StatIQPregame to final — all in one place

Confidence pool

Rank every game in the week. A correct pick scores the number you put on it — your top pick is worth 13, your last is worth 1. Highest total that week wins the pool.

What this number is, and what it isn’t

Every probability below is the bookmaker moneyline with the vig removed — each book’s own two-way price de-vigged, then the median across the 9-odd books that priced both sides. It is the market’s number, not ours.

We are not claiming to beat the market. We tested a winner model against 2,680 real closing prices and the market won every year, so we stopped building one. What this tool does is the part the market doesn’t do for you: rank the slate, help you differentiate from the other people in your pool, and keep the score honestly.

Source: nfl_team_2024_oc.json @ T-1h. De-vig method: power. How we measure.

Your pool

Differentiation

If your entry matches someone else’s exactly, your scores are identical every week — an exact tie, not a close race. To finish ahead of them you have to differ. These are the two ways to do that, and they carry very different risk.

1 · Rank orderSmall lever

Move games up and down your board without changing who you pick. Against someone else the swing goes as (r − s)²: two slots apart on one game is worth 4.

0% off the price ordercosts 0.00 expected points
2 · Side flipsBig lever

Take the underdog. The swing goes as (r + s)²: on a game you both rank 8 that is 256 against 4 — 64× the variance from one decision.

0 underdogs of 13costs 0.00 expected points

A suggestion, not an optimum. The crossover from "maximise expected points" to "maximise variance" is not something we can state — it depends on your slate, your payout and how alike your pool is. Move the sliders and watch the win/tie split.

This entry

68.7
Expected points
of 91 possible
± 11.7
Typical spread
one standard deviation
0.0
Given up vs the default
default expects 68.7
15.1%
Win the pool
tie for first 3.5%
56.9%
Finish top three
average finish 3.5

Win and tie chances come from 1,500 simulated weeks against 11 sampled opponents. Those opponents come from a MODEL of how pools behave, described below — not from observed picks.

This is the consensus entry. Against anyone who submits the same thing your score is identical every week — mean 0, spread 0. That is an exact tie every week, decided by whatever tiebreak your pool uses, not by your picks. Move a knob to change that.

The slate

PtsPickGameWin prob BooksField est.vs defaultResult
13KCKC @ CAR85.3%988%~WON +13
12WASDAL @ WAS84.2%993%~LOST 0
11HOUTEN @ HOU78.4%989%~LOST 0
10MIANE @ MIA77.4%988%~WON +10
9DETDET @ IND76.8%978%~WON +9
8GBSF @ GB72.5%1084%~WON +8
7TBTB @ NYG71.3%971%~WON +7
6DENDEN @ LV70.1%1069%~WON +6
5PITPIT @ CLEprimetime64.8%967%~LOST 0
4MINMIN @ CHI61.3%957%~WON +4
3BALBAL @ LACprimetime59.9%1060%~WON +3
2PHIPHI @ LAprimetime59.3%1059%~WON +2
1ARIARI @ SEA52.1%1044%~LOST 0

Sign in to log this entry and have it graded when the games finish.

How it actually scored

62
This entry
of 91 attainable · 9-4
62
Honest default
of 91 attainable · 9-4
0
Difference
one week — nothing to conclude from it

This is 2024 week 12, a week that has already been played, priced at the odds an hour before kickoff from our own banked capture. One week is a coin flip’s worth of evidence about anything — it is here so you can check the arithmetic against games you remember, not as a result to read into.

Where the “field estimate” comes from

It is a model, and nobody has checked it against real picks — including us. Yahoo, ESPN and CBS all publish pick percentages inside their own products, and we could not confirm we are allowed to reuse them, so we have no observed data at all. What you see is a behavioural prior: crowds take favourites more often than the price says, over-back home teams, under-back road favourites, and pile onto the nationally televised game. The direction of those biases is well documented; the sizes are our assumption. Rows marked ~ are modelled.

Your own pool almost certainly shows you its pick distribution once picks lock. Paste it in and the model is switched off for those games — real numbers from your pool beat a general model of pools every time.

While you’re here: what a moneyline parlay costs

People who like ranking moneylines tend to get offered a parlay of them. The book’s margin on each leg compounds. At a standard −110/−110 market, before anyone is right or wrong about anything:

LegsExpected return
1-4.55%
2-8.88%
3-13.03%
4-16.98%
5-20.75%

Five legs is −20.75% against −4.55% on a single, purely from compounding the same margin. Real moneyline parlays are worse than this table: lopsided games carry more margin than −110, and same-game correlation is re-priced by the book, never by you. This is arithmetic, not a prediction — and a pool costs nothing to enter.

Confidence pool · StatIQ Sports