Introducing Total Points
Guest column by Alex Vigderman
Prelude: Whither Matt Ryan?
Matt Ryan won the 2016 NFL MVP award with an incredible campaign. He set career bests in completion percentage, yards, touchdowns, and interceptions, and led the league in touchdown rate and whichever raw or adjusted yards per attempt metric you prefer. In 2017, his numbers fell off substantially across the board, including a near-halving of his touchdown total. This year he is off to a torrid start, on pace to at least match the numbers he put up two years ago. So what gives? What happened last year?
It turns out that last year wasn't much of a down year at all, at least not one that was reflective of Ryan declining. In 2017, Falcons would-be pass-catchers dropped more than 50 percent more passes than they did the year before (or are on pace to drop this year) while Ryan's catchable pass rate was about the same. Add in that his receivers averaged just under a yard less per reception after the catch, cutting his totals quite a bit.
There were some serious red zone troubles, though, and that ineptitude crept into 2018 as well. Ryan's catchable pass rate in the red zone dropped off by nearly 10 percent from 2016 to 2017, and his receivers' catch rate on catchable passes dropped by more than 12 percent. These numbers have an outsized impact perceptually because of where they came on the field, but in such a small sample they're hardly predictive. What we'd like to be able to do is determine what a player is really responsible for and make sure he gets credit for it, no more and no less.
Sports Info Solutions has developed a system called Total Points that seeks to do just that using our wealth of charting data and the Expected Points framework. Using that system, Ryan was the best passer in the NFL in both 2016 and 2017, which makes his statistical rebound in 2018 more of a continuation of a trend than anything. In the space below, we'll introduce the system and how it's used to evaluate every player on the field.
Introducing Total Points
Before we get to the more complex Total Points system, let's make sure that we're set on the concept of Expected Points.
Football is a very tough sport to evaluate statistically. To resolve the issue of the overall complexity of the game, many analysts have settled on a common language when talking about the impact of different events during the game. That language is Expected Points (EP). At any point in time in a game, the Expected Points is the average value of the next score of the game, regardless of when it occurs or who scores. This allows us to understand how valuable just a change in field position is, regardless of whether there was a score on that drive or not.
The value of any individual play can be calculated as Expected Points Added (EPA), the difference in EP before and after a given play. Having EPA as a currency allows all sorts of different play results and events to be compared apples-to-apples, taking into account the fluctuating value of yards and downs as the other changes. And this allows for really interesting analysis when applied on a holistic level. For example, EPA shows that a "staying ahead of the chains" run of 4 yards on first down is most often a negative play. The rare exceptions include plays at the outer end of field goal range.
Where the typical EPA model struggles, though, is player analysis. "Traditional" EPA-based analysis of players might read like, "Player X was worth 27 EPA catching passes in 2017," but that really means "Player X's team gained 27 EPA on plays in which Player X caught a pass in 2017." There isn't any real attribution of who is responsible for that added value. Which is why SIS sought to develop a more useful and robust model that really takes advantage of the wealth of data available, a model called Total Points.
Total Points takes the common currency of EPA and distributes the value gained or lost on a play to the different players involved based on their impact on the play. Using the wealth of SIS charting data available, this affords an unprecedented assessment of the value provided individually by each player on the field.
Take these two incomplete passes as an example:
1. The quarterback misses on a quick slant, throwing it way behind the receiver.
2. The quarterback is forced to escape the pocket after the left guard blows a block. Despite that, he delivers a strike to a receiver streaking down the sidelines who drops it.
Most traditional evaluation systems would consider these plays basically the same. The pass was incomplete in each case. But so much more happened to inform the evaluation of the players involved.
For Play 1, the quarterback was offered a clean pocket and still threw off-target, so Total Points absolves the receiver of blame and punishes the quarterback. For Play 2, the situation is the opposite. The pass was on-target and the receiver dropped it, so he gets blamed while the quarterback receives credit as if the receiver had caught it. On top of that, the offensive line's failure will make the quarterback look even better in that scenario.
These are two plays with the same in-game outcome but very different player performances, and Total Points helps distinguish that. As an example, assume the EPA on both plays above was -0.5. Using basic EPA, the quarterback and receiver would both be assigned -0.5 EPA, the full value of the play. With respect to Total Points, they might be respectively assigned -0.5 and 0 Total Points for Play 1 but 0.2 and -0.7 Total Points for Play 2, which takes into account their individual contributions to the outcome.
A Tour Around Total Points
Without digging too far into the details, here is an overview of the different considerations entering into Total Points. While Total Points is the name for the overarching system, offensive Total Points are called Points Earned and defensive Total Points are called Points Saved. This distinction helps clarify between totals of players on different sides of the ball that would otherwise be on the same scale.
When looking at these positional leaders, it's important to remember what Total Points is best suited for when it comes to player analysis. Total Points is useful to differentiate between players at the same or similar positions, especially within similar aspects of the game -- for example, comparing the performance against the run of a linebacker and a defensive tackle. This isn't as true when crossing between these aspects. For example, because of the added value of passing from an Expected Points perspective, players involved in the passing game will have more extreme Total Points (good or bad) than players involved more in the running game.
Also, these leaderboards are for 2017 to give a sense for what full seasons look like. With a new metric comes a whole new scale to internalize, so we figured it's better to use full seasons to start off.
As was touched on in the pass examples above, the relationship between the quarterback and the receiver can get fairly involved. The core assumption of passing Points Earned is that each throw has a certain expected outcome based on information like the route, the depth, and the coverage. From that point, the passer and receiver split responsibility for how well they perform above that expectation. Throwing off-target passes and deserved interceptions (caught or not) will bury a signal-caller, while he will be rewarded for leading receivers to more yards after catch and making something out of a broken pocket. The receiver is then evaluated in the context of his route, so 5 yards after the catch on a screen nets less credit than the same amount on a curl.
|Passing Points Earned Leaders, 2017||Receiving Points Earned Leaders, 2017|
|Player||Team||Points Earned||Player||Pos||Team||Points Earned|
|Matt Ryan||ATL||62||Keenan Allen||WR||LAC||43|
|Philip Rivers||LAC||59||DeAndre Hopkins||WR||HOU||37|
|Tom Brady||NE||55||Julio Jones||WR||ATL||36|
|Case Keenum||MIN||51||Rob Gronkowski||TE||NE||35|
|Ben Roethlisberger||PIT||43||Michael Thomas||WR||NO||35|
|Drew Brees||NO||42||Antonio Brown||WR||PIT||32|
|Carson Wentz||PHI||35||Marvin Jones||WR||DET||25|
|Alex Smith||KC||31||Tyreek Hill||WR||KC||25|
|Russell Wilson||SEA||28||Travis Kelce||TE||KC||24|
|Jimmy Garoppolo||SF||21||Davante Adams||WR||GB||23|
Our poster boy Matt Ryan tops the 2017 Points Earned leaderboard among passers thanks to an excellent on-target throw rate (second to Drew Brees among quarterbacks with at least 200 attempts) and correspondingly very few deserved interceptions. His No. 1 target, Julio Jones, doesn't get the same honor because so many of those on-target throws that went his direction ended up being dropped. Brees' go-to target Michael Thomas also makes the receiver leaders thanks to sure hands and some tough runs after the catch.
Philip Rivers put a ton of throws in the defenders' hands but countered that with a lot of throws that led to strong yardage after the catch (which benefited the pass-catching leader, Keenan Allen). Carson Wentz and Jimmy Garoppolo make the list despite shorter seasons; their prorated full-season performance would have put them in the conversation for the top handful at the position were they given the opportunity.
Offensive and Defensive Front
Line play remains one of the more elusive facets of the game from an analytical perspective. Total Points seeks to evaluate the offensive line by starting from a baseline of a clean play and punishing blockers for indicators of poor play, like blown blocks, running back bounces, and other point-of-attack events (e.g., QB hits, run stuffs, and batted passes). Pass-rushers and in-box defenders are correspondingly credited for being involved in those disruptive events and docked for allowing too many cleanly-blocked plays. Each team's performance in the trenches then informs how the players behind them are evaluated; poor run blocking makes a good run look better, and an excellent pass rush makes good coverage look less impressive.
One thing to keep in mind is that while SIS collects a good amount of data on these otherwise under-represented position groups (especially on run plays), there is still a fair amount of uncertainty about how responsible any given player is for some events on the field. Total Points reflects this by producing a lower Points Saved or Points Earned on those plays, so that there isn't too much credit being given one way or another. As a result, performance by the offensive line and defenders on run plays will have a much tighter range of Total Points than others. As SIS expands its data collection even further this will become less and less of a factor.
|Offensive Line Points Earned Leaders, 2017|
Brandon Brooks made his first Pro Bowl last year with the third-lowest Blown Block Rate among guards with at least 300 snaps at the position. There is less credit to go around on run plays because of their per-attempt inefficiency, but he trailed only teammate Jason Kelce and Max Unger in Points Earned in the run game. Brooks shared the 2017 Points Earned lead with 2016 All-Pro Jack Conklin, who leads all offensive linemen in Points Earned the last two years. And Joe Thomas went out on top thanks to yielding two blown blocks in 446 snaps, the lowest among linemen with at least 300 snaps.
|Defensive Line Points Saved Leaders, 2017|
Linemen who can disrupt the passing game get big contracts and big Points Saved totals. New $100 million man Aaron Donald was one of the best in the league at sacking the quarterback and forcing fumbles, and he gets his due here. Calais Campbell and Demarcus Lawrence far and away led defensive linemen in blown block responsibility, but Lawrence's four forced fumbles made him the top dog in this group. The Jaguars' stellar overall defensive effort in 2017 was keyed by the performances of Campbell, Yannick Ngakoue, and Malik Jackson, who each contributed against the run as well as anyone on the leaderboard.
Rushers are so clearly impacted by the play of the offensive line that much of the time analyzing running backs is spent trying to separate their performance from that of their linemen. Evaluating the offensive line's Points Earned separately allows the evaluation of the back to start with this already taken care of. Total Points acknowledges the role that the back has in his yards before contact and bounce/cutback rate, two metrics that have tended to be attributed more to the line in the past. Analysts using other Expected Points models have contended that running backs are largely irrelevant from a value perspective in the modern NFL. Total Points for the most part agrees, although those who are able to consistently break through tacklers and catch the ball out of the backfield will still be valued by Total Points.
|Running Back Points Earned Leaders, 2017|
Kareem Hunt burst onto the scene last season, making a lot of big plays both on the ground and through the air. Dion Lewis, now with the Titans, was very well-acquitted by SIS charting last season, running successfully between the tackles and through contact. And his relatively lofty total includes a fairly hefty debit for the quality of the blocking in front of him. The pass-catching exploits of Le'Veon Bell, Alvin Kamara, and Rod Smith drive their inclusion on this leaderboard, showing how important it is to bring versatility to the table at a time when running is less valuable.
Unsurprisingly, Total Points rewards ball-hawking defensive backs for defensing passes and forcing interceptions. Beyond that, though, the sure tacklers and shutdown defenders finally get their due statistically. Like with receivers, defensive backs are evaluated compared to the expected yardage on the route, so making a quick tackle on a slant or shallow cross does not go unnoticed. Along with that, each receiver and defender in coverage is evaluated for how likely he is to be targeted given elements like where he lines up, how deep he is, and the type of coverage on the play. Defensive backs who consistently deter throws are credited for doing so, and are correspondingly dinged for being peppered with targets.
|Pass Coverage Points Saved Leaders, 2017|
Second-year star Jalen Ramsey made offenses pay for targeting him heavily, getting big interceptions and making stops after the catch en route to the top spot on this list. Jimmy Smith hasn't played a full season since 2015 and won't this year either, but he can cover as well as anyone and is a sure tackler who will make an impact when available. Two rookies, William Jackson and Marshon Lattimore, showed incredible promise in different ways in Year 1. Jackson was a terrific target deterrent in a limited role in Cincinnati. Lattimore overcame a higher rate of offensive attention by forcing turnovers.
For More Information
The main place to find these new metrics is on the SIS DataHub, the one-stop shop for the full suite of SIS charting data. There are player and team leaderboards for every facet of the game (passing, rushing, pass rush, run defense, etc.), and each of them can be filtered by dozens of different variables ranging from down-and-distance to run blocking scheme. You can register for a one-week trial to explore the breadth and depth of data available from Sports Info Solutions. You'll also hear Total Points numbers quoted weekly on the Off The Charts Podcast. And give us a follow on Twitter @SportsInfo_SIS to get updates on these statistics as well as a host of other tidbits.
Alex Vigderman is a Senior Research Analyst at Sports Info Solutions.
6 comments, Last at 19 Nov 2018, 9:28am
#3 by schmoker // Nov 17, 2018 - 12:37pm
This is a fascinating new stat. I follow sports not as a team or player fan anymore, but as a stathead. I'll be following to see where this goes.
The first question that pops up for me is why is it that Ben Rothlisberger is such a total points outlier compared to everyone else, and compared to EPA? He seems to be the only QB getting penalized so harshly in this new stat. Setting it at 100 throws, he goes from 6th by EPA to 24th in TP. I'm more than willing to believe 6th is too high, but 24th seems bananas.
Do you think that TP might take a QB who would rate highly on a bad team and harshly and unfairly penalize him for being surrounded by great players? Obviously that's going to hurt a QB somewhat when divvying up credit for a particular play, but would it in fact make it impossible to accurately judge a QB on a loaded team?
I'm having a hard time with a stat that presents Ben as being worse this year than Eli Manning, Derek Carr and BROCK OSWEILER!
Would the Steelers really be better with any of those three as their starter?
#4 by mrt1212 // Nov 17, 2018 - 2:27pm
That is a curious thing and makes me wonder if somehow, the mesh between tactics and receiving talent, all of it, on the Steelers really does tilt the success more towards being the receivers in the big picture. Like, yes Ben does get the ball there but so much of the Steelers is about what the receivers do to get it and proceed to do after that and what they do is a cut above the rest of the NFL for a while.
At least this season they are far and away the leaders in YAC and in previous seasons been anywhere from middling in absolute YAC to excelling over the past 5 seasons.
#5 by schmoker // Nov 18, 2018 - 11:37am
I can believe that is the case, and it certainly makes sense to me that Ben is artificially enhanced by both the line and the receivers he has. But I still can't see how he drops below Osweiler, among others. That speaks to a bug in the system. But I'm not doing the charting, so I'd just like to hear it explained. Maybe I could be convinced. I've never watched the film in my life, so I could easily be missing something. But it sure does seem like decision making alone favors Ben over a number of the quarterbacks listed above him. (Which is not to say that sometimes he does things that make me say, "Wait, what?" But all quarterbacks do that. Heck, Aaron Rogers and Brady have both done that a number of times this year.)