BBTF's Newsblog Discussion :: Studeman: Postseason Probability Added

Baseball Primer Newsblog
— The Best News Links from the Baseball Newsstand

Friday, January 16, 2009

Studeman: Postseason Probability Added

That’s been our quest: to measure the impact of Howard’s timely performance. We’ve answered the question of how “important” September games truly are (I prefer the word “dramatic”) with our drama index. Now, we’re going to combine the drama index with WPA to develop PPA (oh no! another acronym!), or Postseason Probability Added. Quite simply, we’re going to multiply each player’s WPA in each game times that game’s drama index. And that’s all the explanation you’re going to get out of me, because it’s that simple.

So how do Pujols and Howard compare?

Player Team WPA DI PPA
Howard, R PHI 2.37 0.75 4.48
Pujols, A STL 6.39 0.63 4.39

Well, look at that: a virtual PPA tie. Even though Howard’s overall Drama Index was only slightly higher than Pujols’, Howard produced in games with the highest index (in other words, “when it counted”) and he completely closed the four-win gap in WPA. Pujols’ production was more even during the year, and the ratio between his WPA and PPA reflects that.

A really fantastic innovation. How long until this shows up on Fangraphs?

Heinie Mantush (Krusty) Posted: January 16, 2009 at 05:04 PM | 29 comment(s) Login to Bookmark
Tags: sabermetrics

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

1. Ivan Grushenko of Hong Kong Posted: January 16, 2009 at 06:04 PM (#3053708)

This Drama Index sounds like it's for girls.

2. WillYoung Posted: January 16, 2009 at 06:33 PM (#3053737)

That Denard Span triple was one of the most exciting moments of my life. The Metrodome was absolutely rocking and I'm getting goosebumps just thinking about how much fun that comeback was.

3. Mike Green Posted: January 16, 2009 at 06:34 PM (#3053738)

For several years, I have referred to the concept of "season leverage", and used it for the evaluation of relief pitchers' careers. Same idea.

4. Clint Posted: January 16, 2009 at 06:37 PM (#3053739)

I've been doing something like this for the last couple of years for my own amusement, except that I do it to see which individual plays most affected the playoff races. I multiply a play's WPA times the amount that each game changed the winning team's postseason odds, according to Baseball Prospectus. It's an interesting exercise. It's sort of like what Hardball Times used to do for their annual in listing the plays with the highest WPA for the year, but it adds in the playoff implications.

Sometimes the method yields obvious plays as a team's most important of the year, and sometimes it doesn't. Usually, the odd results come when a team runs away with a playoff spot, so that early season games move the BPro playoff odds more than late season games -- then you can get May home runs that work out to be the big hit of the year.

You can also conversely calculate which play adversely affected a team's playoff chances the most.

5. DL from MN Posted: January 16, 2009 at 06:38 PM (#3053745)

I've been waiting for a comprehensive WPA database of postseason performance for the history of the World Series and associated playoffs. It seems to me that the idea of WPA makes the most sense in that context. Is anyone working on such a thing?

6. Designated Sitter (GGC) Posted: January 16, 2009 at 06:40 PM (#3053747)

I enjoyed this three part series. I thought of something similar after Morneau won the MVP a couple of years ago.

From the comments at Ballhype, this sounded pretty labor intensive.

7. Greg (U)K Posted: January 16, 2009 at 07:26 PM (#3053786)

Agree with #5

I've often thought the ideal use of WPA is in determining playoff series MVPs.

8. Designated Sitter (GGC) Posted: January 16, 2009 at 07:41 PM (#3053805)

DL, amstat.org used to have a page about World Series MVP's but I can't find it. But it only went back a few years.

9. Davo Malvolio Posted: January 16, 2009 at 07:53 PM (#3053822)

That Denard Span triple was one of the most exciting moments of my life. The Metrodome was absolutely rocking and I'm getting goosebumps just thinking about how much fun that comeback was.

Oh, couldn't agree more. I'd grown up in North Dakota, so had only seen 3 baseball games in my life before moving to St. Paul in the summer. It was by far the greatest sporting event of my life!

10. Fridas Boss Posted: January 16, 2009 at 07:56 PM (#3053829)

I'm going to the play role of broken record and say that this doesn't do anything for me in terms of making Ryan Howard "closer" in MVP-worthiness to Pujols.

Look at that WIDE gap in avg WPA! Pujols contributed more value in each and every game which DIRECTLY affected the drama index of later season games for the Cards. Drama index isn't an independent variable here, the players have a say in creating that index.

It just seems faulty to me that Howard performing poorly early and setting up the drama for the Phillies creates a POSITIVE point in his MVP candidacy.

Thanks for an interesting series, studes!

11. DL from MN Posted: January 16, 2009 at 08:12 PM (#3053846)

Sounds like the answer is _no_. This could be a really fun project. What effort does it take to convert the play-by-play into WPA for a single game? From the period of 1903 to 1968 there were fewer than 400 playoff games played. From 69-94 we're looking at ~450 more. From 1995-2008 is about another 400 games. That's approximately 1200 playoff games in total. If we broke it down with 10 volunteers and each converted one game per day then it would be done in a few months.

12. DL from MN Posted: January 16, 2009 at 08:27 PM (#3053856)

Looking over at Fangraphs they seem to have quite a bit of this back several years. I would love to work with Fangraphs to get all the postseasons done using their tools. Then I'd like to create a searchable player database of WPA.

13. JPWF13

Posted: January 16, 2009 at 08:30 PM (#3053863)

I'm going to the play role of broken record and say that this doesn't do anything for me in terms of making Ryan Howard "closer" in MVP-worthiness to Pujols.

what! Does this mean you don't think Bucky Dent was the true 1978 MVP either?

14. ugen64 Posted: January 16, 2009 at 08:43 PM (#3053872)

"I'm going to the play role of broken record and say that this doesn't do anything for me in terms of making Ryan Howard "closer" in MVP-worthiness to Pujols."

maybe it doesn't in your view or mine, but I think part of the exercise was to put ourselves in an MVP voter's shoes and see which players really performed the best down the clutch.

I think some of the more interesting conclusions are that Beltran and Mauer were still undervoted, even using this criteria.

15. Sean Forman Posted: January 16, 2009 at 09:26 PM (#3053902)

Baseball-Reference postseason box scores have WPA for all playoff games in history. We haven't taken the next next step of calculating batter WPA because of not taking the step to get baserunning plays properly distributed. For pitchers we do have the results.

Top 20 pitchers in wpa.

| Mariano Rivera      |   454 | 10.0188 |
| Curt Schilling      |   531 |  4.3455 |
| John Smoltz         |   865 |  3.7090 |
| Orel Hershiser      |   547 |  2.8173 |
| Andy Pettitte       |   949 |  2.7569 |
| Art Nehf            |   323 |  2.6739 |
| Roger Clemens       |   867 |  2.6084 |
| Orlando Hernandez   |   457 |  2.5879 |
| Tom Glavine         |   923 |  2.5200 |
| Red Ruffing         |   355 |  2.4589 |
| Mike Stanton        |   232 |  2.4527 |
| Jonathan Papelbon   |    98 |  2.4242 |
| Rollie Fingers      |   241 |  2.3369 |
| Herb Pennock        |   220 |  2.3165 |
| Allie Reynolds      |   322 |  2.3162 |
| Jim Palmer          |   531 |  2.2562 |
| Christy Mathewson   |   406 |  2.1452 |
| Fernando Valenzuela |   271 |  2.0356 |
| Waite Hoyt          |   360 |  1.9368 |
| Bill Hallahan       |   174 |  1.7937 | 

for the WS.

| name_common       | plays | wpa_def |
+-------------------+-------+---------+
| Art Nehf          |   323 |  2.6739 |
| Red Ruffing       |   355 |  2.4589 |
| Herb Pennock      |   220 |  2.3165 |
| Allie Reynolds    |   322 |  2.3162 |
| Christy Mathewson |   406 |  2.1452 |
| Mariano Rivera    |   130 |  2.1409 |
| Rollie Fingers    |   142 |  1.9846 |
| Waite Hoyt        |   360 |  1.9368 |
| Bill Hallahan     |   174 |  1.7937 |
| Bob Gibson        |   316 |  1.7489 |
| George Earnshaw   |   252 |  1.6891 |
| Jack Morris       |   224 |  1.6331 |
| Mike Stanton      |    88 |  1.6320 |
| Lefty Grove       |   206 |  1.5823 |
| Sandy Koufax      |   215 |  1.5088 |
| Chief Bender      |   352 |  1.4593 |
| Carl Hubbell      |   207 |  1.4391 |
| Monte Pearson     |   135 |  1.3549 |
| Tommy Bridges     |   201 |  1.2767 |
| Tom Glavine       |   229 |  1.2428 | 

If you give all baserunning to the batter.

+-----------------+-------+---------+
| name_common     | plays | wpa_off |
+-----------------+-------+---------+
| Pete Rose       |   310 |  2.6047 |
| Lou Gehrig      |   153 |  2.2942 |
| Albert Pujols   |   232 |  2.1907 |
| David Ortiz     |   286 |  2.0459 |
| Charlie Keller  |    80 |  1.9528 |
| Manny Ramirez   |   474 |  1.8563 |
| Babe Ruth       |   171 |  1.8015 |
| Gary Carter     |   134 |  1.7777 |
| Paul Molitor    |   137 |  1.7582 |
| Dave Henderson  |   147 |  1.7356 |
| Thurman Munson  |   141 |  1.5987 |
| Mickey Mantle   |   275 |  1.5897 |
| Lenny Dykstra   |   138 |  1.5804 |
| Edgardo Alfonzo |   147 |  1.5604 |
| Kirk Gibson     |    96 |  1.5600 |
| George Brett    |   191 |  1.5160 |
| Frank Baker     |   108 |  1.5049 |
| Willie Aikens   |    54 |  1.4842 |
| Billy Hatcher   |    67 |  1.3893 |
| Johnny Bench    |   207 |  1.3859 |
+-----------------+-------+---------+ 

WS alone

| name_common    | plays | wpa_off |
+----------------+-------+---------+
| Lou Gehrig     |   153 |  2.2942 |
| Charlie Keller |    80 |  1.9528 |
| Babe Ruth      |   171 |  1.8015 |
| Mickey Mantle  |   275 |  1.5897 |
| Frank Baker    |   108 |  1.5049 |
| Reggie Jackson |   117 |  1.3544 |
| Gene Tenace    |    76 |  1.2398 |
| Kirk Gibson    |    26 |  1.2157 |
| Willie Aikens  |    29 |  1.2046 |
| Tim McCarver   |    89 |  1.1941 |
| Jimmie Foxx    |    73 |  1.1846 |
| Duke Snider    |   154 |  1.1160 |
| Johnny Evers   |    86 |  1.0614 |
| Dwight Evans   |    61 |  1.0251 |
| Gil Hodges     |   151 |  0.9884 |
| Yogi Berra     |   302 |  0.9581 |
| Scott Brosius  |    76 |  0.9326 |
| Pepper Martin  |    61 |  0.9296 |
| Lonnie Smith   |   129 |  0.9024 |
| Mookie Wilson  |    29 |  0.8875 | 

Lack of baserunning obviously helps Lonnie Smith.

16. gator92 Posted: January 16, 2009 at 09:35 PM (#3053916)

I think this could work better as a "story" stat if it were a little less volatile. One do or die game can overwhelm a lot of slightly less crucial games (see Danks in original article).

Also, suppose Danks had pitched exactly the same throughout the season, except that in a couple of his losses, he pitched better and the White Sox actually won those games. Because they were 2 games better in the standings, the final game of the season would be pretty much meaningless, meaning his pickup of 7.5 PPA points in that final game would not have happened. So, Danks pitches better overall, but his PPA score goes down. I don't like stats that get worse when performance gets better.

17. Sean Forman Posted: January 16, 2009 at 09:37 PM (#3053919)

Then I'd like to create a searchable player database of WPA.

What would such a db be able to do?

18. Toolsy McClutch Posted: January 16, 2009 at 09:45 PM (#3053929)

I am a WPA whore.

19. Mike Green Posted: January 16, 2009 at 09:57 PM (#3053943)

#15,

The Rivera number is, of course, an important part of the reason why discussion of the merits of closers begins with him. It is unsurprising to see Schilling and Smoltz next, and the rest in essentially an unhelpful knot.

20. DL from MN Posted: January 16, 2009 at 09:59 PM (#3053946)

Terrific, I figured things were that far along at least. I'm asking mainly as a HoM voter. In most HoM voting the postseason results are basically ignored or added as a fudge factor. I'd love to add in the WPA results for postseason performance for my HoM evaluations, probably zeroing out any negative WPA. I've hesistated giving postseason credit because there was no easy way to do it without potentially missing someone and because it was inconvenient to calculate at best.

21. DL from MN Posted: January 16, 2009 at 10:02 PM (#3053949)

One thing I would also probably do is normalize the postseasons to a single "pennant". Mariano may have helped win 10 postseason games but he only contributed to 1 pennant per year.

22. "Soiled Hands" Zoid Posted: January 16, 2009 at 10:41 PM (#3053990)

The normalization to 1 pennant is a good idea. Right now, I'm just looking at World Series WPA and giving guys extra credit for the increasing difficulty to win that 1 pennant as you add playoff series . The math is beyond me but I would love to check out the results. Much thanks to Mr. Forman for those lists and everything else.

23. DL from MN Posted: January 16, 2009 at 10:51 PM (#3053997)

If Mr. Forman added the postseason WPA to the baseball reference pages for each player I would be most happy (like $50 in page sponsorships happy).

24. jcnyc Posted: January 17, 2009 at 12:01 AM (#3054044)

Someone send this to the HOF voters so they have evidence of Jack Morris' WS clutchiness!

25. studes Posted: January 17, 2009 at 12:42 AM (#3054057)

Thanks for the link.

DL, amstat.org used to have a page about World Series MVP's but I can't find it. But it only went back a few years.

Exactly right, GGC. That was Jay Bennett's work. He was one of the early pioneers in WPA (though he called it something different). Wrote a paper in the early '90's using WPA to show that Shoeless Joe didn't throw the World Series.

I multiply a play's WPA times the amount that each game changed the winning team's postseason odds, according to Baseball Prospectus.

As I said in the second article, I have a problem with this approach. If two teams are fighting it out at the end of the season and they both win, then their playoff odds barely change, yet the players did perform in "high drama" situations. In this case, I think of WPA as WPA/LI, and the Drama Index as LI. I multiply the two to get PPA.

26. Zach Posted: January 17, 2009 at 12:51 AM (#3054061)

It would be neat to see Holliday's numbers from 2007. He was clutch all the way down the stretch and every game was must-win for a month. His headfirst slide into home in game 163 is probably the most thrilling play I'll ever see in person.

27. Walt Davis Posted: January 17, 2009 at 01:19 AM (#3054071)

The Rivera number is, of course, an important part of the reason why discussion of the merits of closers begins with him. It is unsurprising to see Schilling and Smoltz next, and the rest in essentially an unhelpful knot.

Well ... maybe. The Rivera number is a counting stat -- the Yanks have played a lot of playoff games over the years. You'll notice he's only 6th on the WS list. Jack Morris pitched in just 3 WS, very badly in one of them and it's probably that lousy series which keeps him from finishing ahead of Rivera on the WS list. Stanton and Rivera both have 6 WS and 20 appearances (Rivera has 8 more IP) and Stanton's "just" .5 WPA behind.

And it just as easily showcases the silliness of the WPA stat. Bob Gibson had 9 WS starts, 8 complete games, 81 IP, 7-2 record and a 1.89 ERA -- hell, the man even has 2 WS HRs. If only he'd been able to do more to help his team win the WS, we could speak of him in the same breath as Mariano Rivera. :-)

28. Scott Lange Posted: January 17, 2009 at 01:22 AM (#3054072)

His headfirst slide into home in game 163 is probably the most thrilling play I'll ever see in person.

I bet it would've been even better if it had actually happened.

29. DL from MN Posted: January 17, 2009 at 04:13 PM (#3054279)

I think you have to take Mariano's numbers and essentially divide them by 3 (12 playoff wins required for a pennant versus 4 for Gibson) to compare to Bob Gibson. I wouldn't use this as much more than a bonus but it is nice to be able to quantify the bonus and apply it fairly. Jack Morris +2 pennant wins gives some more separation between him and Dennis Martinez. Orel Hershiser might get into the conversation again with his numbers. As I said before I would probably zero out a lousy performance in a given pennant series. I don't think you lose credit for being in the postseason.

You must be Registered and Logged In to post comments.

<< Back to main

Friday, January 16, 2009

Studeman: Postseason Probability Added

Reader Comments and Retorts

Support BBTF

Bookmarks

Hot Topics