Sports Betting Guide

How bets, prices and bet sizing work, and how to use each tool in the Lab. New? Start with Your First Bet.

Welcome to Gambletron AI

Gambletron AI is a sports research lab. We run the numbers on every game - trends, matchups, prices, and what our models found - so you can make your own call, whether it's your first bet or you've been at it for years.

New here? Start with Your First Bet

Five steps, in order: decide your bankroll and unit, pick a game you know, check the price, size the stake, then log and grade it. It links to the right tool at each step and remembers your progress.

Start Your First Bet

What the Numbers Say

Only money you can afford to lose. A bankroll is money set aside for betting - never rent, bills or savings. Losing streaks happen even with a real edge.
Flat stakes are the safe baseline. Until a record shows a consistent edge, the same stake on every bet limits the damage from a wrong estimate. Kelly sizing assumes the edge is real.
Line shopping is the simplest edge. A half-point or a few cents of juice between books compounds over hundreds of bets.
A record beats memory. Logging every bet, win or lose, is the only way to see whether a process works.
Parlays cost more. Every leg adds the book's margin, so parlays pay less than their true odds. A single bet pays that cost once.

Go deeper in Learn

Each article explains one of these ideas with the math shown.

Moneyline

The simplest bet - pick which team wins. The odds tell you how much you win.

Example

Yankees -150 vs Red Sox +130

To win $100 on NYY you risk $150. A $100 bet on BOS wins $130.

Common use: Games with a clear winner, when a moneyline pays enough vs. the spread.

Point Spread

The book sets a margin. The favorite must win by more than the spread; the underdog must lose by less (or win outright).

Example

Chiefs -6.5 vs Raiders +6.5

Chiefs must win by 7+ for that bet to cash. Raiders cover if they lose by 6 or less (or win).

Common use: NFL and NBA, where big favorites price steeply on the moneyline.

Over / Under (Total)

Bet on the combined score of both teams going over or under a number set by the book. Doesn't matter who wins.

Example

Total: O/U 8.5 runs (MLB)

Over 8.5 cashes if the two teams combine for 9+ runs. Under cashes at 8 or fewer.

Common use: MLB and NHL, where predicting a winner is hard but pace of play is steadier.

Parlay

Chain multiple bets together. All legs must win. Pays more but is much harder to hit.

A 3-leg parlay where each leg has 50% true odds only hits 12.5% of the time - but books pay as if it's higher to entice you.

Common use: Small-stakes entertainment. Not a core strategy.

Arbitrage

Book two different sportsbooks against each other - one on each side - so that both sides pay out about the same, whatever the result - if every bet is accepted at those prices. Use the Tools → Arbitrage to find and calculate these.

Same payout on either side, if both bets stand Requires accounts at multiple books Books may limit accounts that arb too aggressively

Player Props

Bet on a single player's stat line: yards, touchdowns, points, hits, strikeouts. Not tied to who wins or the final score.

Example

Baker Mayfield Over 245.5 Passing Yards -115

Cashes if he throws for 246 or more. Fewer than 246 is a loss regardless of the game outcome.

Common use: when a specific matchup (a receiver against a weak corner, a pitcher against a light-hitting lineup) reads better than the season-long line.

Coverage varies: college player props aren't offered in every state, and fewer books price them than pro sports.

Futures

A single price on a season-long or tournament outcome — division winner, league MVP, championship. Money sits until the outcome resolves.

Example

Chiefs to win Super Bowl +550

A $100 bet returns $550 if they win it. If they don't, the stake is lost when the season ends.

Common use: finding a price early before the market catches up, or hedging a live position closer to the finish.

Betting Glossary

Every term you'll see on this site, in plain English. The same definitions appear as tooltips wherever you see dotted-underlined words.

+EV: A price that pays more than its fair price (positive expected value). A finding, not a guarantee: the fair price is itself an estimate.
3-on-3 overtime: Regular-season overtime: five minutes of 3-on-3, then a shootout if it's still tied. Overtime numbers here count the whole OT period, including any 4-on-3 power play.
3rd down conversion: The share of third downs turned into a first down or touchdown.
aDOT: Average depth of target - how far past the line of scrimmage a quarterback's throws travel on average. High means a downfield passer; low means a short, quick game.
ANY/A: Adjusted net yards per pass attempt - passing yards plus 20 per touchdown, minus 45 per interception, minus sack yards, divided by attempts plus sacks. One of the best simple measures of passing efficiency.
arbitrage: Prices at different books that, staked on every outcome, pay back a little more than the total stake whichever side wins - but only if every bet is accepted and settles at those prices. Rare, and books often move or limit the line quickly.
ATS: Against the spread - whether a team covered the point spread, not just whether it won.
ballpark factor: How much a specific stadium inflates or suppresses scoring, relative to average. Coors Field boosts runs; Oracle Park suppresses them.
bankroll: The total money you've set aside for betting. Never bet money outside your bankroll.
best price: The highest payout any major US sportsbook is offering on that side right now, and which book has it.
blitz rate: How often a defense sends five or more pass rushers.
books: How many major US sportsbooks are pricing this exact bet. More books means a sturdier fair price; with only two, one book is the whole comparison.
both sides: The Over is priced above fair at one book and the Under at another. Each side's fair price comes from the other books, and here the books disagree on the line. Each is a price gap on its own, not a lean either way.
bullpen fatigue: How heavily a team's relief pitchers have been used in recent days. A worn-out bullpen is more likely to give up late runs.
closing line: The final odds available just before a game starts - considered the most accurate market price.
CLV: Closing Line Value - how your odds compare to the final line before the game starts. Beating the close consistently is the strongest evidence of a real edge.
Corsi: Shot-attempt share at 5-on-5 (CF%): every shot on goal, miss and blocked shot, for vs against. A puck-possession measure; above 50% means a team controls play.
CPOE: Completion % Over Expected - a quarterback's actual completion rate minus what's expected given each throw's difficulty (distance, coverage). Isolates accuracy from a scheme that only calls easy short throws.
devig: Removing the sportsbook's built-in commission from a price to see what the market really thinks the odds are. Comparing a book's devigged number to our own estimate is where an edge shows up.
EPA: Expected Points Added - how much a single play changed the offense's expected points, averaged across every play. The single most predictive public football stat; above 0 means a play (or a team's plays overall) gained more than expected.
ERA: Earned run average - earned runs a pitcher allows per nine innings. Lower is better.
EV: Expected value - how much a price pays compared with its fair price, per unit staked. +3% EV means that if the fair chance is right, you'd average 3 cents of profit per dollar over many bets like it. Math on market prices, not a prediction of any one result.
explosiveness: How big a team's successful plays are, in points added per successful play (College Football Data's IsoPPP). For a defense, lower is better.
fair price: What a price would be with the sportsbooks' commission removed, averaged across the other books pricing it. Shown as a chance (37.6%) and as odds (+166). A book paying more than the fair price is where EV comes from.
FPI: ESPN's Football Power Index - an opponent-adjusted power rating (expected point margin vs. a league-average team on a neutral field). Used here as a free cross-check alongside the model's own composite, not a replacement for it.
GSAx: Goals saved above expected: the expected goals a goalie faced minus the goals he allowed. Positive means he stopped more than an average goalie would have on the same shots.
half kelly: Betting half of what the Kelly Criterion recommends - the standard pro approach, trading a little growth for much lower variance.
handedness split: How a hitter or lineup performs against left-handed versus right-handed pitching. Some lineups are dramatically better against one side.
havoc rate: The share of a defense's plays that end in a tackle for loss, a forced fumble, an interception or a pass broken up.
hedge: Betting the opposite side of an existing bet to reduce how much you can lose (or win) on it.
high-danger chances: Shot attempts from the most dangerous spots (mostly in close and in the slot). HDCF% is a team's share of them at 5-on-5.
implied probability: The win probability suggested by the odds. -110 implies 52.4%. If you think the true chance is higher, the bet has value.
juice: Same as vig - the sportsbook's cut baked into the odds. Lower juice means better prices for you.
K-BB%: Strikeout rate minus walk rate - the cleanest single number for how much a pitcher dominates a lineup. Higher is better; 15%+ is very good.
K/9: Strikeouts per nine innings pitched.
Kelly: The Kelly Criterion - a formula for how much of your bankroll to stake based on your edge. Most pros bet Half Kelly (50% of the formula) to reduce variance.
line shopping: Comparing odds across several sportsbooks to get the best price on the same bet. The single easiest way to improve results.
line yards: Rushing yards credited to the offensive line: full credit for the first few yards of each run, less for yards past that, none past 10. Isolates blocking from the running back.
man coverage rate: How often a defense plays man-to-man coverage rather than zone.
market probability: The chance a sportsbook's price implies, with its commission still in. The market's view, not our model's.
middle: Betting both sides of a game at different lines so both bets can win at once (e.g. Over 8 and Under 9 - a 9-run game wins the Over and pushes... at 8.5/9.5 both win on 9).
model score: The model's 0-100 rating of how strongly the data lines up for a side. Research, not a pick: a game only becomes an Official Pick when its score clears the model's bar and its other criteria are met.
moneyline: A bet on which team wins the game outright. No point spread involved.
no contest: A pick that never got a result - the game was postponed, suspended, or voided. It counts as neither a win nor a loss and is left out of the record.
no-vig line: The fair odds after removing the bookmaker's commission - the market's true probability estimate.
opponent-adjusted: A stat reweighted for the strength of the opponents actually faced, so a team that played four tough defenses doesn't read the same as one that played four weak ones.
OPS: On-base percentage plus slugging percentage - one number for how often a hitter gets on base and how much power he hits for.
OPS+: On-base Plus Slugging, adjusted for ballpark and era, where 100 is exactly league average. 120 means 20% better than average.
over/under: Whether the combined final score went over or under the posted total.
parlay: Multiple bets combined into one. Every leg must win. Payouts multiply but so does the risk - books profit most on parlays.
passer rating: The NFL's standard 0-158.3 quarterback rating, built from completion %, yards per attempt, touchdowns and interceptions per attempt. It ignores sacks and rushing.
passing downs: 2nd-and-8 or longer and 3rd/4th-and-5 or longer - situations where the defense knows a pass is likely. Success here shows how an offense handles being behind schedule.
PDO: 5-on-5 shooting % plus save %. League average is about 1.000, and teams far from it usually drift back - it mostly measures luck.
penalty kill %: The share of times shorthanded a team didn't allow a power-play goal (the NHL's official figure).
power play %: Power-play goals divided by power-play opportunities (the NHL's official figure).
power success rate: How often a short-yardage run (3rd or 4th and 2 or less, or at the goal line) gets the first down or touchdown.
PPA: Predicted Points Added - college football's version of EPA (from College Football Data): how much each play changed the offense's expected points, averaged per play. Above 0 is better than expected.
pressure rate: How often a quarterback is hit or sacked (faced) or a defense hits/sacks the opposing quarterback (generated), as a share of dropbacks.
prop: A bet on something other than the final result - usually a player stat like hits, strikeouts, or touchdowns.
push: A tie against the spread or total - your stake is refunded.
red zone TD rate: The share of trips inside the opponent's 20-yard line that end in a touchdown rather than a field goal or nothing.
returning production: The share of last season's production that is back on this year's roster. Context for how settled a team is, not a measure of quality.
ROI: Return on Investment - profit divided by total amount wagered. A long-term ROI above 5% is professional level.
run line: Baseball's version of a spread, almost always 1.5 runs. Backing the favorite on the run line means they must win by 2 or more; the underdog covers by losing by 1 or winning outright.
sack rate: Sacks as a share of dropbacks - how often a quarterback (or the line in front of him) gives one up.
sharp: A professional or winning bettor. Sharp money moves lines; books respect it.
SIERA: Skill-Interactive ERA - an estimate of how well a pitcher has actually thrown, based on strikeouts, walks, and contact quality rather than the runs that happened to score. Lower is better, and it predicts future performance more reliably than ERA.
SP+: Bill Connelly's opponent-adjusted efficiency rating for college teams: points better or worse than an average FBS team. Offense and defense have their own ratings; for defense, lower is better.
spread: A handicap that levels the matchup - the favorite must win by more than the spread; the underdog can lose by less (or win).
square: A casual/recreational bettor. Square money follows favorites, overs, and popular teams.
standard downs: First downs, 2nd-and-7 or shorter, and 3rd/4th-and-4 or shorter - situations where the offense can run or pass. Success on these shows how well an offense stays on schedule.
steam: A sudden, sharp line move across many books at once - usually triggered by professional (sharp) money.
streak: How many decided bets in a row have gone the same way, most recent first.
stuff rate: The share of runs stopped at or behind the line of scrimmage.
success rate: The share of plays that gained enough yardage to keep the offense 'on schedule' (roughly 40% of yards to go on 1st down, 60% on 2nd, 100% on 3rd/4th) - a steadier measure of consistency than EPA alone, since one huge play can inflate EPA without the offense actually staying on schedule.
talent composite: 247Sports' rating of the whole roster's recruiting talent. A higher number means a more talented roster on paper.
tier: The model's own label for how strong an assessment is (for example 'No Play'). Official Picks are the only graded calls.
time to throw: Average seconds from snap to release. Quick throws beat pressure; long holds create big plays and sacks.
total: A bet on the combined final score of both teams - Over or Under a line set by the book.
turnover margin: Takeaways minus giveaways, per game.
unit: Your standard bet size, usually 1-3% of bankroll. Measuring in units (not dollars) keeps discipline and makes records comparable.
velo trend: How a pitcher's recent fastball velocity compares with his season norm. A drop can signal fatigue or injury.
vig: The bookmaker's built-in commission (also called juice). A -110/-110 line has ~4.5% vig - you must win 52.4% just to break even.
WHIP: Walks plus hits per inning pitched. Lower is better; about 1.30 is average.
win rate: Wins divided by decided bets (pushes left out).
xERA: Expected ERA - what a pitcher's ERA should be based on the quality of contact he allows. A gap between xERA and real ERA usually means luck, good or bad, that tends to even out.
xFIP: Expected Fielding Independent Pitching - rates a pitcher only on what he controls (strikeouts, walks, home runs), ignoring how his defense played behind him. Lower is better.
xGF%: Expected goals share at 5-on-5: a team's expected goals divided by both teams' combined. Expected goals weight every shot by how often a shot like it scores, so it measures chance quality, not just volume. Above 50% means a team is out-chancing opponents.
xwOBA: Expected Weighted On-Base Average - how well a hitter or lineup should be producing based on how hard and at what angle they hit the ball, stripping out luck and defense.
xwOBA+: xwOBA allowed, adjusted for park and league to a 100 scale. 100 is average; below 100 means the pitcher is limiting quality contact better than average.
yards per attempt: Passing yards divided by pass attempts.