Buy The Book
How often do you see someone try and bowl over the catcher these days, a la Pete Rose and Ray Fosse in the 1970 All-Star Game?
How often do you see that when the runner would have been safe by a mile if he had just slid in like 99.9% of players do these days?
Try and guess who made that ignoble and boneheaded play. If you saw or heard of the play, obviously you can’t guess…
For stealing third base in the top of the 9th inning, with no one else on base, 0 outs, and your team down by 2 runs?
That is what Torii Hunter attempted tonight (and he was thrown out).
Possibly the least smart thing I have seen a player do in a long time, at least in a situation where they had more than a split second to make the decision…
Great stuff from Jeremy.
It’ll feel good to retire one day, knowing there’s some a great group of talented and hardworking analysts out there, who have jobs where they can sneak in all this great work.
The internet is as much about adding value to corporate america’s efficiency as it is about removing value by surfing and posting.
I quite enjoyed this Davey Lopes interview on baserunning.
He asks:
Scenario: You are the runner on 2nd base and there is 1 out. The score is 3 to 3 and it is the 4th inning. A fly ball is hit to right center field and the right fielder will attempt to catch the ball. As a runner you have average Major League speed. If you choose to tag, you will be safe at 3rd base, but you will have to slide. The next batter up hits in 7th in your lineup and he is a career .250 hitter with 10 HR’s and 60 RBI’s. What do you do and why?
My answer:
Ensberg gives the blueprint.
Let’s pick one out and see what do you do if you are on second base and 0 outs. My natural inclination is to do what Ensberg says. Let’s work it out.
Runner on 2B, 0 outs is run expectancy of 1.189 runs. Let’s say it’s a 90/10 play, where there’s a 90% chance of an out, and a 10% chance of a hit or error.
Option1: You tag up, and move to 3B on the hit/error (gives you RE of 1.904). And when there’s an out, you’ll be able to advance 10% of the time. So, 81% you are stuck at 2B (RE of 0.725) and 9% of the time you get to 3B (0.983). Add it up, and the RE is 0.866 if you tag up.
Option2: You move half way, and react. On the hit/error, you score (gives you RE of 1.953… notice how this number is not very different from just staying at 3B). On the out, let’s make it that you get back safely 87% of the time (0.725), and are doubled-up 3% of the time (0.117). Add it up, and the RE is 0.825 if you move half way. Even if you are NEVER doubled-up, your RE is still only 0.848.
So, here are the breakeven points: if you tag up and there’s NO chance for you to advance, it’s the same thing as going halfway and having a 1% chance of being doubled up. So, being doubled-up at the expense of scoring is a huge killer. Huge.
That’s on a play where there’s a 90% chance of the batter being out.
***
What about on an almost sure hit? If you wait to tag up, the RE is 1.79. If you move halfway, you can get doubled-up SEVENTY percent of the time, and STILL break even.
Conclusion: the level of your lead off 2B on a flyball is going to be directly influenced by the chance that the ball is going to fall in. I think this was obvious to most of us, but I was surprised to the extent that you can just take off. This goes to what to do on line drives. If on a low line drive and the infielder has a 10% chance of making an out, you can only take off to the extent that you’d be able to make it back to 2B at least 30% of the time.
I didn’t realize I’d be so fascinated by something that seems so implicitly benign.
I *know* we’ve talked about this, and I know we’ve come up with a good conclusion. I just want to point out this:
Using the “veer to the right just before the base” approach, a typical player will take 22.2 seconds to complete an in-the-park home run. That same player using the optimal strategy of a constantly curving path can round the bases in 16.7 seconds.
22.2 seconds? How old is this typical player? The diameter of a circle that encompasses the 4 bases is 127.3 feet (square root of two times 90^2… see kids? math is useful). The circumference of a circle is 2PIr, or PId (that’s pi times diameter… see kids? listen to your math teacher, and you can write a blog like I do). So, a circle path is almost exactly 400 feet. We know that Olympic runners run 100m (328 feet… kids, please, stay in school) in around 10 seconds. 400 feet would be 12 seconds (technically you’d want to separate the first 10 m from the other 90 m, but we’re looking for accuracy, not precision). Dirt, cleats, uniform, base-touching, non-Olympic-speed runners.... bump that up to 14 or 15 seconds if you like. Who’s running 400 feet in 22 seconds?
Opposing coach:
“It took two incredible plays to get them. I understand why they were being aggressive. It took two absolute perfect plays to get people so more times than not, they’re going to be safe. It just took two incredible plays.”
If it takes a perfect throw to get the runner out, it was likely that the decision to send the runner home was correct. How often do you get the perfect throw? 10% of the time? 15%? As long as the breakeven point is at least 85%, and you reason that you need a perfect throw to get the runner out, then you send the runner, and you don’t feel bad about it.
I did not see the play, nor have I figured out the breakeven points for the play in question. Someone want to break out The Book and look at the corresponding chart?
Two academicians say yes, and Phil Birnbaum says no.
That’s what Mike Lowell said. Actually, we can put it in. This is what the win expectancy matrix was set up for.
Let’s look at this chart. It’s the bottom of the 12th (look at the bottom of the 9th), the batter just flied out to make the 2nd out, and Scutaro would 99% of the time remain on 1B. That sets the win expectancy at .562. But Scutaro actually went to 2B, for a win expectancy of .610. So, we credit Scutaro with +.048 wins.
There you go, Mr. Lowell, it is now officially in the boxscore. If it’s tangible, we can measure it. And Scutaro going to 2B is tangible and measurable.
All we need to do is get the scorers to make a better notation in the boxscore so we can separate these plays better. If it was a deep fly ball that any runner would have made it, we give 100% of the credit to the batter. If it was a flyball that essentially makes the Scutaro play like a SB attempt, we give 100% of the credit to the runner. The tough ones are the in-between plays where the split is not so easily done.
By the way, had Scutaro been thrown out, the odds go down to .500, or a drop of .062 wins. So, it’s a very heads up play by Scutaro, as he only needed to be safe 56% of the time to breakeven. So, the issue is not that Scutaro went for it, but rather, why don’t more runners go for it on those plays? As Scutaro said: do or die.
***
Also, someone else pointed out to me that they IBB the next batter, raising the win expectancy from .610 to .613. Why did it go up, if the lead runner wins the game in either case? And, having a runner on base now makes the outs at three bases possible. I didn’t look into it, other than guess that it’s because two walks wins the game.
Jeff.
In an epilogue, I just love this stat:
Last year, when there was NOT a man on on second, runners on first scored 42.6% of the time on a double. When there was a guy on second, runners from first scored 45.3%. You read that right, there is no such thing as clogging the bases.
That stat was from the BJ Handbook.
There’s dozens of saberists who do this, and there’s dozens of ways, none of which more right than any other. And in the end, on a career level, it’s not going to matter. To the extent that we want to discuss this, let me tell you how Sean does it and how I do it, and then we can see if we can come up with a consensus.
Let me give you Sean’s method, since it’s so simple to explain: give the entire change in run expectancy to the lead runner whose base-state changes. So, 1b/2b, double-steal: runner on 2b gets the full credit. You’ll get a little quirk with runners on the corners, the runner on 1B steals, there’s an error, and the runner on 3B scores: the runner on 3B gets the full credit.
My method is to give the entire change in run expectancy, regardless of who actually moves off the base, in this way: always to the lead runner who is on base, except if it’s a 1b/3b situation (in which case, the runner on 1B gets the full credit). In Sean’s case and my case, regardless of who gets thrown out, the charge goes to the designated runner.
You can also simply split the credit to all runners on base. I think this is what Fangraphs does (or did?). I used to do this.
As I said, you can make a reasonable argument in anything discussed, here or whatever will follow. I’m more interested to see if we can come with a consensus or not.
With David Lauria giving us all the good questions.
I’ve always had it on my to-do list, and, well, Max did it. But, I’m skeptical of the results as it pertains to Brock:
Baserunners: Willie Wilson, Vince Coleman and Lou Brock were the best once you factor in the battery against which they were stealing (Wanna know about Rickey? Fifth, behind Tim Raines; best among the active players? Carl Crawford, barely missing the top 10); finally the worst three on the basepath, Retrosheet era, are Duane Kuiper, Minnie Minoso and Greg Gross.
As I wrote in the comments:
Great stuff Max. I’d be interested to hear more about Brock v Raines. The gap between the two is 130 SB and 161 CS. It’s hard to believe that Brock can be in the same ballpark as Raines. What you are proposing is that Brock ran against much tougher pitcher/catchers.
Pizza Cutter has an article at ESPN and BPro, and, I wonder if he made a faux pas:
Without going into all the algebra, it says Perlozzo needs to be 73.2 percent sure that Rollins will make it before he sends him. So, if third-base coaches leaguewide are playing the game correctly, we should see that about 73 percent of the runners in this situation wind up scoring.
No, that is incorrect if “this situation” is man on 3B and less than 2 outs. The MINIMUM success rate has to be 73.2%, meaning that if the third base coach is 73.2% to 100% sure of making it, then you send him. So, the average will be more than 73.2%. How much more? Well, there are tons of gimmes here. Let’s say that one-fourth of the time you have a gimme (say 96-100% success rate), one-fourth of the time you have a great chance of scoring (say 83-87% success rate), one-fourth of the time are the borderline plays (say 70-76.4% success), and one-fourth you have a less than 50/50 chance (say 0-50% success rate).
So, if this is true, then this is what happens:
98% success x 25% frequency x 100% attempt
85% success x 25% frequency x 100% attempt
73.2% success x 25% frequency x 50% attempt
25% success x 25% frequency x 0% attempt
The average is 87.8% for this illustration. For that third case, you can argue that you should run 0% or 100% of the time, and it won’t matter because it’s so close that it’s borderline. And therefore, the success rate if you never run on the borderline plays will now become 91.5%, and if you always run on the borderline plays, it’s 85.4%. That is, under this illustration, your run expectancy is maximized when your overall success rate is anywhere from 85.4% to 91.5%.
On the other hand, if he means “this situation” that includes the distance of flyball and speed of runner, such that these parameters would lead to a 73.2% success rate for these subset of plays (my third line in the 4-line illustration), then okay. That’s not how I read the line the first time, which is all well, because it gave me a chance to make the point. Re-reading it, I guess Pizza could have meant it the right way.
The rest of the article doesn’t go back to Pizza’s quoted line above, and is an otherwise good piece of trying to show that the thirdbase coach is too conservative. The knockout is this line:
Here’s an interesting one: What would happen if third-base coaches just sent everyone, playground-style, on these potential sac flies, regardless of whether it was a good idea? It turns out that teams probably would score more runs than they do now.
Indeed, it was rare that it was a bad idea to send the runner, even after controlling for the distance of the fly ball and the speed of the runner. It was almost always the case that the chances of the runner succeeding were above the break-even point.
The article audience doesn’t lend itself to Pizza presenting his evidence here. Maybe he’ll show it to us, though I have a vague recollection that we talked about this when he was at statspeak.
In his career, he has 83 SB and 11 CS. Last year, he was 23-0, as Tommy points out. I remember many many many years ago, when I was a teenager, Raines was still in Montreal, and I just got my first computer. One of the first things I did was try to calculate the breakeven point for steals, by quality of stealer. For example, let’s say that Tim Raines tried to steal only one base all season. Obviously, he would take the most favorable one, the one where the pros outweighed the cons the most. I would give him .999 SB and .001 CS. Now, what if he stole in the two most obvious steal situations. Maybe in this second one, he would be successful .995 times and CS .005 times. See where I’m going here?
So, I fired up my Lotus 1-2-3 spreadsheet, and put in some numbers and tried stuff out. I tried to get to 70 SB and 9 CS, which is a Raines staple. In order to get there, I started him at .999, and dropped by .003 for each steal attempt (0.00289548847776695 if you want to be Excel-exact). By the time I got to the 79th steal attempt, I was at .773 SB. The sum of these 79 was 70 SB. So, I reasoned, Raines was still leaving steals on the table, based on this analysis. What if I kept going? On the 80th steal attempt, he’d be at .770 SB. By the time I get to his 104th attempt, he’s at .701. Basically, we might say that he reaches his personal breakeven point at the 104th attempt. And, adding up all of these steals gives us 88 SB and 16 CS. In some sense, I’m saying that a guy who steals 70 and gets caught 9 times would steal 88 and get caught 16 times.
Interestingly, Raines once had a 90-14 season. Sweet, right?
And, what if Raines decided that his body couldn’t take it, and instead, attempted only 55 steals? In that case, I simply keep the same pattern, but stop him on the 55th attempt. In that case, the sum of his 50 best steal situations is 51 SB and 4 CS (actually 50.6, 4.4). And, interestingly, Raines once had a 50-5 season. I know, I know. I love this stuff, especially if I can apply it to Raines.
Back to Utley. He averages about 15 steals and gets caught 2 times. In order to get that, I can start him as a .999 base stealer, like Raines, but drop him by .0146 after each attempt (about 5 times the degradation rate of Raines), so that he ends up at 15 SB, 2 CS. His last attempt has a success rate of .766. This is what his chart looks like:
Attempt Success
1st 99.9%
2nd 98.4%
3rd 97.0%
4th 95.5%
5th 94.1%
6th 92.6%
7th 91.1%
8th 89.7%
9th 88.2%
10th 86.8%
11th 85.3%
12th 83.8%
13th 82.4%
14th 80.9%
15th 79.5%
16th 78.0%
17th 76.5%
TOTAL 15.00 SB
If I extend it until his last attempt is just over 70%, I get 18 SB, 3 CS.
18th 75.1%
19th 73.6%
20th 72.2%
21st 70.7%
That basically becomes the maximum attempts he should make. UNDER THIS ASSUMPTION. There’s no reason I needed to start him at .999. I could have started him at .9055, and dropped him at .003 like Raines. In that case, his 17th attempt is at 85.9%. In order to get to 70% as the marginal rate, I’d have to extend him to 72 attempts (and he’d end up with 58 SB and 14 CS). As you can see, it depends.
I have to believe that runners are mostly running at their optimal frequency levels, if not pretty close to it. Anyway, I spent (or wasted) lots of hours trying different runners like this. It was loads of fun.
Crashburn gives it to us:
It’s highly unlikely that the Phillies lucked their way into teams as consistently elite as their defensive squads have been.
...
From 2002-07, the Phillies were either first or second in the NL in drawing walks.
...
I’ll conclude this with perhaps the most damning bit of evidence that the Phillies are Sabermetrically-inclined: base running.... the Phillies have been not only elite but once again consistently elite. It’s one thing to have a fluke season here and there but the Phillies are incredibly consistent.
...
The Phillies’ success rate on the base paths will astound you:* 2004: 79%
* 2005: 81%
* 2006: 79%
* 2007: 88%
* 2008: 84%
* 2009: 81%Once again, not just elite, but consistently elite.
Whoah. Really? From 2004-2009, the Phillies have stolen 701 bases and been caught 151 times. Tim Raines for example is 808/146. Joe Morgan is 689/162. Kenny Lofton is 622/160. Willie Wilson is 668/134. These 4 guys averaged 697 bases and caught 151 times, numbers virtually identical to the Phillies. Imagine that. The Phillies as a team, as efficient base stealers are somewhere between Raines, Morgan, Lofton, and Wilson.
Aug 31 15:28
Fans Scouting Report: Update
Sep 02 15:54
The two uncertainties of UZR
Sep 02 15:17
Mail: rWAR v fWAR
Sep 02 14:59
Roger Federer
Sep 02 14:59
It’s hard to beat the crowd (Vegas in this case) no matter how smart you think you are
Sep 02 14:57
Could Rob Dibble have been a comp for Strasburg?
Sep 02 14:15
WOWY Teachers
Sep 02 13:37
Who’s Waldo?
Sep 02 08:36
Team Elin
Sep 02 01:19
Can someone tell me why Trevor Hoffman is still allowed to pitch?
THREADS
August 31, 2010
Fans Scouting Report: Update
September 02, 2010
The two uncertainties of UZR
September 02, 2010
More on Morris and Cox, bloggers / editors
September 02, 2010
It’s hard to beat the crowd (Vegas in this case) no matter how smart you think you are
September 02, 2010
Roger Federer
September 02, 2010
Ryan Howard: Mr September
September 02, 2010
WOWY Teachers
September 01, 2010
Jose Bautista
September 01, 2010
Workload Regularity Score
September 01, 2010
Strasburg II
Recent comments
Older comments
Page 1 of 199 pages 1 2 3 > Last »Complete Archive – By Category
Complete Archive – By Date