Glossary

Simulated games

The sims play each game many times: 20,000 for MLB, 10,000 for NFL. Every chance they show is the share of those games where it happened.

Typical

The middle result of the simulated games. Half land above it, half below.

Middle half

The range the middle 50% of simulated outcomes land in.

Likely range

The range 8 in 10 simulated outcomes land in.

Hold

The book's built-in cut, also called vig or juice. The two sides of a line add up to more than 100%. The extra is the hold.

Market %

The chance the sportsbook prices imply, with the hold taken out and combined across books.

Edge

How far the model's chance sits from the market's, in percentage points. ▲ means the model rates that side higher than the books do.

Fair price

The model's chance written as odds, with no hold. A book price better than fair is in your favor.

Best price

The best odds any tracked sportsbook offers on that side right now.

Outside model range

The model and the market are more than 12 points apart in MLB, or 6 on NFL game lines and 18 on NFL props. A held market's own row carries a mark (tap it for details): a dashed yellow ring is Pending, a red slashed circle is Held. Once approved, a market shows normally with no mark. MLB has no review: a yellow dot there means shown with a warning. Numbers built from that market show plain, or a dash when held.

Pending

A market outside model range waiting on review. Its model numbers are hidden until then, and free views always hide them.

Held

An NFL market outside model range that review held back for this slate. Prices still show; model numbers do not.

Free game

One game each slate, open to everyone in full. MLB: earliest national-TV game. NFL: Thursday, else the earliest national window.

Projected lineup

The lineup the MLB sim uses before the real one posts: the team's last lineup against a starter of the same hand.

Unusual here

Where tonight's sim breaks from what a team has actually been doing, against its own Season, L30, L14 or L7 baseline.

Outliers

Players whose projection tonight sits well above or below their own recent form.

L30 / L14 / L7

The last 30, 14 or 7 games.

Win Probability

The Dafty model's chance a team wins, from its predicted run difference.

Tier

Max, Strong, Moderate or Lean. Moneyline tiers come from win probability. Prop tiers come from expected value. See Confidence Tiers below.

Unit

1% of bankroll. Sizes are shown in units (u), rounded to quarter units.

Kelly Size

A stake sized to the edge and the odds. MLB picks and props use one-eighth Kelly, capped at 5u.

EV

Expected value: average return per unit if the model's chance is right. +5% is 5 cents per dollar.

Model Score

A 0–100 trust check from similar past games. Strong 70+, Fair 45+, Weak below.

What's driving it

The five inputs that moved a pick's win probability most. Bars to the right push it up. Bars to the left pull it down.

Stale

Odds more than 45 minutes old.

Lock

An MLB game locks 45 minutes before first pitch. The price and size at lock are what Results grades.

Push

A bet that lands exactly on the line. The stake comes back and counts as neither a win nor a loss.

P&L

Profit or loss in units, at the price taken.

ROI

Return on investment: P&L divided by the units staked.

Moneyline

A bet on who wins.

Run line

Baseball's spread, usually 1.5 runs. The favorite at −1.5 has to win by 2 or more.

Spread

Football's handicap. The favorite has to win by more than the number to cover.

Total

Over or under a posted number of combined runs or points. Also called the O/U.

F5

First five innings only. A tie after five is a push on the moneyline.

NRFI / YRFI

No run in the 1st inning, or a run in the 1st, by either team.

Player prop

A bet on one player's stat line, like strikeouts, total bases or receiving yards.

One-way

A prop the book prices on one side only. Its hold is estimated from that book's two-sided lines.

ERA

Earned Run Average: earned runs allowed per 9 innings. Lower is better.

FIP

Fielding Independent Pitching: an ERA-style number built only from strikeouts, walks, hit batters and home runs. Strips out defense and luck.

WHIP

Walks plus hits per inning pitched. Lower is better.

K%

Strikeout rate: share of plate appearances that end in a strikeout.

BB%

Walk rate: share of plate appearances that end in a walk.

K–BB%

Strikeout rate minus walk rate. One of the best quick reads on a pitcher.

Whiff%

Share of swings that miss. Higher is better for a pitcher.

Barrel%

Share of batted balls hit at the speed and angle that most often go for extra bases. Lower is better for a pitcher.

Avg EV

Average exit velocity: how hard the ball comes off the bat, in mph.

Hard-Hit%

Share of batted balls hit 95 mph or harder.

GB%

Share of batted balls hit on the ground.

Stuff+

Our pitch-quality grade, built from Statcast run value per pitch. 100 is average. Higher means nastier stuff.

IVB / HB

Induced vertical break and horizontal break: how far a pitch moves, in inches, compared with a pitch thrown without spin.

Velo Trend

Recent fastball velocity against his season baseline. Positive means he's throwing harder.

Season IP

Innings pitched this season. Context for workload and sample size.

Leverage

How much a relief spot matters. High leverage is the 7th inning on with the game within 3 runs.

Exposure

The chance each reliever pitches tonight, in the sim.

AVG / OBP / SLG

Batting average, on-base percentage and slugging percentage. On a game page they are this matchup's projection.

ISO

Isolated power: slugging minus average. Measures extra-base pop.

xwOBA

Expected wOBA: a hitter's value from exit speed and angle, not where the ball lands. Lower favors the pitcher.

xBA

Expected batting average from the quality of contact.

OAA

Outs Above Average: outs a defense makes beyond what an average fielder would, from Statcast. Higher is better.

Catcher framing

How often a catcher gets borderline pitches called strikes.

Platoon

How a hitter or pitcher does against the same hand or the opposite hand.

Log5

A way to combine a hitter's rate and a pitcher's rate into one rate for their matchup. The Pitches tab uses it pitch by pitch.

Threat

The hitter with the best matchup against tonight's starter, ranked by matchup xwOBA.

Pitcher edge / Hitter edge

Which side a matchup read favors: the starter or the batter.

OVR

Overall rating, 0 to 100 across MLB. A team's blends offense 35%, starting pitching 30%, bullpen 20% and defense 15%.

Rating tiers

Elite 90+, Great 80+, Above Average 70+, Average 60+, Below Average 50+, Struggling under 50.

Talent

A player's true-talent estimate, pulled toward average until he has enough sample to stand on his own.

Form

How a player has done over his recent stretch. Ratings lean more on form as the season goes on.

Signals

Rule-based reads on a game from Statcast data, built once both lineups are confirmed. They sit beside the model and don't feed it.

Signal strength

++ strong positive, + moderate positive, −− and − negative, ◆◆ and ◆ context. Context reads describe the park, weather or umpire and favor neither side.

Reliability

On skill reads, R is that tier's average effect against the rule's baseline, and n is how many past firings it rests on.

Trend Ledger

Each side's records straight up, on the run line, on the total and in the first five, measured against each record's own baseline. Context, not picks.

EPA

Expected points added: how much a play changed a team's expected points. The base of every NFL team rating.

Game Script

How the NFL sim expects a game to unfold: its shape, which way it tilts, how drives end, pace, and what happens if the script breaks.

Matchup anatomy

Each team's simulated rates next to the same rates against an average defense.

If the script breaks

The final margin and total for each halftime score the sim reaches often enough to read.

Explosive plays

Gains of 20 yards or more. In the matchup anatomy, explosive runs are carries of 10 yards or more.

Cover, push, miss

How often each side covers the posted spread, and how often the margin lands exactly on it.

Key numbers

Margins NFL games land on most often, led by 3 and 7. A spread on a key number pushes more often.

vs avg defense

A player's projection with the opponent's defense set to league average, so you can see what this matchup adds or takes away.

Softness

Ranks the 32 defenses 0 to 100 by EPA allowed per opportunity to a position. Higher means more given up.

Role stability

A player's offensive snap share by week. A role change shows here before the box score.

Touchdown scorer

Anytime and first touchdown lines. Books post the Yes side only, so the site shows Yes only.

Flagged

A Ledger line 3+ points from the market at 2+ books. Moneylines, spreads and TD scorers are never flagged.

Paper ledger

The NFL Ledger records what the sim flagged and at what price. Nothing in it is placed.

Bets are sized by the edge and the odds, not the tier. MLB uses one-eighth Kelly, capped at 5u. Lean and Moderate dogs get half. Under 0.1u is no bet.

Moneyline picks

A Dafty model pick's tier comes from its win probability for that side. Higher tiers mean stronger conviction.

Max

Win probability of 72% or more. The model's rarest and strongest picks.

Strong

65% to 72%.

Moderate

58% to 65%.

Lean

52.4% to 58%. Just past the break-even rate at standard −110 juice.

Dog vs. Favorite

Lean and Moderate picks split by price. A pick at +100 or longer is a Dog; anything shorter is a Favorite. Dogs get half stake.

Player props

Prop tiers come from expected value at the best two-sided sportsbook price. No tier inside sim noise, under two books, or at −250 or shorter.

Props Max

Expected value of 12% or more.

Props Strong

9% to 12%.

Props Moderate

6% to 9%.

Props Lean

3% to 6%.

Totals

Over/under picks return for the 2027 season. Until then totals show the model's number for reference, with no stake.