There are many tools people use when evaluating a player. This includes film review, situational stats, overall stats, advanced analytics, and more. Here at Pro Football Network, we evaluate players through our proprietary model called the Offensive Share Metric (OSM).
We created this model because it can often be difficult to tell how well a player is performing during an NFL game. Conventional statistics are helpful, but they almost never tell the full story. Fans often claim that a certain player is only good because of their teammates, or that they are being held back by the players around them.
Related | Relative Athletic Scores (RAS) Database
PFN OSM is a way to see if those claims are actually true. The OSM grades measure how much of a player’s statistical production they were actually responsible for. For example, a wide receiver that doesn’t drop passes and breaks a lot of tackles will have a higher grade than one who did the opposite. So, a player with a higher grade was more responsible for their own production than a player with a lower one.
These facts make the OSM useful for gauging an individual player’s actual effectiveness in a game, separate from the performances of their teammates. For more on how the Offensive Share Metric is graded, please scroll to the bottom of the page below the player database.
Minimum Qualifications
The following are the minimums required for players to be recorded in the NFL Next Gen Stats database. As such, these are also the minimums for a player to qualify for a PFN OSM score.
Quarterbacks: 15 pass attempts per game | 128 attempts through Week 17
Running Backs: 10 rushing attempts per game | 85 attempts through Week 17
Wide Receivers/Tight Ends: 5 targets per game | 43 targets through Week 17
PFN Offensive Share Metric Database
The table below shows the PFN OSM grades for the 2016-2019 season. A player with a 40+ grade is considered elite, one with 30-39 is very good, 20-29 is good, 10-19 is average, and 9-below is poor. Mobile users: This table is best viewed in landscape mode.
More on the PFN Offensive Share Metric
Our favorite thing to do when discussing football is to open up a debate about a particular player. You know what we mean, where one person is yelling that Player A is the best player they have ever seen and the other is screaming that the player is merely a product of the system or his supporting cast.
If Player A were on any other team, they would be horrible. We see this time and again when players are traded or signed as free agents. They get to their new team and their production takes a hit, or they keep steaming along. The problem is twofold.
First, how do you determine if Player A is that good or bad? Or, worse, is he merely a good fit for a particular system? If it’s the latter, then your team and his production has dropped, that team has made a big mistake.
Football is the ultimate team game. That makes it inherently difficult to determine the individual performance of different players. If you try to use generic stats like passing yards and completion percentage, you end up using stats that speak to the result of the combined efforts of multiple players. The same is true of their offspring – third down completion percentage, fourth quarter passer rating or, even worse, wins,
Misleading Stats
Let’s look at the most common example – a quarterback throws a beautiful ball down the sideline on 3rd down to an open receiver who has two steps on his defender. He drops it. Through no fault of his own, the quarterback’s completion percentage takes a hit.
His third down completion percentage takes a hit. Now, he doesn’t have as many yards as he should have. Again, none of that had anything to do with anything he did individually.
Likewise, let’s examine a wide receiver or tight end who gets three yards of separation from the nearest defender and catches a pass that only traveled seven yards in the air. After the catch, the player then broke two tackles en route to 25 yards after the catch. The receiver did everything after the catch on his own. The broken tackles and the 25 yards after he caught the ball were all a result of his behavior.
In the box score, however, the quarterback gets equal credit as the receiver. He gets a 32 yards “pass,” and the receiver gets a 32-yard catch. Take it a step further. Let’s say the receiver scored on the play. Now the quarterback gets credit for a 32-yard touchdown pass.
That one stat will count towards whether or not he is considered for MVP, even though his effort on the play amounted to nothing more than throwing the ball seven yards in the air to a wide open receiver.
Offensive Share Metric (OSM)
It is these all too common scenarios that led us to come up with a metric that measures how well a player performed relatively ONLY to how he performed in areas solely under his control. We call this type of analysis “football behavioral analytics.” Behavior is something all humans have and do. It is what we do when the environment presents us with a set of circumstances that cause us to behave in a certain way.
We continue to act in this way based on whether or not we found that behavior to be successful for achieving our needs. In football, we use this to look only at one player’s football behavior in a given environment (system, supporting cast, etc.) and determine how well they performed in that environment, regardless of what the other players around him do.
And thus, the Pro Football Network Offensive Share Metric (PFN OSM) was born. This metric is rated out of 100%. No one player will ever have close to a grade of 100. That would mean that player threw the ball, then caught it and somehow also rushed with it simultaneously. So if the grade seems low, this is why. But to help you better understand what each grade means, we’ve created a scale, which you can find above.
How is OSM calculated?
OSM is formulated by looking at a combination of mostly the NFL’s very own Next Gen stats and some others. Doing this allows us to see how well a quarterback or receiver did with what only he could control. This metric accounts for things like air yards, completion probabilities, and differential, aggressiveness, and more.
If a quarterback performed poorly in these areas, then it is logical to conclude that he is not the reason for his team’s offensive production. Instead, we would conclude that others on the offense played a more significant role in the success. Likewise, if a receiver is poor at getting separation, drops more targets than he should and doesn’t get the YAC he should, their OSM will rate lower, as he is mostly in control of those outcomes.
If a running back has a large per carry average, maintained that while facing eight or more men in the box a larger percentage of the time, and is efficient at getting north and south quickly, he will rate highly because those are things under his direct control under challenging environments.
To be clear, this metric is not a power rankings type metric. If quarterback A has a higher rating than quarterback B, it does not necessarily mean that quarterback A is better than quarterback B. It is just that B is less responsible for the production of his offense then A is for his.
Since all offenses are different, place different burdens on their players, and that no two supporting casts are the same, it would be unwise to try and use this as a ranking method for direct player-to-player comparisons from different teams.
Thinking of it in different terms
PFN OSM is used merely to provide deeper context to box score stats and help fans see who on their favorite team is standing out and who is maybe not pulling their weight. Think of it like a quadrant grid. If player A has a high ranking and the overall production of the offense is ranked high, they would appear in the top right.
Conversely, if player B is ranked low and their offense is ranked low, they would appear in the bottom left. If Player C ranked high, but the offense ranked low, you would conclude that the player did the best he could with what he was asked to do and work with.
Likewise, if Player D ranked low, and the offense ranked high, you would conclude that the player did not necessarily pull his weight and underperformed. You would also find that offense as a whole benefited from the supporting cast than that one player.
It’s all just part of the puzzle
This is a brand new and unique way of looking at football players and we expect it to take some time to catch on. PFN OSM is one tool of many that should be used to evaluate a player, with none more important than film.
Our stance has always been this – film first, numbers second. If the numbers don’t match what is seen on film, something is likely wrong with the numbers. That was taken into account when developing our proprietary formula for OSM. This is also why we have a dedicated film room section – to help all of the pieces of the puzzle come together.