How scoring works
Anyone can be right about the favorite. The Take Makers score rewards being right when it was hard to be, and costs the most for missing what everyone saw coming.
1. What counts: on the board
Every prediction from an episode is pulled out as a take. But a take only counts if a sportsbook or a prediction market (Vegas, DraftKings, FanDuel, Kalshi, Polymarket) would plausibly let you bet on it: a side, a total, a player prop, a season future, an award, a coach getting fired.
Injury and lineup reports, “the defense plays well,” “he could become one of the best”: those are off the board. They stay on the episode and still get graded, but they don’t count toward records, heat or score. Each counted take shows its market: Spread, Moneyline, Total, Player prop, Future, Award, Event.
2. Heat: how bold it was when said
Every take gets a heat score from 0 to 10, to one decimal, judged at the moment it was made and never with hindsight. 5.0 is a coin flip; lower means most people expected it; higher means it went against the grain. In order, heat comes from:
- Odds said on air. If the host quotes a price (“Coleman anytime touchdown at +240”), that price sets the heat.
- The betting market. For game picks, the closing moneyline, spread and total for that game.
- Published odds. For futures, awards and player props, the odds a sportsbook posted on or before the day it was said, found on the web.
- An informed estimate. If nobody priced it, a model judges its chance from that day’s records, injuries and news.
- A coin flip. If there’s nothing reliable to go on, it’s 5.0.
Odds always have the sportsbook’s cut taken out, and nothing from after the take was said is ever used.
3. Firmness: how clearly it was called
“The Rams cover −3.5” is firm. “I could see the Rams winning by a touchdown, maybe” isn’t. Firmness (0–100) measures how clearly the take was stated and how cleanly it settles. It’s shown on its own and doesn’t change the points.
- 90–100 · Settles cleanly from a box score, standings, or a stat line
- 65–89 · Clear outcome, but there’s a hedge, a range, a condition, or a long wait
- 40–64 · You have to pick a stat to stand in for words like “suck” or “really good”
- 0–39 · Vibes, “should,” conditionals, or things with no record to check
4. Grading
Once its game or season is over, each take is settled: correct, wrong, or push (exactly on the number). Game picks settle straight from ESPN box scores; everything else is checked against reported results with the source linked. A take with several legs (“the Bills and the Lions both win”) is correct only if every leg is. Admins can review and reverse any grade.
5. The take score
Heat and the result make one number per take, from 0 to 100. It works like a bet’s payout: heat is 100 minus the take’s chance, so a long shot pays a lot and a sure thing pays little. A wrong take scores 0; nothing is taken away.
Correct: + heat × 10
Wrong: 0
Push: 0
So a bold call that comes true earns the most:
Chalk (heat 1.5)
Hot (heat 8.0)
Correct
+15
Saw it coming
Everyone had it. A little credit.
+80
The call
Went against the grain and was right. The most points.
Wrong
0
Blew the gimme
Missed what everyone saw coming. Nothing.
0
Swung and missed
Bold and wrong. Nothing, but nothing lost for taking the shot.
6. Points and the rating
A taker’s points (Pts) are the sum of their take scores. Points reward volume: someone who talks a lot can pile them up. Standings list the most correct takes first; the rating weighs everything at once, one whole number from 0 to 100, where 50 is average, built from every stat:
| Part | Weight | What it measures |
|---|---|---|
| Points per take | 45% | How many points each call earns, on average |
| Win rate | 25% | Correct out of correct + wrong |
| Boldness | 15% | Average heat: how far from consensus they go |
| Volume | 15% | How many graded takes they’ve put on the record |
Small samples are pulled toward average: every taker starts as if they had a few break-even calls, so two lucky picks barely move the rating while thirty good ones move it a lot. A rating also needs a track record: until a taker has 10 graded takes, their rating is pulled toward 40, below average, and rises to its full value as they reach 10.
Everything counts graded, on-the-board takes only. A pick repeated on another episode or show counts once; a take shared by two hosts counts for each. Tables for a single day, game or episode are too small to rate anyone, so they show no rating (a multi-leg take counts on the day of its last leg). Click any column to sort by it. See the standings →
7. Your calls: tail or fade
Signed in, you can tail a take (you think it’s right) or fade it (you think it’s wrong) until its game kicks off. Every call is timestamped. Tailing a take that hits or fading one that misses is a win; your record lives on your page.