(Editor’s Note: Apologies for the delay on this, as we try to run this feature every Thursday. My fault – not Jon’s. Thankfully, given last night’s win, this piece is timed impeccably. Enjoy!)
Off-seasons are notorious for banal story lines, and while the Indians made their share of big-time player acquisitions this off-season (Mike Redmond, anyone?), most of the talk this winter has dealt with Manny Acta and how he might leave his stamp on the 2010 ball club. It seems the unanimous chorus from the Cleveland faithful was something along the lines of: “Hey, he’s not Eric Wedge! YAY!”
But now that we’re playing baseball games that actually count, I thought we might get a bit more specific than his non-Wedgian status and take a look at some of Acta’s thoughts on in-game managing decisions. Let’s start with his thoughts on the sacrifice bunt. Here’sActa:
“Bunting is pretty outdated. Everybody scores so many runs nowadays, it doesn’t make sense to play for one run unless it’s late in the game and it’s close. I hardly ever bunt early in a game, unless it’s with a pitcher. A big inning can win you a game. One run in the third inning can’t, unless you have Pedro [Martinez] pitching.”
So let’s look to see how bunting works in real-game situations and try to draw some conclusions as to whether Manny’s onto something here.
Obviously, the point of a sacrifice bunt is to give up one out to advance a runner one base—typically from first to second, but occasionally from second to third. To weigh the value of giving up that out for the advance of the runner, let’s look at a table that explores how many runs ON AVERAGE a team scores until the end of an inning in all of the various base-out states:
We’ve looked at charts like this before, but let me refresh your memory. In the first column, you’ve got all the possible combinations of baserunners; zeros represent unoccupied bases while an x indicates a baserunner. So looking at the first entry, when there are no runners on and nobody out, the team scores about 0.517 runs (multiply that by nine, and you get 4.7, or the average number of runs scored per team, per nine-inning game).
Remember that a “successful” sac bunt would be moving a runner up one base and costing your offense one out. So take a look at a few bunting scenarios. The first would be with a runner on first and nobody out. The starting run expectancy in that scenario is 0.897 runs. Now, imagine that you complete a successful sac bunt. You move over to the “1 out” column and down to the “0x0” row (runner on second). What’s our new run expectancy? 0.691. So a “successful” sac bunt costs our team .206 runs to the end of the inning.
Manny could be onto something here!
We were likely to score more runs before the bunt. Why? Well, because in baseball, outs are precious commodities. The game doesn’t end after a certain amount of time, but after 27 outs. So you don’t want to go giving them away unless the payout is significant.
I’ll let you look at the other scenarios for yourself, but here are the run expectancy changes of a sac bunt, by the various base-out states:
0 outs, runner on first: -0.206 runs
1 out, runner on first: -0.199 runs
0 outs, runner on second: -0.159 runs
1 out, runner on second: -0.319 runs
0 outs, runners on first and second: -0.055 runs
1 out, runners on first and second: -0.307 runs
As you can see, a sac bunt never increases your run expectancy, at least not in an average run environment (where both teams are likely to score about 5 runs per game). And this is assuming that all sac bunts are successful, which is obviously not the case. So it looks like Acta is, indeed, correct when it comes to his bunting philosophy.
But, as you might already be shouting, run expectancy isn’t always the best way to measure things. After all, you may be less likely to score multiple runs in an inning by giving up an out, but still be more likely to score a single run. And sometimes one run makes all the difference.
To check out this scenario, let’s look at a different chart. This one presents “win expectancy”—i.e. how often the batting team won the game—for games where the team on offense is down by exactly one run in the seventh inning or later (including extra innings; quoted from The Book):
So again, with no one on and no one out, teams won approximately 35.3% of the time in these one-run, late-game situations. With the bases loaded and nobody out, they won 68.7% of the time.
Let’s look again at our bunting scenarios. It looks like bunting a runner from first to second with no outs means that you’re less likely to win the game (.431 winning percentage vs. .399 winning percentage). There are certainly some closer calls in this chart, but overall, it again looks like Acta’s thoughts seem to coincide with what the numbers indicate: bunting is more likely to cost your team a chance to win than to increase it. Or, according to Manny himself:
“If I have enough data…I can go by the stats, because they don’t lie. I mean, it’s been proven to me that a guy from first base with no outs has a better chance to score than a guy from second base with one out. That’s been proven to me with millions of at-bats. So I don’t like moving guys over from first to second unless there’s a pitcher up or it’s real late in the game. … I’m telling you right now you’re not going to be seeing me bunting guys from first to second in the middle of the games. I’ll be straight up to you guys, I’m not going to be running all over the place just so 25,000 people in the stands are saying that I’m aggressive.”
Hmm. Seems smart and well-reasoned. So far, so good.
Now let’s go back to the first chart to take a closer look at what the numbers have to tell us about stealing bases.
Obviously, a successful stolen base will increase your run expectancy. After all, it’s always better to have a runner on second than first, so long as it doesn’t cost you an out. You can check it out for yourself, but I would guess this will come as no great surprise.
However, getting caught stealing a base is really costly—much more than a sac bunt. For example, with a runner on first and one out, you’re likely to score about 0.549 runs to the end of the inning, but if that runner gets caught stealing, you’re run expectancy goes down to 0.108 (2 outs, nobody on). So getting caught in that scenario costs you .441 runs. When you think of it that way, stealing starts to look pretty costly.
So next I want to compile how many runs you gain by stealing successfully versus how many runs you lose by getting thrown out. The chart below is just a permutation of the one above; if you need the exact calculations, just ask in the comments, but you should be able to figure this one out:
Starting # of outs
The way to read this is, “If you successfully steal second with nobody out, you add 0.24 runs to your inning, but if you’re thrown out you cost your team 0.62 runs.” Look at stealing third with two outs: “If you’re successful, you add 0.02 runs to your inning, but if you’re caught, you cost yourself 0.69 runs.” That makes sense to those of us who were told by our coaches, “Never make the third out at third base.” With two outs, you’re just not that much more likely to score from third than you are from second, so it doesn’t make much sense to try to steal in this situation.
But since I promised you some extra math this week, let’s add an extra row to this chart. What I’d like to determine is how often a team must be successful for a stolen base attempt to makes sense? At first glance, it looks as if a caught stealing is significantly more costly than a successful steal is valuable (positive .24 for a success; negative .62 for getting thrown out). For example, when stealing second with no outs, if you were successful twice in a game but thrown out once you would have the following run expectancy equation:
0.242 + 0.242 – 0.617 = -0.134 runs
So even if you’re successful 67% of the time (2 for 3) you’re still costing yourself runs over the course of a game! So how should we figure out how successful a team needs to be to break even? Why, algebra of course! (I warned you about the math; skip the next two paragraphs if you’re ascared.) If we’d like to know the breakeven point, let’s set the right side of this equation to zero and calculate how many successful steal attempts you need out of 100 to balance the scale. That equation would look like this:
A*(0.24) + (100-A)*(-.62) = 0
.24A – 62 + .62A = 0
.88A = 62
A = 71.9
(For those following at home, A represents the number of times out of 100 you’re successful and 100-A is the number of times you’re caught stealing. Obviously, they add to 100 in our scenario.)
Solve for A, and we get…71.9! So when there is a runner on first base and nobody out, you should only steal if you expect to be successful 72% of the time. Less than that, and you’re costing your team runs. I’ll do the heavy lifting and solve the rest of these “break-even” points for the various scenarios.
Exactly what we expected! Stealing third with two outs is stupid, unless it’s a sure thing. Otherwise, we want to be successful between 67% and 76% of the time, depending on the situation.
“We will run selectively. I think one of the things that doomed this club [Washington] last year is that they were first in caught stealing. I am not going to be running all over the place just because 25,000 people in the stands are saying I am aggressive while people are getting thrown out on the bases. Not everybody will have a green light here. The guys who are going to run are the guys who are going to prove to me that they will be successful most of the time trying to steal a base.”
He’s really worried about those 25,000 people, huh? And, according the author of the article, “For Acta, a 70% success rate marks the line between helpful and damaging.”
You think he’s seen these numbers before? I’d bet yes.
Before we go, let’s take a lookat how the Indians did stealing bases last season, to see whether they were successful enough to merit their attempts. (I’ve included every baserunner who attempted a steal last season.)
Any surprises here? Grady got thrown out far too often, and ended up costing us more runs than he created on the bases, though perhaps that’s a result of his abdomen injury. Same with Brantley, though his is obviously a small sample—he was successful over 90% of his time in Columbus in 2009. Choo and Cabrera are the obvious stand-outs, and I would expect Acta to be giving the green light to those two going forward, as well as Brantley, considering his minor league success.
Just for reference, here’s how the Indians performed last season relative to their American League foes:
I’d say they were right around average with regard to their success rate, but with far fewer attempts than the average AL team. Perhaps this was a result of too many ABs going to Dellucci, Garko, and DeRosa—three notoriously slow runners? This year, with Brantley and a healthy Sizemore? I’d guess more attempts and a slightly better success rate, but only time will tell.
We’ll leave you with one more Acta quote, since some readers have wondered whether he’s a “numbers” guy or an “old school” guy:
“I want to win. More than being statistically-inclined, I’m very open-minded. If someone can show me things that I didn’t already know, I am willing to change. I’m not stubborn. If the statistical evidence shows I’m wrong, and it helps my team win baseball games, then I would be a fool not to listen.”
Refreshing, isn’t it? It’s nice to know that our manager takes his job seriously. He sounds like he’ll keep an open mind. When the numbers are there, he’ll use them. When they’re not, he has 20 years of real-world, baseball experience to rely on. He doesn’t seem to accept the false choice between statistical analyses on the one hand, and experiential knowledge that only “old school” baseball men have on the other. This seems like a good sign to me. And if all else fails, at least there won’t be any more grinding.
Next time we’ll take a look at Justin Masterson and his pitch types, since by then we’ll hopefully have some pitch/fx data to play with. See you then!