How DiamondEdge researches NFL player props

A weekly board, rebuilt until kickoff

The board for a new week is built on Tuesday morning (11:00 UTC, which is 7 AM Eastern during daylight time) and rebuilt on the same schedule Wednesday through Saturday, with a second pass at 21:00 UTC (5 PM Eastern) Tuesday through Friday and a final pass on Sunday at 14:00 UTC (10 AM Eastern). Each rebuild picks up the latest injury designations, practice participation and depth charts.

A rebuild only touches games that have not kicked off. Once a game starts, its cards are frozen exactly as they were shown, and that frozen version is what gets graded. Days whose games are already over drop off the board, so by Sunday night only Monday's game is left.

The line is an N+ target set from the player's own data

Every card carries a DiamondEdge line, which is a threshold DiamondEdge sets from the player's own game log so recent games can be measured against one consistent number. It is not a sportsbook line and it carries no price. The card shows it as a target such as 62+ rushing yards or 5+ receptions, where the number is the line when the line is whole and the line rounded up when it is not.

Props that are too obvious to research never reach the board: receptions below 2+, receiving or rushing yards below 15+, rushing plus receiving yards below 20+, and passing yards below 100+ are left out. Touchdown markets start at 1+, which is a real question.

What the rating is built from

The 1 to 5 diamond rating uses only the card's published research, so the number you see is the same one the server computed. It starts from how often the player reached this exact line in their last 10 decided games: 90 percent or better starts at 4, 80 percent at 3.5, 70 percent at 3, 60 percent at 2.5, 50 percent at 2, and anything lower at 1.

The projection then moves it: a projection above the line adds a diamond, a projection below the line takes one away. A clean last 5 (all five most recent games at or above the line) adds half a diamond. The result is rounded to a whole diamond and capped by data confidence: limited data cannot rate above 3, standard above 4, high above 5. A player with fewer than 4 decided games gets no rating at all rather than a guess.

Recent-game clears, shown one game at a time

• A clear is a recent game in which the player's stat was at or above the DiamondEdge line. The card shows the individual game values, newest on the right, and which of them cleared.
• The rating reads the last 10 decided games; a game with no usable value is skipped rather than counted either way.
• When 20 games are stored, the card and its sheet offer an L10 / L20 switch so you can see whether a short window agrees with a longer one.
• A clear count is context about what already happened. It is not a probability for the next game.

Defense vs position and game environment

The write-up brings in matchup context that the rating itself does not use. For NFL, the opposing defense is measured by what it has allowed per game this season to the relevant position: passing yards to quarterbacks, rushing yards to quarterbacks and running backs, receiving yards and receptions to backs, receivers and tight ends. Once enough defenses have played enough games to rank them, a defense in the most generous quarter shows up as a supporting signal and one in the stingiest quarter shows up as a limiting signal labeled a tough matchup. Touchdown lines do not get a defense read.

Game environment covers the venue roof (dome, retractable or outdoors), each team's season points scored and allowed from final scores, and, for open-air venues inside the forecast window, the National Weather Service kickoff forecast. Wind of 15 mph or more, a 50 percent or higher chance of rain, or a temperature at or below 32F is flagged as a caution for passing and receiving props. Point spreads and game totals are deliberately not used anywhere in the research.

Confidence labels and the case against the prop

Every card carries a data-confidence label: high, standard, limited, or line unavailable, with a short reason. The label describes how much usable data sits behind the card and how stable it is. It caps the rating, but it is not a win probability.

Each card also carries a plain-English 'Why this could miss' line built from the card's own data: recent misses, a cold last 3, close calls, a thin sample, or an unsettled status. Pro readers also see the backend's cautions and a note when the projection sits below the line.

Voids and grading

Every free pick and every Top Pick (4 diamonds or higher) is graded against the final box score the morning after its game, using the same N+ target the card showed. If the player's stat reached the target it is a win, otherwise a loss. Because the target is always a whole number, there is no push in football.

A void is a card whose player did not record a stat line for that game, such as a player who was inactive. Voids are counted separately on Results and are never treated as a win or a loss. A card dropped from the board before its game day, for example a player ruled out mid-week, is not graded at all.

Free versus Pro on the NFL board

• Free viewers get three free picks a week: the top-rated cards on the board, never the same player twice, spread across different games when the slate allows. Those three show their rating, projection, line read and summary.
• Every other card shows its evidence to free viewers (the line, recent game values and clears, confidence, and signals) but the rating, projection, line read and summary are held back, and locked cards are ordered by recent hit rate rather than rating.
• Free viewers see how many props on the board rate 4 diamonds or higher, but not which ones. Pro opens every rating, all Top Picks, and the full research breakdown on each card.
• The full breakdown covers a summary, why the card surfaced, the statistical case, recent form with sample size, role and usage, the matchup, supporting signals, risk factors and a research conclusion.

A repeatable weekly NFL workflow

1. 1Open the board after the Tuesday build and note the week's free picks and the count of 4+ diamond props.
2. 2For any card, read the individual recent game values before the clear count, and check whether the last 5 tell the same story as the last 10.
3. 3Read the data-confidence label and its reason. Treat limited data as limited, not as neutral.
4. 4Check the matchup and environment signals: defense vs position, roof, points, and any weather caution on a passing or receiving prop.
5. 5Read the 'Why this could miss' line and decide whether the case against outweighs the case for.
6. 6Re-check on Friday or Saturday after injury designations settle, then treat the card as final at kickoff.
7. 7After the week, read Results to see how free picks and Top Picks graded by week and by market.

Common pitfalls

• Reading the rating as a probability. It is a research signal built from clears, projection and confidence.
• Treating a Tuesday card as final. Injury designations and depth charts move until kickoff, and so can the card.
• Comparing an NFL clear rate to a baseball one. Ten NFL games is most of a season; the sample is small by design.
• Assuming a weather caution exists for every game. Forecasts only appear for open-air venues inside the forecast window.
• Counting a void as a loss. A player who did not record a stat line is neither.

Frequently asked questions

Keep reading