Every metric, statistic, and betting term used on Havocrate, defined in plain English. CLV, Brier, PPA, EPA, success rate, Kelly stake, juice — all here without jargon.
The final betting line a sportsbook posts before kickoff. The closing line is the sharpest public price because it reflects all the money bet up to that moment.
Example
If Texas opens as a 2-point favorite over Georgia and the line moves to Texas -3.5 by Saturday morning, then Texas -3.5 is the closing line.
How much better your bet's number was than the closing line, in points or in cents. Positive CLV means you got a better number than the market consensus. The single best metric for evaluating long-term betting skill.
Example
You bet Georgia +3 at 11 AM Saturday. By kickoff the line closes at Georgia +1.5. You beat the closing line by +1.5 points — that's +CLV.
The sportsbook's margin built into a bet's price. Standard juice on spreads is -110, meaning you risk $110 to win $100. Books take ~5% of action as commission.
Example
On a -110 spread, you need to win 52.4% of bets just to break even after juice.
The long-run average outcome of a bet, expressed as a percentage of stake. Positive EV means you're projected to profit on average; negative EV means you'll lose money over time.
Example
A bet with +5% EV at $100 stake returns +$5 on average per bet over many trials.
A formula for sizing bets that maximizes long-term bankroll growth. Full Kelly is mathematically optimal but high-variance. Half Kelly and quarter Kelly trade some growth for less drawdown.
Example
With a +9.8% EV bet at -110 juice, full Kelly says risk 10.8% of bankroll. Quarter Kelly says risk 2.7% — much smoother variance, slightly slower compounding.
Bets placed by professional or high-volume bettors with documented edge. When sharp money piles onto one side, sportsbooks move the line quickly. The opposite is 'public money' or 'square money.'
Example
If 70% of public bets are on Team A but the line moves toward Team B, sharp money is on Team B.
A rapid, simultaneous line move across multiple sportsbooks. Almost always triggered by one big sharp player placing a bet at a respected book, then other books follow to stay in line.
Example
Texas moves from -2.5 to -3.5 across DraftKings, FanDuel, and BetMGM within 90 seconds. That's steam.
When the line moves opposite to where the public is betting. Strong indicator that sharp money is on the other side.
Example
70% of bets are on Alabama -7, but the line moves to Alabama -6.5. The book is signaling sharp money likes the underdog.
A non-standard spread or total offered at adjusted odds. Buying off the key numbers (3 and 7) often unlocks +EV at standard pricing the main line doesn't have.
Example
Main line: Georgia -2.5 (-110). Alt: Georgia -0.5 (+115) — you're crossing the key number 1 toward zero for plus money.
An offshore sportsbook based in Curacao, considered the sharpest book in the world. Pinnacle accepts professional bettors without limiting them, so its line is the closest public proxy for true market price.
Example
When Havocrate measures CLV, we compare to Pinnacle's closing line as the reference.
Las Vegas sportsbook known for opening the earliest college football lines each week and welcoming sharp action. Often the first US-licensed book to post a CFB number.
Example
Circa's Sunday opener is one of the most sharply-priced US lines available to non-offshore bettors.
Bill Connelly's opponent-adjusted college football rating system. Combines offensive, defensive, and special-teams efficiency with strength of schedule. Published weekly during the season. Higher is better; 0 represents an average FBS team.
Example
Indiana's 2025 SP+ of 32.4 ranks #1 nationally — top tier offense and elite defense.
ESPN's Football Power Index. Predictive power rating that estimates how many points better than average a team would be against a typical FBS opponent. Updates weekly.
Example
An FPI of +18 means the team would beat an average FBS team by ~18 points on a neutral field.
A continuous power index originally invented for chess. Each game's outcome and margin updates each team's Elo score relative to expectations. Self-correcting over time.
Example
A team that starts at Elo 1500 and wins by 14 against a 1600 opponent moves up — the system pays more for upsets.
Simple Rating System. Margin of victory per game adjusted for schedule strength. Older, simpler than SP+ or FPI but still informative.
Example
An SRS of +12 means the team beats an average team by 12 points after adjusting for who they played.
Predicted Points Added (sometimes Expected Points Added). The change in expected points before vs after a single play. The foundation of modern football analytics.
Example
A 50-yard pass from your own 25 to the opponent's 25 has a much higher PPA than a 5-yard gain on third-and-1.
Percentage of plays that 'succeed' — 50% of needed yards on first down, 70% on second down, 100% on third or fourth down. A volume measure of consistency.
Example
A 50% success rate is excellent. Elite offenses run 55-60%.
Average PPA on successful plays only. A team can be consistent (high success rate) without being explosive (lots of 20-yard gains), or vice versa.
Example
Texas Tech 2025 explosiveness was high because their successful plays averaged big gains, even if they failed often.
Percent of opponent plays that result in a tackle for loss, forced fumble, interception, or pass break-up. Measures defensive disruption.
Example
Georgia's 2025 havoc rate of 24.1% led the country — they got hands on a lot of plays.
A modeling technique that runs a game outcome thousands of times under randomized conditions to produce a distribution of likely results. Havocrate runs 10,000 simulations per game.
Example
Of 10,000 simulated UGA vs TEX games, UGA wins 58% and covers the spread in 58% of outcomes.
A calibration metric for probabilistic predictions, ranging from 0 (perfect) to 1 (worst). Measures how close your predicted probabilities are to actual outcomes.
Example
A Brier of 0.241 means when you say 60%, you're roughly right. Lower = better calibrated.
How well predicted probabilities match observed frequencies. A well-calibrated model that says 70% should hit at roughly 70% over many trials.
Example
If your model says 60% twenty different times, 12 of those games should hit if you're calibrated.
The range of values within which the true result is expected to fall, given the model's uncertainty. 80% CI means 8 of 10 simulated outcomes land in this range; 95% CI means 19 of 20.
Example
UGA's 80% confidence interval is −10 to +4 — most outcomes are within that spread.
The full shape of likely outcomes from a simulation, expressed as the frequency of each margin bucket. A way to show variance beyond a single point estimate.
Example
If UGA wins by 3-7 in 28% of simulations, that bucket dominates the distribution.
The number of observations a metric is based on. Small samples are statistically noisy; large samples are more reliable.
Example
A 54.6% positive-CLV rate over 348 bets has a ±2.7 percentage point margin of error at one standard deviation.
The expected variability in a statistic due to random sampling. Smaller SE = more reliable estimate.
Example
An ATS record of 55-45 (55%) over 100 games has SE of ~5 points — you'd need many more games to know if 55% is real or luck.
Football Bowl Subdivision — the top tier of NCAA Division I football. 134-138 teams across 11 conferences. The level Havocrate covers.
Example
Alabama, Ohio State, and Boise State are all FBS programs.
Football Championship Subdivision — the second tier of Division I football. Smaller scholarships, playoff system instead of bowls.
Example
North Dakota State and James Madison (before its FBS move) are FCS programs.
The four richest FBS conferences: SEC, Big Ten, ACC, and Big 12. Replaced 'Power 5' after the Pac-12's 2024 collapse.
Example
Texas, Oklahoma, Alabama, Ohio State are all P4 programs.
The five non-Power-4 FBS conferences: American Athletic, Mountain West, Sun Belt, Mid-American, Conference USA. Smaller budgets, smaller TV deals.
Example
Boise State and Memphis are G5 programs that often punch above their weight.
The NCAA system that lets college athletes change schools without sitting out a year. Massively changed roster building since 2021.
Example
Carson Beck entered the portal after the 2024 Georgia season and landed at Miami in 2025.
Name, Image, and Likeness — the right of college athletes to earn money from endorsement deals, signed in 2021. Reshaped recruiting and retention.
Example
A 5-star QB might command a $1M+ NIL package to attend a top program.
Percent of a team's prior-season production (snaps, targets, yards) that returns to the roster the following year. Strong predictor of preseason rankings.
Example
A team returning 80% of their PPA is in much better shape than one returning 30%.
The area from the opponent's 20-yard line to the goal line. Red zone efficiency (TD% on red zone trips) is a key offensive metric.
Example
A 70% red zone TD rate is elite; 50% is average.
Late-game minutes where the outcome is no longer in doubt, and stats get inflated by both teams playing backups. Most modern analytics filter garbage time out.
Example
A 38-3 game in the fourth quarter — anything after that point is garbage time.
Have a term we missed? Let us know. We add definitions as questions come in. Statistical definitions derive from CollegeFootballData.com's glossary and Bill Connelly's published methodology for SP+.