Introducing Individual Player Stats

You may have noticed some minor cosmetic changes to the site this morning. The menu has been reorganized and the site is more tightly integrated. But the big news today is that Advanced NFL Stats is introducing individual player stats pages. Advanced stats for quarterbacks, running backs, wide receivers, and tight ends for the last 10 seasons are now available. Just use the 'Stats' menu item to navigate to the position you want to see.

Most readers will be familiar with the stats. There's Win Probability Added (WPA) and Expected Points Added (EPA), plus per play versions of both stats. The usual suspects top the lists--Manning, Brady, Faulk, Tomlinson, Moss, Gonzalez--but there are a few surprises.

There's also an all new stat--Success Rate (SR). This is as simple as it sounds. It's the percent of plays in which the player participated that result in an increase in net expected point advantage. SR measures consistency as opposed to the total magnitude of each play's outcome. Together with WPA and EPA, SR can help paint a more complete picture of how a player performs.

For receivers I've also included Targets and Target Percentage--the share of a team's pass attempts thrown to a receiver. The NFL started tracking targets as an official stat just last season, but I've tallied them up for previous seasons using the play-by-play text description. There's also Catch Rate (CR), the simple percentage of passes thrown to a receiver that are caught. These aren't exactly 'advanced' stats, but they're things you usually don't see on stat sites.

I'm planning to add a few more stats for each position. For receivers I'll add %Deep, the percentage of passes attempts thrown greater than 15 yards downfield. I'm also going to add a stat for receivers inspired by Bill James' Range Factor. This would roughly be a receiver's catch rate compared to his team's overall accuracy to the same position (WR or TE), and adjusted for Target Percentage and the number of deep attempts thrown to him.

Right now all the stats include the playoffs, but I plan to make an additional option to show the regular season, the playoffs only, or both. Also, for now I've limited the players to 'qualified' players--those who have participated in a minimum number of plays, but there will be another option to show all players. And as I add more stats, I plan to have separate pages for advanced stats and conventional stats.

There will be more enhancements--player pages, team pages, opponent adjustments, special teams, and more. It's always a work in progress. Suggestions and comments are welcome.

  • Spread The Love
  • Digg This Post
  • Tweet This Post
  • Stumble This Post
  • Submit This Post To Delicious
  • Submit This Post To Reddit
  • Submit This Post To Mixx

26 Responses to “Introducing Individual Player Stats”

  1. Anonymous says:

    First!

    Great!

  2. Martin says:

    Almost everybody says that the Colts running-game is among the worst in the leauge, but both Addai and Donald Brown is in top 16 i succes rate, over players like D.Williams, L.McCoy and M.Jones-Drew. Also Donald Brown is #6 in WPA/P. Is it me reading the stats wrong, or is Addai and Brown better than most peapole thinks?

  3. Anonymous says:

    >>Is it me reading the stats wrong, or is Addai and Brown better than most peapole thinks?


    Perhaps its because everyone plays to stop the pass, giving these guys a better chance. Its not necessasrily that they are so good, it may be more that the Colts don't run much, and when they do, they benefit somewaht from teh element of surprise.

    I'm not saying these guys are no good, but play calling can have a big impact of stats.

  4. Ryan says:

    Martin,
    Those Colts RBs are used a lot more sparingly. It's interesting to note that almost everyone on the top of the SR list splits carries with another back, usually on a team with a good QB. I'm guessing workhorse backs get a lot more negative plays, since they're always running on first down and such... whereas the Colts have Manning to take off a lot of that pressure. SR% probably has more to do with coaching, putting RBs in position to make positive plays when least expected.

    It's an interesting stat to consider, but you'd have a tough time convincing me that that guys like Buckhalter and Moats are top 5 backs. WPA and EPA look like much better predictors of actual success. I'd like to see a stat that someone in the comments suggested a few posts back, where players (and coaches) get WPA/EPA added or subtracted, compared to the normal outcome of what would be an optimal playcall in a given situation, rather than a straight "plus or minus." Barry Sanders, for example, surely had more negative runs than positive runs, but they were so often HUGELY positive that it's not really fair to judge him on SR alone.

  5. James says:

    I'm impressed by the amount of work you put into this site!

    Can you find the WPA for coaching decisions?

  6. Ian says:

    Brian, great addition to the site.

    Some enhancement ideas:

    1) In the quarterbacks table add passing touchdowns.

    2) A table of most successful and least successful plays by EPA and WPA.

    3) This is related to Ryan's and James' posts: Take every punt, field goal, and 4th down conversion attempt. Create a table or two showing the decisions that were most and least supportable using your stats.

  7. Zach says:

    Is it possible to show WPA or EPA per like 500 plays for QBs or per 300 plays for RBs? .0039 WPA/play doesn't mean a whole lot, but 1.5 WPA/300 gives you a reasonable number.

    Also, I love the %Deep and Range Factor ideas.

    What's the league average success rate? 50%, I presume? What about for each position?

  8. Brian Burke says:

    Great ideas. Love the one about tallying WPA on coaching decisions. I already do that in house, so it's something I can roll out in the future.

    Eventually there will be separate pages for conventional and advanced stats, so I can include TD passes, fumbles, and all the other stuff.

    To answer Zach's question, there are actually more failures than successes, 58% to 43%, and that's even counting zero EPA plays as a success. (Zero EPA is playing par football.) I was going to explain this later, but here it is. Successes tend to be bigger successes than failures tend to be big failures, so when taken as a total, all plays net out to near zero EPA.

    The solid majority of plays are actually tiny, tiny 'failures'--incomplete passes and even 1,2 or 3 yd runs. They add up quickly, but bring down the average magnitude of failures.

  9. 81Trucolors says:

    Brian,

    Your site is amazing. Please consider adding a key to the individual player stats pages.

  10. Brian Burke says:

    You had me at "amazing." By "key" do you mean creating a link for each player's career? Sort of like this?

  11. bsencore says:

    This is fabulous. And individual player pages like the one you link to above for LT would be amazing.

    Also, and maybe this is what 81Trucolors is saying, but adding a "key" to the stats would be nice. If you get linked directly to the table of information but haven't been reading up on WPA, EPA, SR(%), its hard to pick up unless you search through the blog.

    Also, noting that the stats include playoffs on the page (at least until you separate it) would be a nice reminder when everything looks off.

  12. Alex says:

    fwiw I'm partial to the databases where hovering the cursor over a stat heading shows the definition of that stat, and then I agree that a creation of a detailed key would be nice too. Additionally, being able to set the filter across multiple years would be nice - to see wpa/epa leaders for 2001-2009, or from 2003-2006, or whatever.

    Another thing I think would be cool (in addition to conventional/advanced stat pages and individual player pages) would be box scores along with the wp charts of games (a la fangraphs).

    Anyways - the more data, the better - this is a really really cool addition and it's really impressive how you've built your site up (again i guess fangraphs is the notable comp).

  13. Brian Burke says:

    Thanks. Great suggestions. Fangraphs was not the original inspiration for my WP graphs, but since I discovered it, it's been the inspiration for a lot of the other things I'm adding. It's a great site with a great design. It's fun to click around there.

    Also I forgot to mention I really like the idea for WP/500 plays for QBs or WP/300 plays for RBs. Stats that go into the ten-thousandths place are never going to catch on.

  14. Anonymous says:

    Brian, on the individual pages, can you also create a box for career numbers. For example on the LT page, you have his individual season numbers, but nothing for total career numbers.

  15. Anonymous says:

    Can you let us choose "Any" for the years, so we can see the most WPA by a QB since 2000?

  16. Buzz says:

    Brian,

    I would like to reiterate all of the great praise that is being thrown out by the posters on this thread. Since i found your site about 2 years ago it has become the staple of my online football research.

    I think your range factor that you mention will be a very interesting stat and might help to start determining how good various receivers really are - at least as compared to their own team. Seperating QB play from WR play (not to mention how important a line is, etc)is about as tough a task as you can have and this is a great step.

    Other items that i think would be great since you asked for suggestions would be to include a total at the bottom of the player page that you showed for LT. It would be even better to have a playoff total and a regular season total with how much weight everyone gives to the playoffs.

    I think a WPA or EPA per game would also both be great additions as alternatives/additions to "per 500 pass attempts" as they would really help paint a picture to how much a player contributed to a specific win.

    A defensive adjustment would be really cool as i imagine it is a lot easier to accrue WPA points vs the lions as compared to someone like the steelers.

    Finally i think it would be very interesting to see some team numbers as well. Maybe you could have a WPA added/lost by offensive rushing, offensive passing, defensive running, defensive passing, and special teams (i suppose penalties too).

    Just a couple of random thoughts. Great job all around!

  17. Anonymous says:

    Brian,

    The links at the top of the page don't work. When I click on Stats, Analysis, etc, it just takes me back to the home page.

  18. bsencore says:

    What AYPA are you using? It doesn't seem to match up with PFR's AY/A or ANY/A -- even for the non-playoff QBs.

  19. Brian Burke says:

    If you don't have CSS style sheets enabled in your browser, you won't get the menu links.

    bsencore-It's the same formula but without the bonus for TD passes. It's:

    (PassYds - SkYds - 45*Ints)/(PassAtt + Sks)

    It would be closest to PFR's ANY/A but without the TD bonus.

    A career total line has been added for the player pages. You can go to:
    http://wp.advancednflstats.com/playerpage.php?playerid=21-L.Tomlinson&pos=RB to see. If you change the playerid and position variable in the link, you can see any player, e.g.:
    http://wp.advancednflstats.com/playerpage.php?playerid=81-T.Owens&pos=WR. It's not live yet because it doesn't account for guys who have changed numbers or names (Ochocinco). So for now, Randy Moss has 3 different pages--one for his time as #84 in MIN, one for his time as #18 in OAK, and one for #81 in NE.

  20. Brian Burke says:

    I should also mention I added '%Deep' for receivers--the % of targets attempted beyond 15 yds. The NFL changed the way it tracked the depth of passes in 2006, so for now I only have it from 2006 on.

  21. James says:

    Hello, font change? Also, I just noticed your dropdown menus (for me anyway) only appear when I scroll over the white space below "Research", "Analysis", etc. and not while I my mouse is over the words themselves.

  22. Nick Dilonardo says:

    RB statistics:

    Yards per carry is problematic - the long, breakaway threat back's YPC can be misleading (although that is not to make a substantive claim on a slow plodder being preferable to a hit or miss type) - but perhaps a YPC based on down and distance?

    I know this point of mine might lack inherency and WPA might make this point moot, but I think it would be worthwhile to look at backs from the 1st and Ten perspective, 2nd and Long, as well as 3rd or 4th and short. I think looking at backs this way could be particularly helpful.

  23. Anonymous says:

    Like Nick said, how about median yards per carry? I remember you had a post about it a while back but hadn't followed up on it.

  24. Paul says:

    This is wonderful stuff Brian

  25. Keith says:

    This stuff is really interesting. I'm actually beginning to do some research on the subject myself (I began with the idea of WPA from baseball into football and then I found that you had been doing the same thing already off of your own win probability model).

    My comment is this: I have the understanding that you're individual WPA is based off of the play-by-play player contribution. Since this doesn't take into account everything off the ball I think it may be incomplete. For example, a running back who picks up a blitzing line backer, enabling the QB to throw a pass is attributed no success from the play. What would be a more accurate individual player rating would be it based on a look at plays with that player in the line-up and that player not in the line-up--potentially compared to standardized league averages--(maybe somewhat like the replacement player stat in baseball). Let me know your thoughts

  26. Dave says:

    Man Jamal Lewis's 2000 yard season comes out as basically league average in terms of WPA and EPA.

    How much do lost fumbles typically impact the WPA and EPA?

Leave a Reply