Anyone who has read my articles knows I have an affinity for NFL statistics. However, I don't value just any stat. I value really only those that correlate to winning NFL games.
Brian Burke of Advanced Football Analytics is one of the leaders in this field as he is perhaps best known for his Win Probability charts.
These charts start with the basic premise that each NFL opponent, in a league designed for parity, starts with a 50% chance of winning the game. The respective talent on each team is not considered — that is what Vegas betting lines are for. Instead, the charts consider only the score, down and distance, time remaining, and so on.
Once the game begins and starts taking shape with scoring, he uses historical data for past games that have a similar set of circumstances to calculate each team's odds of winning.
For instance, here is the Win Probability chart for the Ravens @ Broncos playoff game from the 2012 season.
You'll see that as soon as Trindon Holliday ran that punt back for a touchdown at the start of the game, the Ravens win probability dipped from just over 50% to a 23% win probability.
When Torrey Smith made his leaping grab over Champ Bailey for a touchdown to tie the game up at 21-21, Baltimore's win probability leapt up to 48%. On the Ravens' final drive of regulation, they had only a 2% chance of winning the game. This shouldn't surprise anyone as teams that are down seven points, 70 yards away, with no timeouts, with one minute left almost never win...
Flacco and Steve Smith dominating in Win Probability Added
Win Probability Added is an awesome stat in terms of evaluating a player's direct contribution to his team's chances of winning the game, which makes it infinitely more interesting than, say, total yards.
Win Probability Added measure's the impact of a single play on the game's outcome. This value can then be ascribed to the players involved (including defense). As plays are run, this value fluctuates up and down over the course of the game.
Let me give an example of how this works.
If on the first play of the game Flacco throws a long touchdown to Steve Smith, both players will receive a big boost in personal WPA because the Ravens' chances of winning that game will have just gone up noticeably.
If on their second drive of the game, Flacco throws a pick-six, Flacco will see a precipitous drop in his personal WPA while Steve Smith will remain the same effectively. By the end of the game, the total sum of these plays is added to give us the final contribution of each player towards winning or losing the game.
Through four weeks, Joe Flacco is leading all NFL quarterbacks in WPA and Steve Smith, Sr. is leading all NFL receivers. In other words, Joe Flacco and Steve Smith have done more to help the Ravens win their football games than all other players at their positions in their games.
Flacco has an average WPA per game of 0.39, which means that Flacco has increased the Ravens' chances of winning each game by 39% on average. He is followed closely by Philip Rivers who is also playing fantastic ball. Steve Smith is averaging a WPA per game of 33% which is far above the next receiver, Jeremy Maclin. That's pretty good.
The rest of our offense is positively contributing, too, with the exception of Jacoby Jones, whose problems I highlighted yesterday.
Who are our Defensive Contributors?
On defense, Burke only considers plays that result in a benefit for the team. Plays that are great individual efforts but the play still results in a negative for the team are not counted. So defenders are treated a little differently but the ultimate takeaway is the same.
On defense, the single biggest contributor in WPA through four games is none other than our rookie Inside Linebacker C.J. Mosley, who ranks 13th among all NFL linebackers with 0.40 WPA. Deangelo Tyson is also having a great season so far with a WPA of 0.32 through four games.
To the surprise of no one, our secondary is not doing so well, with the exception of Jimmy Smith.
Stats are great but if they don't correlate to winning, their ultimate value is diminished. We want to know a true picture of a team's strength. Stats like total yards will always be misrepresented in so-called "team rankings" because they correlate poorly with winning and thus don't really tell us much.
While stats like DVOA are great for evaluating a team's true ability in lieu of popular metrics like yards, stats like WPA give us a great idea of who is contributing individually to the win column.
And wins are the stat all of us care about the most.