Despite a career MLB line of .306/.339/.480, and despite having some very good overall years, Robinson Cano can really frustrate Yankee fans. In addition to what seems likes maddening inconsistency in general, Cano has hit worse with runners on base in almost every season of his career so far, with 2007 being the lone exception.
Here's how Cano's splits in this category have looked so far in his career.
Babip: Batting average on balls in play
wOBA: Weighted on-base average
FWIW, the AL has typically hit better with men on base than without, at least looking over the last few seasons. The difference isn't huge, but it's generally been in the 5% range.
The difference in wOBA between Cano with the bases empty (---) vs. Cano with men on base is .068. Even if you were to completely ignore the fact that weights of wOBA are different with runners on base vs. the bases empty (in other words, positive offensive events are worth more when runners on base and less when the bases are empty), the difference between those two wOBAs over around 1500 PAs is close to 90 runs, roughly 40 runs over a 650 PA season.
When looking at Cano's actual value to the team, this is a real and persistent problem that has made him less valuable then what a context-neutral metric would have said. The question I want to look at here is if there is some reason to think there is more going on here than the vagaries of batted balls and sample size when breaking down a player's performance into subsets that fit into neat little buckets.
The sample size thing is important here. While it may feel like 1500 PAs in both splits is significant, it's still not quite enough to start thinking we're seeing definitive proof. We generally need at least 2000 PAs in a split, but even then we have to regress them somewhat, depending on the split and the player and factoring in the fact that by the time a player has accrued those 2000 PAs, he may be a different player than he was when he accrued the first n of them. So keep that in mind when looking at the numbers that follow.
Generally, we think of luck in the batter's box in terms of BABIP (batting average on balls in play). While it's more nuanced than that, we can see that Cano has a BABIP of .346 with the bases empty and a BABIP of .294 with men on base. However, if we use Pitch F/X data to try and break down his performance by batted ball type, it would look like this.
Pitch F/X data is only from 2007 on, and isn't complete for those years, but it's pretty close. Unfortunately, Pitch F/X does not break out hits by batted ball type. However, we can try to extrapolate the total batted balls by dividing the outs for each type by the average percentage of outs when each type is hit, which are:
Fly balls are outs 79% of the time.
Ground balls are outs 72% of the time.
Line drives are outs 26% of the time.
Pop ups are outs 99% of the time unless Luis Castillo is under it.
Using those figures, we'd get a revised batted ball distribution like this.
Extrapolating batted ball types in this way introduces some uncertainty into this, although I suppose you could say that there's also uncertainty in the classifications of batted ball types on the margins. Anyway, keep in mind the fact that although this data is presented empirically, there's some fuzziness in here.
You can probably already tell this by looking at the numbers, but that type of batted ball distribution is pretty similar in both cases, and does not support a BABIP difference of .052.
Delving a little further into Pitch F/X, we can look at what Cano does in the batter's box depending on whether there are men on base or not to see if his actual approach is changing.
Update: Charts below have been updated to include missing columns.
|Split||Pitch||#||max||min||avg||ball %||stkS%||foul%||stkC%||In play, out(s)%||In play, no out %||HBP %|
|Men On||All Fastballs||1382||99.4||78.7||91.6||34.2%||3.8%||24.2%||16.4%||14.3%||3.5%||0.3%|
|Men On||Cut fastball||61||92.5||78.0||87.2||23.0%||6.6%||16.4%||16.4%||32.8%||3.3%||0.0%|
|Men On||Split-finger fastball||3||87.6||84.0||85.3||66.7%||0.0%||0.0%||0.0%||33.3%||0.0%||0.0%|
#: number of times pitch was thrown as recorded in Pitch F/X
max: highest recorded starting velocity
min: lowest recorded starting velocity
avg: average recorded starting velocity
ball %: percentage of time pitch was taken for a ball
stkS%: percentage of time pitch was swung on and missed
foul%: percentage of time pitch was fouled off
stkC%: percentage of time pitch was taken for a called strike
Here's the pie chart version of the last two rows.
In general, it looks like he's a little more likely to swing at pitches with runners on. He takes pitches 48.9% of the time with no one on base, and 43.6% of the time when there are runners on, although that could be due to the fact that he's more likely to see a strike when there's a runner on base. But I don't know if a difference of 5% here is necessarily all that meaningful.(Note: After revising the data to include the missing outcomes, this is no longer true. Cano takes a pitch 49.1% of the time when there are either runners on or not)
Honestly, I expected to see more of a split here in the underlying data, but it's just not there.
Cano's results to this point with runners on base are markedly worse than his results with the bases empty, but it's not because of any obvious change in his approach in the two scenarios, unless I'm missing something here or not considering something that I should be.
I guess this is encouraging, because it means we really shouldn't have any reason to think that Cano will continue to hit as poorly with men on base as he has so far.
Another update: As suggested by sam, here's a look at the pitch locations against Cano with men on vs. with the bases empty. I don't think it shows much if anything, maybe that he gets more pitches outside with runners on, but here it is anyway.