Wider band = less predictable = lower confidence. The band is drawn from the same chance we publish and grade against, so how much of it sits past the line is exactly that percentage — where that lands away from the raw projection, the card says so.

Also shown as: band

Versus this opponent#

Past games against tonight’s opponent. Genuinely independent of our projection, but usually a handful of games — treat it as colour, not evidence.

Your scenario#

What the numbers look like using the minutes you set instead of ours. It’s labelled as yours on purpose, and the model’s own call stays on screen next to it so the two never get mixed up.

Best price / line shopping#

The same bet is priced differently at different books. We show the best number we can find among the books you told us you actually use. Getting a better price on a bet you were making anyway is the one advantage nobody argues about.

Also shown as: line shopping · best under price

Books pricing it#

How many sportsbooks quote BOTH sides of this game. Only those can be de-vigged, because a book’s cut can only be removed using its own two prices. Two or three books is a quote, not a consensus, and we label it as thin rather than dressing it up.

Also shown as: book count · thin market

Break-even#

The win rate a price needs just to come out level, before anyone is ahead. About 52.4% at −110. Clearing it is the bar a bet has to pass; clearing it on paper is not the same as clearing it in reality.

Consensus line#

The number most books agree on, rather than any single book’s. Using it stops one outlier book from making a play look better than it is.

Expected value (EV)#

The average result if you made this bet many times, from our hit-% estimate at that price. It’s only as good as the estimate behind it — an estimate, not a promise. Negative EV means the price doesn’t justify the bet.

Also shown as: +EV · value

Line freshness#

When we last refreshed this price. Prices move and props get pulled, so an older one may not exist at the book any more — always confirm the live number before you bet it.

Also shown as: updated

Market maturity#

Early lines go up with small limits and a wide cut, and they move a lot. A number that looks great the minute it opens often doesn’t survive the market filling in — which is why some picks are watched rather than priced.

Also shown as: thin market · matured

Moneyline#

The simplest bet there is: which team wins, no points involved. The price tells you what the book charges for each side.

Also shown as: h2h · win market

No-vig (fair) price#

A book’s two prices add up to more than 100% — the extra is its cut. Strip that out and you get the no-vig, or fair, price: the market’s actual read on the chance. We anchor our probability to that number and then let our model nudge it, because the market is a strong starting point and pretending otherwise is how people lose.

Also shown as: fair price · de-vig · fair-to-fair

No-vig win probability#

The bookmaker’s moneyline with its margin stripped out. A book prices both sides to add up to more than 100% — that surplus is its cut. Remove it and you get the market’s honest estimate of who wins. It’s the market’s number, not ours, and we don’t claim to beat it.

Also shown as: de-vigged · fair probability · no-vig

Vig (juice)#

The sportsbook’s built-in cut, baked into the price on both sides. It’s why a genuine coin flip priced at −110 both ways loses money over time — you have to clear the cut before you’ve done anything.

Also shown as: juice · hold · book margin

Win probability#

The market’s chance that a team wins, with the bookmaker’s cut taken out. We take every book’s own two prices, strip its margin, and use the middle number across books. It is the market’s opinion, not ours — we have no game-winner model, and when we tested one against real closing prices the market beat it every year. On game winners this is the best estimate available, which is exactly why we publish it instead of a forecast of our own.

Also shown as: win % · de-vigged moneyline · fair win chance

Beat close#

How often the price we took was better than where that side closed. It says something about our prices, and nothing at all about whether those bets won.

Closing line#

The last price before the event starts, once all the money is in. It’s the market’s best-informed number, which is why we measure our own prices against it instead of against our results.

Also shown as: close

Closing line value (CLV)#

Whether the price moved your way after you bet — a sign you got a good number before the market caught up. It is evidence about the price you took, not proof of profit: you can beat the close by less than the cut you paid and still lose money.

De-vigged CLV#

CLV measured fair price to fair price, with the book’s cut removed from both ends. It’s a smaller sample than the raw figure but a cleaner comparison — the raw number flatters us, because part of it is just the cut shrinking as a market matures.

Also shown as: fair-to-fair · raw CLV

Line move#

How the number moved between when we logged it and when the market closed. Toward our side means the market drifted our way afterwards; away means it didn’t. Both are recorded.

Also shown as: moved toward StatIQ · moved away

Locked price#

The exact price we recorded at the moment we published a play. Everything is graded at that price — never at a better one we found afterwards.

Also shown as: locked · our price

Opening price#

The first price we saw on the board for that side. Comparing it to what we locked shows what waiting was worth; comparing it to the close shows which way the whole market drifted regardless of us.

Also shown as: open

95% CI#

The range the true number plausibly sits in, given how few plays we have. If the range crosses zero, the honest read is “we don’t know yet”, however good the middle number looks.

Append-only record#

Nothing is ever deleted or edited once it’s logged. Plays settle — win, loss, push or DNP — and stay exactly where they were, losses included. A record you can edit is not a record.

Flat 1u#

Every play measured at the same size, one unit, so the record reflects the picks rather than clever sizing. It’s a measurement convention for comparing honestly — not a recommendation for how you should bet.

Also shown as: measured (1u)

Hit rate#

The share of graded plays that won. It says nothing about price — you can win most of your bets at bad numbers and still be down, which is why we show it next to price measures rather than alone.

Also shown as: win rate · % hit

Locked vs watching#

Locked means we recorded a price and it is in our record. Watching means we like the number but haven’t taken a price yet, usually because the market is still thin. You can bet a watched number — it just isn’t ours yet, so we can’t claim it either way.

Also shown as: WATCH · PLAY · watch-only

Model era#

When the model that produced these picks started. Picks from an older version are kept unedited but counted separately — blending them would misrepresent both the old model and the new one.

Also shown as: current model · prior model · record start

Pending#

Logged before the game, not yet graded. It will settle to a win, loss, push or DNP. It never disappears, whichever way it goes.

Push#

The result landed exactly on the line, so the bet ties and your stake comes back. It counts as neither a win nor a loss, and it stays on the record.

ROI#

Return per unit risked across a set of graded plays — what actually happened, not what we expect to happen next. Small samples swing hard in both directions, so read it with the number of plays sitting beside it.

Also shown as: net P/L · units P/L

Sample size (n)#

How many graded plays a number is built on. Below a minimum we hide the rate and show the raw counts instead, because a hit rate off a handful of plays is noise wearing a number’s clothes — and that cuts against us exactly as often as it flatters us.

Also shown as: n= · rate withheld

Tracked vs measurement-only#

Tracked means it counts in our win/loss record. Measurement-only means we froze a price purely to measure ourselves against the close on a bigger sample — it never counts toward our record, good or bad. Neither is a recommendation to bet.

Also shown as: NOT BET · CLV sample · model-qualified

Void / DNP#

The player never played, or the market was pulled, so there’s nothing to grade. The row stays on the record and settles at zero — a DNP is not a play we dropped.

Also shown as: DNP · VOID · NO ACTION · PPD

W-L-P-V#

Wins, losses, pushes, voids, in that order. We show pushes and voids rather than quietly dropping them, because dropping them inflates a win rate for free.

Also shown as: record · W-L-P

Arms (A and B)#

Two versions of the same pre-registered rule graded side by side — a broad one and a narrower one — so we can tell whether the idea works or just the width. A pick can appear in both, so their records overlap and must never be added together.

Also shown as: Arm A · Arm B

Backtest#

Running a rule over past seasons to see what it would have done. Useful for ruling ideas out, and very easy to fool yourself with — almost anything can be made to look good in hindsight.

Backtest preview#

Real output from our model on past games, shown before a sport has a live record. It is history, not a track record, and it is labelled that way everywhere it appears.

Forward test#

Grading a rule on games that hadn’t been played when the rule was written. A backtest can be tuned until it looks good; a forward test can’t, which is the only reason it’s worth anything.

Frozen model#

The model file is fixed and fingerprinted, so the version that produced week one is provably the same one producing week nine. Retraining mid-test would invalidate the whole thing, which is why we publish the fingerprint.

Also shown as: model artifact · sha256

Measurement only#

We are measuring something in public, not recommending it. Nothing on a page marked this way is a bet we’re telling you to make, and no return is being claimed.

Play = 0#

A row logged at zero stake. It’s on the record so it can be graded and counted honestly, but no bet was placed on it and none is suggested.

Pre-registered#

The rule was written down and frozen before the season, then published, so it can’t be quietly adjusted once results come in. It’s the difference between a test and a story told afterwards.

Ask / all-in price#

The ask is the cheapest price anyone is offering to sell you that contract, in cents out of 100. Kalshi then charges a trading fee on top, so we show the all-in price — ask plus fee — because that is what leaves your balance and what the odds beside it are calculated from. Both are live, so confirm them on Kalshi before you act.

Also shown as: orderbook · marketable ask · all-in

Kalshi#

A regulated exchange where you trade contracts on outcomes instead of betting against a book. Contracts are priced in cents out of 100, and a “25+” market is the same idea as an over: YES is the over, NO is the under.

Also shown as: YES / NO · contract

Kalshi fees#

Kalshi takes a trading fee that scales with the contract’s price, and it comes straight out of your return. We price it in, because a trade that looks break-even before fees is a losing trade after them.

Parlay#

One bet made of several legs where every leg has to win. The payout is bigger because the chance is smaller — stacking legs multiplies the book’s cut too, so it isn’t a way to turn a small edge into a big one.

Pick’em#

Apps where you choose several players to go more or less than a number and the whole entry pays out together. There’s no price to shop — the payout is fixed by the app, so a better number is the only thing you can control.

Power vs Flex#

Power: every pick must hit for a bigger payout. Flex: you can miss one (on larger entries) for a smaller payout.

Thin market#

Only a couple of contracts are resting at that price, so you could never get real size filled. An “edge” you can’t actually take isn’t an edge — it’s usually a stale quote, which is why we hide those rather than count them. Separately: if we say we couldn’t read a market, that’s our request failing, not a thin market — we keep those visible and don’t count them as thin.

Also shown as: phantom price · liquidity · price looks off · couldn’t read

Correlated bets#

Bets that move together — same game, same player. Stacking them isn’t spreading risk: one blowout or one quiet night busts all of them at once, so treat them as one bet when you size them.

Differentiation#

Deliberately submitting something the rest of your pool won’t. If your entry matches someone else’s exactly, your scores are identical every single week — an exact tie, not a close race. Differentiating costs you expected points and buys you the variance you need to finish first. How much you should buy depends on how many people are in the pool and what it pays.

Also shown as: contrarian · leverage

Exposure (at risk)#

How much of your bankroll is tied up in bets that haven’t settled. It’s the number that decides how rough a bad night feels, and it’s easy to lose track of one small bet at a time.

Also shown as: at risk

Injury watch#

A teammate is listed out, so minutes and usage could shift. The line may not have caught up yet — which cuts both ways, and is a reason to look closer rather than to bet bigger.

Questionable (Q)#

Listed Questionable on the injury report — they may play, may be limited, or may sit. If they sit, the prop usually voids; if they play limited minutes, the line often hasn’t priced that in.