Behind the Numbers: K-Rod and the Saves Record
Baseball July 17th. 2008, 11:15am
Anaheim closer Francisco Rodriguez (aka, K-Rod) has a major league-leading 38 saves at the All-Star break. The record is 57 set by by Bobby Thigpen of the White Sox in 1990. With just 67 games remaining in the season, and K-Rod needing just 20 to break the record, what are the chances he pulls this off? Cornell student Max Wasserman, our numbers specialist, explores K-Rod’s chances.
Closers are to baseball teams what Velma is to Scooby-Doo. They don’t really add much but everyone remembers their name.
Okay, sorry for the analogy, but I’m behind on studying for the GRE. Anyway, this week’s adventure into the maths involves closers, or more specifically, Francisco Rodriguez. K-Rod is the best relief pitcher with an incongruous initialized nickname since Byung-Hyun “B.K.€ Kim. But unlike Kim, Rodriguez is good outside of any categories. At the all-star break, Rodriguez has amassed a staggering 38 saves in relief for the Angels, that’s 10 more than second-place George Sherrill. Thirty-eight saves through 95 games puts Rodriguez on pace to shatter the record of 57 saves in a season, set by Bobby Thigpen for the White Sox in 1990.
However, a Rodriguez record-breaking season may not be quite so inevitable. Save opportunities, in most cases, depend on a team only having a slight lead in the last inning. And in the long term, close games, especially one-run games, come down to luck over anything else. This year, the Angels have been by far the luckiest team, with an 18-13 record in one-run games and an 18-5 record in two-run games. At the All-Star break, the Angels were a major-league greatest seven games above their Pythagorean projection. It would be quite unlikely that the Angels will continue their close-game-winning ways in their remaining 67 games. But that is what they will need to do in order to give Rodriguez a shot at the record. In 1990, the White Sox finished the season seven games above their Pythagorean expectation as Thigpen set the saves record. In 2002, when John Smoltz set the NL record with 55 saves, the Braves finished five games above expectation. And when Eric Gagne tied that record the next season, the Dodgers finished a relatively unlucky two games better than expected.
So the task is to predict how many saves K-Rod will have at the end of the season. Now, here’s where I describe the math techniques I used, so if you don’t want to read it, don’t care, or Wheel of Fortune is on, feel free to skip this section. I’ll let you know when you can pick it up again. Also, I’ve decided to use format “K’Rod†for Rodriguez’s nickname because it makes him sound more like a Klingon.
Math
To obtain an estimate for the number of saves Rodriguez will have in the Angels’ remaining 67 games, we need to determine the probability that he will get a save in any game. This probability is dependent on the Angels entering the ninth inning with a one, two, or three-run lead and Rodriguez not blowing the save. Both runs scored per game and runs allowed per game are simulated by Weibull distributions, a fact that was proven by Brown Professor Steven Miller, who also derived James’ Pythagorean formula. Using these two Weibull distributions adjusted for the number of runs scored and allowed per game for the Angels this season, we can determine three fairly accurate probabilities for the Halos entering the last inning with a one, two, or three run lead. And since probably nobody is reading this, I can type whatever I want Opaque Llama Uterus. [Ed. We love this guy!]
The probability distribution for runs per inning was determined by Baseball Prospectus’ Keith Woolner, and it resembles an exponential distribution, though not perfectly. To use Woolner’s distribution, we need an estimate of Rodriguez’s true ERA. We would normally use Bayes’ theorem to do this using Rodriguez’s current ERA (2.36) and his predicted PECOTA ERA (2.79). However, the theorem requires a distribution for ERA, and as far as I could research, none has been determined. In lieu of a Bayes predicted ERA, the mean of those two ERAs (2.575) is used.
The resulting probability of Rodriguez getting a save in any game is found to be 0.243. Multiplying that value by the number of remaining games gives and expected value for number of saves Rodriguez will get between now and the end of September. The expected value was adjusted still more though by factoring in games that go to extra innings. The probability of Rodriguez getting a save in an inning past the 9th is 0.023. Multiplying this probability to the number of remaining Angels road games, 32, and adding it to the previously found expected value gives us our prediction. Obviously, more could be done in the prediction since we did not take opponents, stadiums, specific players, or other methods of getting saves into account, but Temperate Soul Reaver Tittie Garmin Chipotle. [Ed. Love Chipotle!]
End Math
The expected value for the number of saves K’Rod will earn after the All-Star break is found to be just under 17 (16.98 to be exact). This means that the prediction for Rodriguez is he will finish the season with 55 saves, two short of Thigpen’s record. Obviously, this doesn’t mean that he can’t break the record. However, he will need his team to continue to play just-good-enough baseball for the rest of the season. The Angels do play the Mariners 10 more times this season, so I guess anything is possible.
Still, I can see how tough the life of a closer must be. Some don’t get many chances to pick up saves and they need to produce when they get that chance. If a closer blows even a couple of saves in their few chances, something that could be chalked up to random chance, they may lose their spot in the bullpen. And yet despite this grim reality, Armando Benitez has somehow managed to keep getting closing jobs. That’s one thing I’ve never been able to figure out. I’m guessing he must give a really great PowerPoint presentation to GMs.
44 Responses to “Behind the Numbers: K-Rod and the Saves Record”
Leave a Reply
You must be logged in to post a comment.

July 17th, 2008 at 11:20 am
o-ver-rated!
July 17th, 2008 at 11:20 am
what are the chances he pulls this off?
80%.
nice post iggy. GO IC
July 17th, 2008 at 11:22 am
wait, we’re not talking about papelboner?
July 17th, 2008 at 11:23 am
Great Stuff.
July 17th, 2008 at 11:24 am
“Opaque Llama Uterus” will be my FF team name. Thank you and god bless.
July 17th, 2008 at 11:26 am
Dwight: What is the capital of Maine?
Andy: The capital of Maine is Montpelier, Vermont, which is near Ithaca, New York, where I went to Cornell.
Dwight: Okay. Also, moratorium on Cornell-talk. Don’t want to hear about it. Forget your personal history and learn the history of this company.
Andy: That should not be a problem. I minored in history at the Ivy-League school which I attended.
July 17th, 2008 at 11:26 am
Scooby Doo, Klingon, Chipotle, and Wheel of Fortune jokes. TBL owes you VIP passes to ComicCon for this one.
July 17th, 2008 at 11:27 am
Please explain, thanks.
July 17th, 2008 at 11:30 am
If every lead in to Max’s posts didn’t include “Cornell student”, I’d give his mathletics zero credibility.
July 17th, 2008 at 11:32 am
Doesn’t everything in life involve luck. I was lucky to wake up this morning. I was lucky to make it work safely. I was lucky the IT didn’t find out about all the NSFW images I have been looking at. Luck Luck Luck. I am cool with all the math (TBL IS ESTATIC…HE LOVES HEARING SMART SOUNDING WORDS!) it is well done, but the reason why not anyone can be a closer and why Mariano Rivera is the greatest of all time is because of their ability to handle pressure. Pressure does exist. They are all humans and talent is only part of the equation. But you are right about saves being a stat predicated on your team. Pretty much every pitching stat except strikeouts is predicated on your team and luck.
July 17th, 2008 at 11:34 am
I find it humorous that James’ formula, Bayes’ theorem, and Woolner’s distribution were used to arrive at a prediction for saves. I wonder what William of Ockham would have to say about all this. Speaking of which, do you think he had a complex theory as to why his razor was spelled like that?
/what are we talking about again?
July 17th, 2008 at 11:35 am
AndyMax Wasserman: The capital of Maine is Montpelier, Vermont, which is near Ithaca, New York, where I went to Cornell.AndyMax Wasserman: That should not be a problem. I minored in history at the Ivy-League school which I attended./fixed.
//iggy, we kid because we love. don’t fire retorts from your blackberry. just enjoy the elliptical machine you’re probably on right now.
July 17th, 2008 at 11:35 am
one of the reasons i got into sports (specifically baseball) in my formulative years was to get away from math.
the way baseball coverage has morphed since, i now have nightmares that the nuns of my youth have returned to torment me for life.
(may bill james rot in hell).
July 17th, 2008 at 11:38 am
For the record, TBL is usually the one who mentions the whole Cornell thing.
July 17th, 2008 at 11:39 am
Being a Cornell student is pretty overrated, I know a few… they just happened to have more money than most of our friends hence they could go to that school. That being said, you sound like you got your shit together so nice work.
Mike NYC I hate you.
July 17th, 2008 at 11:42 am
Oh, im definitely the one who throws in Cornell. I guess i could say college kid.
July 17th, 2008 at 11:42 am
Hey, thanks phat!
/um, what?
July 17th, 2008 at 11:45 am
Good stuff Max, and just so y’know, I actually enjoy reading the mathematical procedure behind it all, so thanks for taking the time to explain that all out
July 17th, 2008 at 11:46 am
what a colossal waste of time.
July 17th, 2008 at 11:49 am
Max’iggy… we all know it’s TBL who sets you up like that. TBL, don’t change a thing. Since 87% of the people here don’t understand his posts, it gives us something else to talk about. or at least a reason to re-post the dwight/andy exchange. for that, we thank you.
July 17th, 2008 at 11:50 am
@Mike: The major factor in determining a team’s win/loss record is runs scored and runs against. After all, that’s why the Pythagorean expectation has proven so accurate. Whenever there are deviations from the expectation, as is the case with the Angels, it is due mostly to record in close games. And close games tend to be more like coin flips than be affected by skill. The 2003 Tigers, for example, had a winning record in one-run games. It has been shown that teams with excellent pitching, like the Angels, do better than average in one-run games, but it still comes down to luck a lot of the time.
July 17th, 2008 at 11:51 am
Wow, iggy and me are not on the same intellectual level. Iggy you ever think about doing the math and gambling on futures at vegas? Or online?
July 17th, 2008 at 11:54 am
I also love Chipotle. And titties. Not so much llamas. Goog kob with this post. I feel like a lot of people assume that because he has 38 saves at the all-star break, they double his total and assume he will finish near 76. Not quite the same level of math.
July 17th, 2008 at 11:55 am
goog kob = good job
July 17th, 2008 at 11:56 am
Thanks, Max. I don’t really agree with this: “And close games tend to be more like coin flips than be affected by skill” — but appreciate the explanation.
July 17th, 2008 at 11:57 am
Would all the math in the world tell you before the season started that he would have 38 saves at this point? If not, then why are the remaining games any different.
July 17th, 2008 at 11:57 am
Yancey Thigpen can be convinced, if given the proper perspective.
July 17th, 2008 at 11:57 am
@phat: You know, just because three-quarters of Cornell comes from Long Island doesn’t mean we all do. Now if you’ll excuse me, I have to go meet Gatsby and Daisy for tea.
July 17th, 2008 at 12:02 pm
No in-between today, folks. Its either TBL and sportsgal acting like Entertainment Tonight, or Wasserman giving us the TBL version of the 11pm program on PBS.
I think I need a nap.
July 17th, 2008 at 12:05 pm
July 17th, 2008 at 12:05 pm
Well that was bad huh?
July 17th, 2008 at 12:07 pm
@nwilson: Problem is, Vegas is doing the same math and setting the odds accordingly so that it’s still basically random. And who needs Vegas anyway? Turning Stone is only 90 minutes away from Ithaca and they get all the best Dave Matthews cover bands to play there.
July 17th, 2008 at 12:10 pm
@ iggy, a year or so ago espn.com had an awesome article about the behind the scenes sports booking places in vegas and how they set odds and the work that went into setting all sports book odds. I came away amazed. I had just never thought about that before. You realize that it’s not just something fun that you should be doing because you think you know something. Chances are, they know more than you.
July 17th, 2008 at 12:11 pm
YAWN
July 17th, 2008 at 12:12 pm
@iggy: if you finished reading my comment I said,
So I was giving you a compliment on the post regardless of the sentence immediately before that.
Only thing possibly worse than DMB is a Dave Matthews cover band
July 17th, 2008 at 12:15 pm
@nwilson26: hell yeah dude, that ESPN article was solid (can’t believe I just said that…). That article made me laugh especially hard at my friends who said “you have to watch the games to bet them” because the article said that the bookies basically look at box scores and use math to come up with their numbers. Even harder laughs at people said “I bet because its fun” hahaha yeah the bookies think its fucking great…
July 17th, 2008 at 12:20 pm
I bet on some college basketball because well it make the game much more interesting if I don’t give a shit about the teams. Fucking eric gordon. Don’t ever get a 10 second violation again up 3 with :40 left in the game or I will KILL you. Anyway I regress, the world is more complex than people think. That was my lesson after reading that article and realizing that it wasn’t as easy as I first thought. Those spreads are always close.
July 17th, 2008 at 12:26 pm
Iggy…I can take the Math, I can take The Cornell shit, I can take the mild dig at the greatest suburb in all the world (not all Long Islanders are wealthy, only most of us)but please…Please don’t tell me a gambling place is great because of a cover band of Dave Matthews. Dave Matthews is the vomitous garbage that I had to endure during my teenage and college years and for some shitty awful reason because the “Grateful Dead” of the fat college girl wear skorts and sandals. It may be the most boring fanbase of anything ever. I wouldn’t want to hear Dave Matthews whine through All Along The Watchtower none the less people pretending to be him doing it…you are a smart kid…go get yourself a better band
July 17th, 2008 at 12:46 pm
You had to study for the GRE?
/Snickers
July 17th, 2008 at 12:49 pm
This post was very un-Bill. Excellent work, Mr. Wasserman.
July 17th, 2008 at 1:37 pm
Can you calculate the actual probability that K-Rod gets 20 more saves this year (and thus breaks the record)? It looks like he’d have to zoom up to getting a save 30% of the time. What are the probabilities he could do that? It seems pretty small given he’d need 17% more games (67 to about 80) to do so, but on the other hand the Angels have a low-scoring offense, which helps his chances. (Relief pitchers are perhaps analogous to placekickers in that a high Saves/FG total is great…a VERY high Saves/FG total is instead often a sign of team weakness in putting away games in emphatic fashion).
July 17th, 2008 at 3:27 pm
Considering how quickly Thigpen went downhill after the record-setting season, wouldn’t his usage pattern concern Rodriguez and his agent? Rodriguez is hitting free agency – he’ll be a premiere young closer and looking to score a long and lucrative contract. If I am Mike Sciosia, I try everything I can to blow Rodriguez’s arm to get wins (via the saves).
Closers come from anywhere and while you (a baseball team) should want your best pitchers throwing the most innings, I believe there is value in having a defined ninth-inning saves pitcher. I would keep my best relievers for the higher-leverage situations, but who cares who pitches the ninth? Mike Williams was a closer and a good one for a couple of seasons, for My sake.
July 17th, 2008 at 4:57 pm
way to do a post that pisses off giants’ fans like myself. first it’s about a guy who helped kill my giants’ world series chances in 2002 and then you have to mention armando. thanks for pouring salt in my wounds. good write up though.
July 17th, 2008 at 5:10 pm
@Roman: That comment was supposed to be facetious, but the commenting system or whatnot didn’t put in my little “/facetious” tag after it. So yeah, just kidding about things.