Defense versus position after one week: why the early tables mislead

What the table claims to show

Defense versus position, usually shortened to DvP, is a table of how many fantasy points each defense has allowed to quarterbacks, running backs, receivers and tight ends. The idea is sound. Some defenses are weaker against the run, some cannot cover slot receivers, and over a season those tendencies show up in the points column.

The problem is the calendar. On the Monday after Week 1, every defense in the table has played one game against one offense, in one game script, with one set of injuries on both sides. The table looks exactly as authoritative as it will in December. It is not.

All figures below come from Sleeper's Week 1 feed, pulled at 9:43 AM ET on Monday, September 14. Thirty teams had played. Denver and Kansas City meet on Monday night. Points are Sleeper's full PPR scoring.

Four Week 1 tables, four different stories

Source: Sleeper Week 1 team defense stats; scores from the ESPN scoreboard.

Carolina and the touchdown problem. Chicago's running backs scored 52.8 points against Carolina, and the whole of it came from two players: D'Andre Swift with 32.4 and Kyle Monangai with 20.4. They scored four rushing touchdowns between them, which is 24 of the 52.8 points. Sleeper's projections for every Chicago back combined came to 20.9 before the game. Some of the gap is a defense losing at the line of scrimmage. A lot of it is a game Chicago won 59 to 37, where the leading team runs the ball late and the goal line carries fall to backs. Touchdowns are the least repeatable part of fantasy scoring, and a one game DvP table is mostly a touchdown table.

Pittsburgh and the quarterback on the other side. Pittsburgh allowed 9.9 points to wide receivers, the fewest in the league. Before crediting the secondary, look at who threw the passes. Sleeper's player file lists Atlanta's top two quarterbacks, Tua Tagovailoa and Michael Penix, as Out, and Cooper Rush, third on the depth chart, started. Atlanta's backs, receivers and tight ends drew 19 targets for a combined 25 air yards, the fewest targets of any team that played before Monday. Receivers cannot score on throws that never go downfield. The same defense allowed 34.4 points to running backs, the fifth most, largely because Bijan Robinson caught 8 of 10 targets. That is one offense's plan, recorded as a defense's weakness.

Minnesota and the single play. Minnesota's defense was second worst against receivers and best against running backs in the same afternoon. Green Bay's Jordan Love threw 42 times for 387 yards, and Christian Watson turned 6 catches into 147 yards and 2 touchdowns, including an 81 yard reception. Remove that one play and the table reads differently. Nothing about the Vikings' run defense was proven by holding a team that threw 42 times to 6.6 points on the ground.

San Francisco and volume. The 49ers allowed 6.52 points to quarterbacks, the fewest. Matthew Stafford threw 25 passes for 155 yards and scored 5.1, and Stetson Bennett played 10 snaps. A quarterback who throws 25 times rarely gets to 20 points against anyone. The same pattern runs the other way in Detroit, which allowed the most points to receivers, 55.6, to a New Orleans offense whose quarterback, Tyler Shough, threw 56 passes.

A better first cut: allowed against expected

The simplest fix for one opponent is to compare what a defense allowed with what that offense was projected to score before the game. For each team we summed Sleeper's Week 1 projections for every player at a position and set that beside the points the opponent allowed.

It helps. Carolina allowed 31.9 points more than Chicago's backs were projected for, the largest gap at any position. Tampa Bay allowed 19.1 points to Cincinnati's receivers against a projection of 45.4, which looks like a shutdown secondary until you remember that Ja'Marr Chase, projected for 20.36, caught 2 passes for 12 yards. Adjusting for the opponent removes one bias. It does not add games to the sample, and one game is still one game.

What to use instead in September

The betting market. A sportsbook's game total and point spread already price both the offense and the defense, and they are set by people who lose money when they are wrong. A player prop is sharper still, because it prices this player against this defense this week. That is why the Parlae start or sit call leans on props first, as the four numbers guide explains.

Personnel over points. An injury in a secondary or on a defensive line is known information that the table has not yet caught up to. The reverse matters too: a defense that faced a third string quarterback is a defense you know little about.

Patience with the table itself. A reasonable rule is to treat DvP as a tiebreaker until a defense has faced four or five different offenses, and to read it alongside who those offenses were.

Applying it to Week 2

Minnesota plays at Chicago. Chicago's defense allowed 34.84 points to Carolina's quarterbacks, the third most, and that will be the headline for anyone streaming Carson Wentz. It is one game against a Carolina offense whose quarterback, Bryce Young, threw for 361 yards. Weigh it lightly.

Carolina plays at Atlanta, which allowed only 14.4 points to Pittsburgh's running backs against a projection of 26.6. One strong game is a reason to look at Chuba Hubbard's matchup twice, not a reason to bench him after 23.7 points in Week 1.

Detroit opens the week at Buffalo on Thursday night. Detroit allowed the most receiver points in Week 1, which makes Buffalo's receivers look like easy starts. The reason Detroit gave up those points, 56 pass attempts against it, is worth more than the rank.

What to do with this

1. When a DvP rank tempts a lineup change, open the box score of the game behind it. Check the opposing quarterback, the pass attempts and the touchdowns before you trust the number.
2. Compare allowed points to what that offense was expected to score. A defense that allowed a lot to a good offense has told you less than one that allowed a lot to a bad one.
3. Let the market set the matchup until October. Check the start or sit page, where calls priced against player props are labelled as such.