Vidro, steal maker
A math-filled exploration into how serious Vidro’s DP problem is. Stats drawn from ESPN’s splits, Fangraphs, and Baseball Prospectus.
Run Expectation (2006)
Runner on first, no outs: .93 runs
Runner on second, no outs: 1.15 runs
No runners on, two outs: .11 runs
32% of balls Vidro puts into play go for hits
49% of balls Vidro puts into play are grounders
Using what I can get out of ESPN’s splits:
He had 30 PA with someone on first
He had 13 PA with runners on first and second
He has 4 PA with runners on first and third
= 47 PA where GIDP was a possibility if there were zero or one out.
He came up with runners on with two out 20 times, and runners on 74 times. That’s 27% of the time (I thought it would be 33% too)
Of those 47 PA, ~13 should be with two outs, so he came up 34 times with a DP possible. He hit into a double play 7 times, so about 20% of the time he has the chance, he’s GIDPing. That doesn’t even count situations where the runner went on the pitch in order to try and stay out of the DP.
For the sake of simplicity, I’m going to use “runner on first, no outs” for the run calculations here.
Man on first, no outs, Vidro swings away
7.3% of the time, he walks (.93 runs expected –> 1.58 runs = + .63 runs)
25% of the time, he singles advancing the runner one base* (.93 runs expected –> 1.58 runs = + .63 runs)
7.2% of the time he strikes out (.93 runs expected –> .56 runs = -.37 runs)
20% of the time, GIDP (.93 runs expected –> .11 runs expected = -.82 runs)
41% of the time, he’s out without wiping the runner out by popping out, flying out, or hitting into a fielder’s choice (.93 runs expected –> .56 runs = -.37 runs)
*using his runners on split here, a .246 average, which is low
Average value of a Vidro PA with a runner on first: -.14 runs
Yaaahoooooooooo! Making some pro-Vidro changes: assume he hits .300, that his non-GIDP outs advance the runner, say, a third of the time, put the power in… I came out at -.1 run/PA with a runner on.
Every time Vidro is up with a runner on first and no outs, the most optimistic scenario I came up with is that he costs the team a tenth of a run.
Vidro is the team’s designated hitter.
The Sacrifice Strategy
Say there’s a runner on first, and Vidro lays down a bunt every time to advance the runner at the cost of an out, and he’s 100% successful at laying down that bunt.
Bunt moves the runner to second: .93 runs –> .74 runs = -.19 runs/PA
You could have a reasonably practiced pitcher pinch-hit for Vidro, lay down a sac bunt, and they’d only be a little worse than Vidro in hurting the team’s chances to score. Vidro is the team’s designated hitter.
All Stolen Base Variant
Ichiro’s career stolen base rate is 81% (244/303). Why not, if he’s on first, have him steal every time to remove the double play? Assume, for purposes of this calculation, that having every other team know that he’s going reduces his success rate substantially.
66% of the time, Ichiro reaches second (+.23 runs)
33% of the time, he’s thrown out (-.64 runs)
= you lose .04 runs/attempt
However… then it changes the whole Vidro at-bat:
Vidro up, Ichiro on second
(using the more pro-Vidro %s)
7.3% of the time, he walks (1.15 runs expected –> 1.58 runs runs = + .43 runs)
30% of the time, he singles advancing the runner one base (1.15 runs expected –> 1.81 runs = + .66 runs)
7.2% of the time he strikes out (1.15 runs expected –> .74 runs = -.41 runs)
20% of the time, out that advances the runner (1.15 runs expected –> .96 runs expected = -.19 runs)
36% of the time, he’s out without wiping the runner out by popping out, flying out, or hitting into a fielder’s choice (1.15 runs expected –> .74 runs = -.37 runs)
Every Vidro PA is now +.03, a swing of at least +.13 runs/ab just by removing the double play
No steal, Vidro’s at least -.1
Steal works, Vidro’s now +.03
Steal fails, Vidro’s PAs also become worth ~.04 (you can just take my word for it or work it out yourself)
The gap between having Ichiro successfully steal second to remove the double play is about the same run value as having Vidro bat normally and having a pitcher lay down a sacrifice bunt in the same situation.
Having Ichiro steal then carries a substantial value to the team above the value of moving 90 feet, by removing the possibility of the double play. So much so that it’s a good move to have Ichiro steal second even if his success rate will be significantly lower: sketching this out, it looks like it’s a good move to have him run all the way down to about 60% success, where normally you want a 75% success rate from your first-to-second stolen bases.
Consider that for a second: our designated hitter is so slow, so unproductive, that it makes sense for the team to pursue what would under normal circumstances be detrimental strategies because they are less harmful than having our designated hitter hit normally.
I sketched this all out a couple of hours, now that I look at the clock, and I still can’t believe that that’s right, and yet… there it is. If there’s a better way to approach the problem, I’d love to hear it. Or, as we say, patches welcome. For more, check out the Fangraphs play log for Vidro.
Anyone mentioning Snelling in this comment thread will be banned. This isn’t about Vidro/Snelling.
Besides this mention, obviously.
What I find amazing is that Hargrove seems to be aware of all this, according to comments of his cited in Baker’s blog, and yet he doesn’t seem inclined to move Vidro in the order so he’s less of a danger to Ichiro.
“You could have a reasonably practiced pitcher pinch-hit for Vidro, lay down a sac bunt, and they’d only be a little worse than Vidro in hurting the team’s chances to score.”
This gives me an idea of a handy way for Weaver to earn his paycheck when he gets back from the DL…
Wow…this has to be a sample size fluke, right? Do you think is DP rate is sustainable?
Nice analysis, but isn’t this all just small sample size theatre? I’d like to see these numbers run again at the end of the year… Sadly, I can actually see them changing for the worse.
A question I’ve been meaning to ask for a while that’s only tangentially related to all this:
This has to do with run expectancy. At what point is a batter hitting so poorly, that bunting the runner over becomes the better move? If I’m hitting .100 with a lot of strikeouts, there’s no way that bunting the runner over doesn’t become the better strategy, right?
That’s always been my problem with run expectancy tabulations and win expectancy charts; they’re fun to look at, but for the percentages to be accurate in a real game scenario, every single ballplayer on the field on both teams has to be exactly average. My suspicion is that the range of talent on every team is so great that as to render the run expectancy percentages close to useless in a given real-game situation.
[Please don’t shoot me down for being some sort of stat-averse luddite. I’m not even close to that.]
By the way, Vidro still remains very much on target to threaten Jim Rice’s GIDP record of 36, set in 1984.
Do you think is DP rate is sustainable?
Of course not. As the season goes on, and he gets hurt and worn down and slower, his GIDP will get higher. By September, it should be approaching 100%
Using the fangraphs play log just to double check Derek’s work, there have actually been 40 GIDP opportunities for Vidro this year. He’s grounding into a DP 17.5% of the time he has the chance to do so. This sounds bad, but it’s not even close to the league lead.
Among players with 100+ PA, Vidro ranks 67th in DP%. Jason Kendall has hit into a DP 30% of the time he has a chance. Brandon Phillips is also killing the Reds, hitting into a league high 11 double plays, 28.9% of the time has a chance to be doubled up.
Other Mariners DP%: Kenji Johjima is at 26.3%, Jose Lopez is at 22.2%, and Richie Sexson is at 18.2%.
Moral of the story: Groundball hitters without foot speed are going to hit into a lot of double plays. They better make up for it in other ways.
Re: Dave
Would the math DMZ applied to Vidro then also apply to Lopez and Johjima or would their secondary average make the “steal every chance” approach wrong?
Their power would make it more advantageous to stand on first base, because there’s a higher chance they’ll hit it over the wall for a home run, or at least drive the ball into the gap for a double.
Since Vidro’s best case scenario is just a single that gets you only to second base, stealing becomes much more appealing with him at the plate, as the patented Vidro single than has a chance to score the run.
8. and 9.
Good catch. Derek’s analysis fails to take in the chance of an extra base hit, which, as miniscule as it is in Vidro’s case, would still skew the numbers slightly.
Dave and Derek: Does it stand to reason that Ichiro’s 81 percent steal rate is at a heavy risk of going down because he’s being “forced” to attempt steals in situations in which the circumstances for a steal are less than ideal? (i.e., lefty with a good move on the hill, good defensive catcher, power pitcher with good slide-step, etc.)
Derek’s analysis fails to take in the chance of an extra base hit, which, as miniscule as it is in Vidro’s case, would still skew the numbers slightly.
From the article
Making some pro-Vidro changes: assume he hits .300, that his non-GIDP outs advance the runner, say, a third of the time, put the power in… I came out at -.1 run/PA with a runner on.
Thom: Yup.
12.
ooop. Sorry, didn’t catch that.
Dave and Derek: Does it stand to reason that Ichiro’s 81 percent steal rate is at a heavy risk of going down because he’s being “forced†to attempt steals in situations in which the circumstances for a steal are less than ideal? (i.e., lefty with a good move on the hill, good defensive catcher, power pitcher with good slide-step, etc.)
I don’t think we’ve actually seen that. He’s run like crazy against panzy armed catchers like Josh Bard, but for most of April, he basically stood on first base like a spectator. He didn’t try to run on Pudge in the Detroit series, for instance.
I think Ichiro’s going to run when he thinks he has a good chance to make it, and he’s not going to run when he doesn’t.
I took the question to mean “Would it go down if he was forced to run all the time?”
And that’s the problem with the run-all-the-time strategy, as I pointed out: it’s only a good move, even with the GIDP-ness, if you can do it at least x% of the time. Against Pudge, say, it’s likely you wait for Vidro to GIDP.
For what it’s worth, in Vidro’s 40 PA with a runner on first and less than two outs, summing his WPA on FanGraphs shows he has +0.234 WPA and +0.97 BRAA. Small sample size, of course, but he has at least been a positive in DP situations to date.
Yea – three of his double plays have come in blowout scenarios where the game was already decided, and thus, they had almost no impact on the team’s win probability.
But Derek’s overall point remains – the team’s DH hits into so many double plays that he basically wipes out all the value his singles add. That’s sad.
Interesting analysis. I’m curious about the following, however:
[quote] Making some pro-Vidro changes: assume he hits .300, that his non-GIDP outs advance the runner, say, a third of the time, put the power in… I came out at -.1 run/PA with a runner on.[/quote]
More for my edification than to raise an issue, what do your numbers look like, specifically, when you “put the power back in” to this analysis? Are you looking specifically at how often he moves a runner from first to third vs. first to second vs. first to home? Also, are those numbers coming from this season specifically? If so, because of small sample size, etc. I worry that’s incredibly misleading. I’m not saying the analysis isn’t sound, but I’m curious to see what those numbers look like.
And also I have no idea how to quote on Word Press, apparently.
Guessed at runner advancement, tossed the almost unnoticeable XBH into the percentages.
Feel free to offer a better number, or analysis based on what you feel are better assumptions.
Hargrove’s quote from last night’s game suggests that he’s aware of this problem. Unfortunately, his solution isn’t to steal with Ichiro, it’s to hit and run with Vidro at the plate. Hence Ichiro’s CS.
Nor has he thought of the more obvious solution: bat someone else second.
Anybody like Johjima as a vast No. 2 improvement? I remember thinking that would be a fine idea last year when they were first trying Beltre there.
There aren’t any good options, really, but no, I don’t really want Joh hitting second. He’s potentially even slower than Vidro, if that’s possible, and also a pretty extreme groundball hitter.
Joh fits best in the #5 hole. Honestly, right now, the best option is probably Jose Guillen.
How did we manage to assemble an offense that skips straight from the number one spot to the number five spot in the order?
[ot]
I guess you could say roster construction is not one of Bavasi’s strengths.
What about batting Broussard 2nd against RHP?
This is the first serious crack I’ve taken at something like this, but those Vidro numbers look a little off to me and I also doubted the very small sample size of the season thus far. As such, here are Vidro’s numbers from 2006 for every possible situation in which he could ground in to a double play. For simplicity sake, I will assume zero outs in all situations.
Also, for the sake of conservatism, we’ll assume the runner(s) always stay in their exact position(s) relative to Vidro on a hit and that, unless otherwise specified, a Vidro out (non-dp) advances no runners (or Vidro replaces a runner on a fielder’s choice). Also, this is using his 2006 numbers, coming up with runners on 199 times and runners on and two out 55 times for 27.6%, meaning he has the opportunity to ground in to a double play on roughly 72.4% of his plate appearances in the following scenarios.
Vidro 2006
~~++–++~~
Runner on first:
PA: 110
H: 27
BB: 9
SO: 6
HBP: 1
GDP: 12
24.5% of the time he gets a hit, breakdown as follows:
16.3% of the time it is a single (.93 RE -> 1.58 runs = +.63 runs)
7.3% of the time it is a double (.93 RE -> 1.97 runs = +1.04 runs)
0.9% of the time it is a home run (.93 RE -> (2 + .54) runs = +1.61 runs)
9.1% of the time he walks or is HBP. (.93 RE -> 1.58 runs = +.63 runs)
5.5% of the time he strikes out. (.93 RE -> .56 runs = -.37 runs)
15.1% of the time he grounds in to a double play (.93 RE -> .11 runs = -.82 runs)
45.8% of the time he’s out sans other-runner-wipe-out. (.93 RE -> .56 runs = -.37 runs)
This totals to: -.06 runs per Vidro plate appearance with a runner on first.
~~++–++~~
~~XX–XX~~
Runners on first and second:
PA: 22
H: 2
BB: 1
SO: 0
HBP: 0
GDP: 3
9.1% of the time he gets a hit, breakdown as follows:
9.1% of the time it is a home run (1.58 RE -> (3 + .54) runs = 1.96 runs)
4.5% of the time he walks or is HBP (1.58 RE -> 2.37 runs = .79 runs)
0.0% of the time he strikes out.
18.8% of the time he grounds in to a double play* (1.58 RE -> .37 runs = -1.21 runs)
67.6% of the time he’s out sans other-runner-wipe-out** (1.58 RE -> .96 runs = -.62 runs)
*Let’s assume he moves the runner over to third in this case.
This totals to: -.43 runs per Vidro plate appearance with runners on first and second.
~~XX–XX~~
~~÷÷–÷÷~~
Runners on first and third:
PA: 14
H: 4
BB: 2
SO: 1
HBP: 1
GDP: 0
SF: 2
28.6% of the time he gets a hit, breakdown as follows***
14.3% of the time it is a single (1.81 RE -> (1 + 1.58) runs = .77 runs)
14.3% of the time it is a double (1.81 RE -> (1 + 1.97) runs = 1.16 runs)
21.4% of the time he walks or is HBP (1.81 RE -> 2.37 runs = .56 runs)
7.1% of the time he strikes out (1.81 RE -> 1.17 runs = -.64 runs)
0.0% of the time he grounds in to a double play.
20% of the time he hits a sac fly**** (1.81 RE -> (1 + .57) runs = -.24 runs)
22.9% of the time he’s out sans other-runner-wipe-out***** (1.81 RE -> 1.17 runs = -.64 runs)
***Assume a hit scores the runner at third.
****Assume only the runner at third advances.
*****Assume no runners advance.
This totals to: +.16 runs per Vidro plate appearance with runners on first and third.
~~÷÷–÷÷~~
~~##–##~~
Bases Loaded
PA: 11
H: 3
BB: 1
SO: 1
HBP: 0
GDP: 1
SF: 1
27.3% of the time he gets a hit, breakdown as follows:
27.3% of the time it is a single (2.37 RE -> (1 + 2.37) runs = 1 run)
9.1% of the time he walks or is HBP (2.37 RE -> (1 + 2.37) runs = 1 run)
9.1% of the time he strikes out (2.37 RE -> 1.65 runs = -.72 runs)
12.6% of the time he grounds in to a double play****** (2.37 RE -> .59 runs = -1.78 runs)
12.6% of the time he hits a sac fly******* (2.37 RE -> (1 + .96) runs = -.41 runs)
29.3% of the time he’s out sans other-runner-wipe-out******** (2.37 RE -> 1.65 runs = -.72 runs)
******In an effort to balance out the types of double play that can happen and their results in this situation (with zero vs. one outs, run scoring or not, etc. etc.) let’s assume it’s a home to first dp so the run doesn’t score but we’re left with runners on second and third.
*******Assume only the runner at third advances.
********Assume no runners advance.
This totals to +.08 runs per Vidro plate appearance with the bases loaded.
~~##–##~~
Putting it all together, in all possible GDP situations:
110pa @ -.06 runs/pa
22pa @ -.43 r/pa
14pa @ +.16 r/pa
11pa @ +.08 r/pa
Totals to:
~ -.08 runs per plate appearance. I actually imagine this would be a little low, based on our assumptions of the conservatism of the runners. The biggest discrepancy would probably be in the outcomes of Vidro’s singles and doubles with a runner on first. We can partially account for this by distributing his RBI for those situations (minus the home runs and sac flys) among the above averages. With a runner on first, Vidro accrued 5 RBI with one home run and zero sacs. We could divide those 3 runs among his 110pa and we get an additional .03 runs/pa. Of course, this would also change the resultant run states of those hits (which were used in the above calculations), but the point is our resultant -.08 runs is probably too low.
So, in summary, while Derek’s analysis is essentially right, I think the numbers are slightly misleading for this small a sample size at this point in the season.
Now, I’m going to watch the game; thank God for TiVo.
How’d I do, by the way?
Well, here’s the big problem with your analysis of 06: you’re looking at, say, “man on 1st” and calculating the chance that he GIDPs as % of GIDP from GIDP/PA.
But with that, as with the other situations, Vidro can’t GIDP if there are two outs. So you’re saying “Vidro is +x with a guy on first” but that doesn’t really apply to “if there are 1 or 0 outs, what’s the chance he GIDPs, and how does that affect your strategy?”
This is why I attempted to go back and extrapolate Vidro’s GIDP % to look at those situations.
The other big problem with Vidro, is that he does not seem to understand what to do when Ichiro is on the move. It is so annoying watching him foul the ball when Ichiro is attempting a steal, and its even worse when he fails to put the ball in play on a hit and run. (No wonder Ichiro is pissed) So yes, the numbers might indicate it is better for Ichiro to try stealing every time hes on first, but even then Vidro will make Ichiro’s life harder.
Looking on the bright side, the Ms DH is so bad that when the Ms play a NL team away they will not suffer much of a down grade in their lineup.
32:
I realized that problem; what I actually calculated was GIDP/(PA x .724). Vidro was in a situation with fewer than two outs for 72.4% of his plate appearances, so I extrapolated that 72.4% of the plate appearances where he could ground in to a double play (the scenarios I laid out) to arrive at the percent chance he would GIDP in the scenarios I go over. This was similar, I thought, to what you went over when you mentioned a similar adjustment. Am I wrong?
Considering that your average MLB player hits about .270 with maybe a 33 or 34% chance to not make an out, it seems intuitive that over a large sample size, the average player that comes to bat with a man on 1st (no outs) would have an average run expected score contribution below zero (like Vidro’s -.14), unless the hitter made up for it by OPS’ing very high. I’m not trying to defend Vidro, but theres plenty of legitimate things to criticize him over without having to reach for something you could say about (my guess) 80% of major leaguers. Basically what all those numbers boil down to is something we already knew- Jose Vidro hits for mediocre OPS and isn’t fast. Being a (merely decent) OBP hitter who is slow, what little value he has comes with the bases empty.
The analogy with the pitcher’s bunt (-.19) was interesting and thought provoking, but flawed because we’re comparing a player to a “productive out,” and it also assumes that a pitcher can get the bunt down 100% of the time. It does reveal however, how NL ballclubs are still able to score a decent amount of runs every year even with a .100 hitter in the lineup every day.
Iiiiiii don’t understand why that’s flawed. It’s not intended to be a serious argument, it’s meant to point out that if you had a totally helpless bat-swinger, they’d be only a little less productive than Vidro if they could bunt consistently.
And as to your other argument – no, you can’t say that about 80% of major leaguers.
Ah. So here’s the thing — what’s the run expectancy after a hitter GIDP with one out?
The average AL hitter’s line is .275/.339/.436
Vidro’s line is .303/.354/.362
The average AL hitter’s about a 1.2 G/F rate, Vidro’s historically been way over that and this season is at 1.49
The average hitter gets fewer hits, but there’s a lot more power, and the outs are less likely to be GIDPs. They also walks more often than Vidro.
Derek re: 37
I don’t really want to go through all the calculations again for GIDP with one out, but I was trying to demonstrate that, despite his lack of speed, I don’t think he’s quite the liability on the basepaths that your analysis painted him to be. Keep in mind that I tried to remain very conservative with my numbers and I still came out with significantly better run expectancies per plate appearance than you did, mainly because of this season’s small(er) sample size compared to 06.
I’m not saying Vidro should be hitting in the 2 hole, or even that I disagree with most of the points of your argument, I just think it’s overly pessimistic in terms of his GIDP rate and performance in general.
If you remove absolutely every other facet of my analysis, you still get his GIDP at about 14% of the time he had the opportunity. While still not ideal, this is significantly lower than the 20% he’s posting so far this season (and I bet that number has already dropped a little after last night’s game). I just don’t think that 20% is sustainable and I also don’t think GIDP is THAT much more of a problem for him than a plethora of other players out there.
39 –
The problem I see is that his 20% GIDP for ’07 so far is actually lower than it really should be. He’s been pretty lucky with a lot of his singles getting through (the anti-Sexson), and Ichiro has been stealing or being sent on hit-and-runs a lot more than usual as of late. So assuming 20% is unsustainable doesn’t necessarily mean it’s not going up instead of down. 😉
Yeah, exactly.
I wouldn’t deny that his 2007 GIDP rate may be too high, but I think there’s enough evidence that it’s if anything a little low. If 2006 stats make it seem like he should be a little better, well, okay. But the 2007 stats show that he’s not so far. You can say “small sample size” if you wish… or we can acknowledge that it is what it is, and wait to see if more information this year helps.
If nothing else, we should be able to say that so far this year, he’s certainly done nothing to show that the team’s optimism about his speed was warranted.
39 and 41:
I completely agree about this, but based on his past, my guess is that he won’t continue to be so bad, but that’s just a guess. I’d be willing to posit that he ends the season at better than 17% GIDP/opportunity.
Wait… I was 39.
What I meant was 40 and 41.
DMZ,
Saying that something is flawed is not the same thing as saying you have no point at all. You have to factor that bunting does not have a 100% success rate. It doesn’t make a huge difference in your argument, but it is a flaw.
Also, I disagree with the idea that having a negative run contribution with runners on is anything remarkable (your emphasis on it leads me to the conclusion that you are inferring as much, though its just my interpretation). The 80% number I was guessing was not the number of batters who are the same level or worse than Vidro, but the % of batters that would post a negative run contribution score with a man on and less than 2 outs. 80% is a total guess, it might be 70% or 60% or 50.1%, but I’d be shocked if a majority of MLB players posted positive average run contribution in that scenario. How many hitters do the Mariners have currently that would post a positive run contribution average in that scenario? My guesses: Guillen, Johjima, perhaps Ichiro. (excluding back-ups like Broussard, Ellison, and Burke)