Founding lessons — The Wizard
Five lessons from the lab's founding research. They carry no picks.
Break-Even Math: Why 52.38% at −110 Is the Whole Game
At the standard price of −110, a picker who wins half the picks loses money. The price, not the pick, sets that result.
A −110 price means risking 110 to win 100. The break-even win rate is the amount risked divided by the amount risked plus the amount won:
110 ÷ (110 + 100) = 52.38%
A picker must win 52.38% of picks at −110 just to break even. At 50%, the picker falls 2.38 points short of that line on every 100 picks. The shortfall is the book's margin. It applies on every pick, win or lose, before any question of football knowledge enters the record.
The break-even line moves with the price. At −105, the break-even is 51.22%. At −120, it is 54.55%. The same picks, graded at a worse price, need a higher win rate to stand still. A record stated without its price states only half the result.
This is why a win rate alone does not grade a sheet. A 53% record at −110 clears the line by a small margin. A 53% record at −120 does not clear it at all. A 51% record at −105 falls short, and the same 51% at a still lower price would fall short by less. The reader needs both numbers — wins and the price paid — to know where a record stands.
It also sets the scale for any claim of skill. A picker a point or two above break-even over a short run has not separated skill from variance. The gap between 50% and 52.38% is small on paper and large over a season, because the margin is collected on every pick.
How to use this: grade any sheet by comparing its win rate to the break-even for the price it actually paid — 52.38% at −110, 51.22% at −105, 54.55% at −120 — not to 50%.
This lesson is AI-produced. The Sports Betting Sheet is a paper record; it places no wagers and is not betting advice.
What Closing Line Value Is
Closing line value, or CLV, measures the price a picker obtained against the closing line. A pick can lose the game and still have beaten the close. A pick can win the game and still have taken a worse price than the close offered.
The close is the market's final number before kickoff. On the bundled sample of 7,276 games from 1999–2025, the closing spread's average error was +0.09 points. The close missed high about as often as it missed low. That record is why the close serves as the anchor: it is a hard number to beat by picking winners alone.
This sheet scores every pick twice. The first score is the result: win, loss, or push against its closing line. The second score is the price: the price as filed against the closing line. For spreads and totals, CLV is recorded in points, and the price difference in cents. For moneylines, CLV is recorded in de-vigged probability percentage points.
The comparison close matters. CLV for this sheet is measured against the de-vigged sharp closing price, with the sharp reference close used where the line capture carries it. Comparing a pick to a vigged close, or to the picker's own book's close, flatters the record. The ledger states which close graded each pick.
Evidence from outside the NFL points the same way on price. An analysis of 87,960 Pinnacle football odds pairs found that beating the close by a given percentage was worth roughly that percentage in yield, with an almost one-to-one slope. That sample is soccer, not the NFL, and the lesson cites it as evidence about price, not about football outcomes.
CLV still needs a sample. This sheet does not characterize a market's CLV before about 200 graded picks in that market. Win-rate claims wait longer: distinguishing a genuine 54% picker from a coin flipper takes about 1,000 graded bets. Before those counts, the ledger prints numbers and pick counts and draws no conclusion line.
How to use this: grade any sheet by asking for both scores on every pick — the result and the price against the close — and ask how many graded picks stand behind any claim made from them.
This lesson is AI-produced. The Sports Betting Sheet is a paper record; it places no wagers and is not betting advice.
Key Numbers: 3, 6, 7, and 10
Not all points on a spread are equal. Games land on some margins far more often than others, and a half-point that crosses a common margin is worth more than a half-point that crosses a rare one.
The shares below come from the full margin distribution in 7,276 games from 1999–2025. A margin of 3 accounted for 15.1% of all games. A margin of 7 accounted for 9.0%. A margin of 6 accounted for 6.0%. A margin of 10 accounted for 5.5%. Margins of 3 and 7 together accounted for 24.1% of all games.
The order has shifted. Margins of 7 fell from 9.9% in 1999–2004 to 8.1% in 2023–25, while margins of 6 rose from 4.9% to 6.4% over the same split. On the full sample, 6 has passed 10 as the third key number. Tables built on older seasons overvalue 7 and undervalue 6.
The value of a half-point is conditional. It depends on the number it crosses, not on the pooled share. When the spread closes at exactly 3, the game pushes 9.0% of the time. Moving from −3 to −2.5 converts that push to a win 9.0% of the time, worth about 8 cents per $1 at −110. At exactly 7, the push rate is 6.5%, worth about 6 cents.
Books charge more than that value to sell the move. Books typically charge 20–25 cents to cross the 3. A buyer pays 20–25 cents for an asset worth about 8 cents. The arithmetic runs the other way when the better number is available at the standard price: taking −2.5 instead of −3, at no extra cost, captures the full value.
Totals have no equivalent concentration. Common totals cluster loosely, without the 3 and 7 concentration found in margins. That difference is part of why spread numbers carry this lesson and totals numbers do not carry it in the same way.
How to use this: grade any sheet by the numbers its picks were taken at — a pick at −2.5 and a pick at −3.5 in the same game are different picks — and treat a paid half-point through 3 as a cost to justify, not a routine purchase.
This lesson is AI-produced. The Sports Betting Sheet is a paper record; it places no wagers and is not betting advice.
Why Teasers Died at the Current Price
A six-point teaser moves each leg by six points. The legs that historically held up cross both 3 and 7: favorites from −7.5 to −8.5 move to −1.5 to −2.5, and underdogs from +1.5 to +2.5 move to +7.5 to +8.5. Those are the Wong windows.
The legs still land at their historical rates. Measured on closing lines from 1999–2025, Wong favorite legs won 72.8% of the time, on a sample of 493 legs. Wong underdog legs won 75.2% of the time, on a sample of 904 legs. Legs outside the Wong windows won 62.5–68.2% across the groups tested, and were never profitable at any price in that record.
A two-team teaser needs both legs to win. The per-leg rate needed to break even is therefore higher than either leg's headline rate suggests. At a −110 teaser price, each leg must win 72.4%. At −120, each leg must win 73.9%. At −130, each leg must win 75.2%. At −135, each leg must win 75.8%.
Read the two tables against each other. At −120, the underdog rate of 75.2% clears the 73.9% requirement, and the favorite rate of 72.8% does not. A dogs-only pair at −120 is roughly break-even on historical rates. At the top of the table, no Wong combination clears the requirement: the underdog rate of 75.2% falls short of 75.8%, and the favorite rate falls short by more.
Current pricing sits at that top end for one book. A dated price check, updated 14 Oct 2026, listed two-team six-point teasers at −120 at DraftKings and Caesars, −120/−130 at BetMGM, −125 at Circa, and −134 at FanDuel. At FanDuel's −134, no Wong combination clears break-even on the historical rates above.
The football did not change. In the same record, 38.6% of games still land on margins of 3 through 7. The price changed, and the price is what removed the margin the legs used to carry.
How to use this: grade any teaser pick by its ticket price first — state the per-leg rate that price requires, then compare it to the rate the legs have actually produced, with the sample size attached.
This lesson is AI-produced. The Sports Betting Sheet is a paper record; it places no wagers and is not betting advice.
Line Shopping Is Arithmetic
Line shopping does not require a better forecast. Two books offer different prices on the same game, and taking the better price changes the break-even line by a fixed amount. The gain is in the arithmetic, not in a read on the teams.
Start with the price alone. Taking −105 instead of −110 on the same pick lowers the break-even from 52.38% to 51.22%. That is a difference of 1.16 points of win rate, obtained without picking any better.
A worked example in the research record shows how the difference compounds. A 53% bettor placing 1,000 bets of $100 earns $1,300 at −110, for a return of 1.30%. At an average price of −107, the same bettor earns $2,710, for a return of 2.71%. A three-cent improvement in average price more than doubles the profit in that example, on identical picks.
The number matters as well as the price. Across books, NFL sides typically differ by 0.5–1.0 points from best to worst, with a maximum observed difference of 2.0 points. Totals typically differ by 0.5–1.5 points, and moneylines by 5–15 cents. A separate tracking count of 14,247 games across 10 US books from 2022–24 found a mean best-to-worst spread gap of 0.47 points. That count is blog-grade evidence, and the lesson treats its decimal as indicative, not exact.
Around a key number, a shopped number can outweigh a shopped price. Getting −2.5 instead of −3.5 spans the full 9% push mass at 3, an 8% win-probability swing in the research record. No change in the pick produces that swing. Only the book offering the number produces it.
Practitioner sources estimate the value of consistent shopping at about +1.5 to +3.0 points of return per year, against single-book betting. The operational rule in the research record is direct: check all eight New York books on every pick, and record the best obtainable price and the consensus price.
How to use this: grade any sheet by the prices and numbers printed beside its picks — a sheet that records only the pick, and not the book and price it was taken at, cannot be graded on the part of the record shopping controls.
This lesson is AI-produced. The Sports Betting Sheet is a paper record; it places no wagers and is not betting advice.
