Methodology
The rules are the product. Every show is graded the same way, from the same 1-unit stake, so the numbers on this site can be compared without knowing anyone’s bankroll.
How a pick is graded
Six rules cover almost every pick made on air.
- Every named play risks 1 unit, whatever the host says they are putting on it.
- A win pays by the American price. A positive price p pays p / 100; a negative price pays 100 / |p|. The result is rounded to two decimals.
- A loss is -1.00. A push or a void is 0.00 and is left out of the record and out of ROI.
- Each rung of a ladder is its own play. A three-rung ladder risks three units, not one.
- Prices are as quoted on air. Where no price was given we use -110, and any price we had to estimate is flagged est. in the picks table.
- Leans, “if forced” picks and passing mentions are not picks and are left out.
Worked examples
The same six rules, on real week 1 picks.
| Pick | Price | Result | Units |
|---|---|---|---|
| Jack Bech anytime TD | +800 | Win | +8.00 |
| Steelers moneyline | -178 | Win | +0.56 |
| Giants +3.5 | -110 | Win | +0.91 |
| Rams -3.5 | -110 | Loss | -1.00 |
| Bears -3 (landed on 3) | -110 | Push | 0.00 |
| Scratched player prop | -110 | Void | 0.00 |
Records, ROI and CLV
Three numbers do most of the work on this site.
- The record counts wins and losses only. Pushes and voids are shown as a third number where they exist, and pending picks are not counted at all.
- ROI is units divided by graded plays, where graded plays are wins plus losses. Units are rounded per play and then summed, so a table always adds up to the picks under it.
- CLV compares the number in the pick to the DraftKings closing line. A plus counts where the line moved toward the show’s side before kickoff, a minus where it moved away. It is filled in only where a closing line is on record, so the plus and minus counts are usually smaller than the number of picks.
Picks are graded independently of each show’s self-reported results. Where a show grades its own card differently, we keep our grade and note the difference in that week’s notes.
Luck or skill
Whether a run is bigger than the swing the prices alone produce.
- Every price carries an implied probability, vig included: a negative price p implies |p| / (|p| + 100), a positive one implies 100 / (p + 100). At -110 that is 52.4%.
- A bettor with no edge at all wins at exactly that rate, so each 1-unit play has a mean of 0 units and a variance of q(1 - q)(w + 1)², where q is the implied probability and w is what a win pays.
- Variances add, so a whole card has a standard deviation of the square root of the sum. Nine no-edge bettors in ten land within 1.645 standard deviations of zero, and that is the no-edge range printed on every show and host page.
- z is how many standard deviations the actual units sit above or below zero. Under 1 we call it within noise, between 1 and 2 leaning, and 2 or more a clear edge.
Three caveats. The vig means a no-edge bettor does not really break even: at -110 they lose about 4.5 cents per unit staked, so zero is a slightly generous bar. We print no label at all under 15 graded picks, which is why Gallant at 5-3 and +16.40 in week 1 reads as too early rather than as a find. And the normal approximation behind the range is weakest at long prices, so a card built on anytime touchdown books is the one to trust least.
Sides versus prices
How much of a total came from the picks and how much from the numbers.
Every graded result is regraded a second time at a flat -110, so a win is +0.91 and a loss is -1.00 whatever was quoted on air. The gap between the quoted total and the flat total is the price edge.
- In week 1 the Anytime TD Show is +18.53 units at the quoted prices and -7.99 at a flat -110, because 11 of its 28 plus-money plays landed. The picks were ordinary; the prices were the story.
- The edge runs the other way too. Even Money shows a negative edge because a teaser winner at -125 pays 0.80 rather than the 0.91 the same win is worth at a flat price.
- Prices we had to estimate are counted in the totals but also listed on their own, so a reader can see how much of a price edge rests on a number nobody said out loud.
The consensus board
Where two or more shows landed on the same position in the same week.
- Positions are read out of the pick text. A side groups by team and ignores the number, so Eagles -5.5 and Eagles -4.5 are one position with both numbers listed against it.
- A show counts once on a position even when two of its hosts called it. The extra plays are shown as in-house repeats, so the units still add up.
- The fade column is what the other side of the same plays would have paid at a flat -110. It is a hypothetical, not a bet anyone made, and it ignores the price a fader would really have been offered.
The board is built by reading strings, so it is the least certain thing on the site. Picks it cannot read are counted in the footnote under the board rather than dropped. Ladders, props, parlays and teasers are out of scope and are counted there too. Positions are never matched across bet types, so a Malik Willis anytime touchdown and a Willis anytime rushing touchdown prop stay apart even though the same thing has to happen for both. Disagreements are limited to totals: a side or a moneyline names one team, so without a schedule we cannot tell that two shows took opposite sides of one game.
How a week is published
One week at a time, once the last game is final.
A week is graded on the Tuesday after Monday Night Football, checked against each show’s own recap, and then published. Nothing appears on the site until the whole week is published, so a partly graded week is never visible. Pending picks — a futures ticket, say — stay at zero units until they settle.
Report a correction
Wrong price, missed pick, mis-graded result: please say so.
Email corrections@podpickstracker.com with the show, the week and the pick. Every correction we make is logged in that week’s notes, so the change is visible on the week page rather than applied quietly.