Moments with Maddon: Recap

Due to my procrastination, we have two weeks worth of "Moments with Maddon" to discuss, so let's get right down to it.  Last week, I asked a question about how, as a manager, you balance statistic and scouting information.  I don't claim to have a specific answer about the "proper" balance between statistics and scouting; I was interested in stirring up some debate and seeing the different opinions people have.  Here's the specific situation I proposed:

It's game-day and you need to determine your lineup.  In particular, you're having trouble with the 2B/RF slots. You're debating between Ben Zobrist, Sean Rodriguez, Gabe Kapler, and Matt Joyce for the two openings, but can't decide which to use.  There is a righty on the mound for the opposing team, which makes you lean towards using Joyce in right field and Zobrist at second in order to get the platoon advantage.  However, your scouts have informed you that in the limited at-bats that Zobrist and Joyce have faced this pitcher, they have looked utterly and completely fooled. Zobrist is 2-19 against him with 7 strikeouts, Joyce is 1-9 with 4 strikeouts, and the scouts insist they both look even worse than those numbers.  They can't hit this guy.

Conversely, Kapler and Sean Rodriguez both have exemplary scouting reports against this pitcher.  Their "swing planes" line up with the pitcher's stuff and they've looked locked and loaded against him in the past, although in only five plate appearances each. 

For argument's sake, let's say each player has been performing to that point of the season at the level of their CHONE predictions.  No player is on a particular hot or cold streak.  Who would you start?

Going into this, my theory was that most people would trust that statistics more than scouting when the disparity in talent between the starter and replacement was large (Zobrist versus Kapler), trusting that the more talented player would still perform better than their potential replacement despite a negative scouting report.  When the talent level of players was more equal (like between Joyce and Rodriguez), people would be more likely to side with the scouting report.  In other words, since the statistics suggest one player would be only a minimal improvement over the other, it'd make sense to trust your scouting report to try and give you an extra edge.

Lo and behold, that's exactly what happened.  The leading response (52%) was "Zobrist and Rodriguez", which is what my theory predicted.  I'm not going to claim to know if this is the "optimal" scenario because I don't know that for sure, but judging from our responses it seems to be one that makes a large amount of sense at least.  Managers are constantly faced with similar decisions of how to blend the stats and scouts together in the way that makes the most sense, so hopefully this got you thinking a bit about how you'd personally go about blending the two.

And now, on to our scenario from two weeks ago.  Here's a refresher:

Bottom of the seventh inning, no outs.  Gabe Kapler is on first, Dioner Navarro is at the plate, and Jason Bartlett on deck.  The Rays are facing the Yankees and the score is currently tied at four.  A.J. Burnett is on the mound, but the Yanks have right-handers Phil Hughes and Alfredo Aceves warming in the pen.  The Rays' bench is still full and consists of Kelly Shoppach, Willy Aybar, Sean Rodriguez, and Matt Joyce.

What's the most optimal strategy in this situation?  Well, let's find out, looking specifically at the two choices that got the most support: pinch-hitting (39%) and sacrificing (33%).

Pinch-Hit

I talked about pinch-hitting some in our wrap-up on the first Moment with Maddon, and the main take-away point was that you should only pinch-hit in situations where the pinch-hitter is a significantly better hitter than the one at the plate.  The reason for this is because hitters see their performances suffer an average of 10% when pinch-hitting, which turns a .330 Weighted On-Base Average (wOBA) hitter into a .300 wOBA batter and a .380 wOBA into a .345 wOBA (which is around league-average).  So please, don't ever suggest that we pinch-hit for Evan Longoria.

In this situation, though, Navarro is at the plate and he's one of the weakest batters on the Rays (career .297 wOBA).  Also, as some astute readers pointed out, he hits a lot of groundballs (career 41.1%) and is a slow runner, which leads to a lot of double-plays.  You'd be risking leaving yourself without a backup catcher in case the game went into extra-innings, but pinch-hitting for Navarro seems to make sense and is a defendable position.

Since the end goal is to increase the Rays' chances of winning, exactly how much would be increasing our chances of winning if we were to pinch-hit for him?  Let's run some numbers using Expected Outcome Analysis, multiplying the percent chance that an event will happen by its win expectancy (WE).  That's probably confusing, so here's the formula I'm using for this situation:

Expected Change in WE = (% chance of hit, walk, or hit-by-pitch * change in WE to 1^st and 2^nd, no outs) + (% chance of out * change in WE to runner on 1^st, one out) + (% chance of double play * change in WE to none on, two outs)

Sure, it's simplistic, but it gets the job done.  When we calculate out the difference between Willy Aybar and Navarro using this formula (and including a 10% hit in production for Aybar), you end up getting .001 change in WE for Aybar and a -.008 change in WE with Navarro.  To get some perspective, if you had a batter like Pena (approx. .380 OBP) at the plate, you'd get a .015 change in WE.  So it's a small difference, but you would definitely be increasing your odds of winning by pinch-hitting.

Sacrifice Bunt

As a believer in sabermetrics, I'm automatically pre-disposed to dislike the sacrifice bunt.  One of the basic tenants of "Moneyball" was that outs are precious as gold and you shouldn't willingly give them up, which is a valid sentiment considering each team only gets 27 outs a game.  However, it's too simplistic to say that teams should never sacrifice; there are certain instances where playing for one run makes strategic sense.  What are these situations where it makes sense to sacrifice?  That's actually a really tough question to answer succinctly and if you want an exact answer, you'll need to read Chapter 9 in "The Book: Playing the Percentages in Baseball".  In general, you don't want to sacrifice early in a game and should only consider it late in a close game.  It depends, though, on the quality of the batter, the count, the speed of the batter, the inning, the score, the amount of runners on base...it's rather mind-blowing how many variables there are.  I wish I could say it was simple, but it's actually quite complex.

That said, here's a summation of the principles to keep in mind, courtesy of "The Book":

"With a runner on first or first and second, and no outs, late in the game, with the infield playing up, all but the best hitters can bunt. Early in the game, if the infield is expecting a bunt, only the weakest hitters should bunt. The speed and bunting proficiency of the batter should be considered. Slow/poor bunters should rarely bunt and fast/good bunters can bunt more often." (pg. 283)

So where does this leave us for our game situation?  Navarro is a weak hitter, a slow runner, and (I believe) a poor bunter.  Actually, I don't really know how good he is at bunting, but my gut tells me he's not that good at it; feel free to correct me.  If he is a weak bunter, though, this really makes it a tough call.  Theoretically this situation would be ideal for a sacrifice bunt, but you also don't want Navarro bunting into a double play or giving up an out too easily. 

How likely do you think it is Navarro will get a successful bunt down?  Forty percent of the time?  Sixty percent?  Twenty Percent?  Personally, I wouldn't have him attempt a sacrifice bunt unless I was sure he had higher chance of putting down the bunt than he did of getting a hit...plus some.  To pick a number, I don't think I'd bunt if I thought he had less than a 50% chance of success.  If he fails on the bunt, your win expectancy drops by 4%; if he's successful, it remains stable.  In essence, you're trading an out for certainty - certainty that you won't hit into a double-play and that your win expectancy won't drop.  If you're not confident about that bunt's success, though, don't even think about it.  You should just pinch-hit.

Conclusion

In the end, I'd have to pinch-hit for Navi.  I don't like his odds of getting down a successful sacrifice bunt and like bobr and others pointed out, it seems like a poor idea to be playing for one run against the Yankees.  If it's the ninth inning, sure, why not?  But in the seventh inning you're leaving them too much time to come back.  Sacrificing does have its time and its place, but this doesn't seem like quite the right spot.

Again, though, these are tough decisions and legitimate arguments can be made in either direction.  Sometimes when you're a manager there's no clear-cut "right" answer, so you have to hang your hat on a decision and hope it turns out for the best.  If it pans out, you'll be hailed a genius the next morning in the papers; if you're wrong, you'll be skinned alive.  C'est la vie.