What does XBT% really tell us?
Posted by Andy on April 27, 2011
I was just perusing the 2011 MLB baserunning splits and did a sort by XBT%, which is the percentage of time that a baserunner advances more than 1 base on a single or more than 2 bases on a double.
The league average so far this year is 41% but check out the teams with the lowest percentage so far:
| Tm | R/G | XBT% ▴ | 1stS | 1stS2 | 1stS3 | 1stD | 1stD3 | 1stDH | 2ndS | 2ndS3 | 2ndSH |
|---|---|---|---|---|---|---|---|---|---|---|---|
| NYY | 5.74 | 24% | 23 | 17 | 6 | 15 | 14 | 1 | 7 | 3 | 4 |
| DET | 4.55 | 29% | 38 | 34 | 4 | 20 | 13 | 6 | 26 | 11 | 14 |
| TOR | 4.32 | 31% | 33 | 25 | 7 | 14 | 9 | 5 | 21 | 11 | 9 |
| MIN | 3.38 | 33% | 42 | 30 | 12 | 14 | 7 | 7 | 29 | 17 | 9 |
| CHC | 4.18 | 33% | 48 | 37 | 10 | 15 | 6 | 8 | 18 | 9 | 9 |
| OAK | 3.48 | 34% | 36 | 29 | 6 | 10 | 7 | 3 | 27 | 9 | 16 |
| FLA | 4.33 | 34% | 45 | 29 | 16 | 12 | 10 | 2 | 33 | 19 | 13 |
| SDP | 2.96 | 36% | 35 | 23 | 12 | 9 | 7 | 2 | 18 | 8 | 8 |
| PHI | 4.32 | 37% | 45 | 37 | 8 | 11 | 6 | 5 | 33 | 11 | 20 |
| HOU | 4.41 | 37% | 48 | 39 | 9 | 13 | 8 | 4 | 37 | 13 | 23 |
The Yankees have by far the lowest percentage. It's an unusually low number as in most seasons the best and worst teams don't usually vary from the average by more than 6 or 7 percentage points.
Initially, you might think that the Yankees' low percentage indicates that they're having trouble scoring runs, but they're not. (Check out the first column above.)
In truth, what's happening is that the Yankees are homering a lot when the guys are on base, and as a result they don't actually have all that many baserunning situations. Note how many fewer situations they have in the columns to the right above, and it turns out that with such a small sample size, they just happen to have had some bad luck so far.
How much does it normally correlate with run scoring in normal cases? Not a ton. I took a look at the 2010 numbers, and there's quite a weak correlation between XBT% and R/G. For those who care, I found just a 0.10 R-squared on a linear correlation.
April 27th, 2011 at 9:13 am
You can't conclusively attribute it to luck. You can only say that we can't conclude yet that they are slow on the bases and that it may be attributable to chance. The difference is of such a magnitude that luck is probably not the best explanation, even with a limited sample size.
April 27th, 2011 at 9:25 am
This makes sense to me. With all the bangers the Yanks have in their lineup you don't want to be thrown out on the basepaths, I think. The odd thing to me is why teams that have trouble scoring (OAK, HOU, etc.) aren't more daring on the bases.
April 27th, 2011 at 9:49 am
I'd bet it correlates with stolen bases to a better degree than runs scored. Teams that don't have big boppers in the lineup are going to take more chances going for extra bases with speed in an attempt to move runners along.
April 27th, 2011 at 10:16 am
Good thoughts in #2 and #3.
I checked the correlation with SB plus caught stealing, and the R-squared is 0.15 for the 2010 data, so there's at least a higher correlation there than with overall runs scored.
There's even a higher correlation with Outs On Basepaths, which is times a runner was thrown out trying to advance on a fly ball or take an extra base on a hit. That R-squared is 0.23.
So I'd say there's merit to the idea that it's aggressive style of play that contributes to all 3 things--a higher percentage of possible extra bases taken, but also more outs on the basepaths, and more attempted steals. Teams that play less aggressively make fewer outs on the basepaths but also take a lower overall percentage of possible extra bases.
I also looked up the correlation between XBT% and HR, and for 2010 there's almost no correlation with an R-squared of 0.0023. I would have expected somewhat of a negative correlation due to the points above--why be thrown out running if you can just sit tight and score on a homer. But the Toronto Blue Jays had both a ton of homers (#1 in MLB last year) and one of the highest XBT% percentages (in a 3-way tie for second highest). If you take them out of the mix, the other 29 teams do show a slight negative correlation with an R-squared of 0.03.
April 27th, 2011 at 10:54 am
I think that this is an instance of correlation vs. causation.
Teams that don't generate a ton of offense are going to be a lot more aggressive on the basepaths, to squeak out more and more runs.
The question isn't, does team X's approach of taking the extra base score them more runs than team Y, but rather, does team X's approach of taking the extra base more often score them more runs than a hypothetical team X that doesn't take the extra base as much (still bearing in mind that there still exists a chance of failure).
April 27th, 2011 at 11:26 am
Blame Jeter 🙂
Yankees 11/45 - 24%
Chavez 2/2 - 100%
Martin 1/1 - 100%
Swisher 2/5 - 40%
ARod 3/10 - 30%
Granderson 1/4 - 25%
Teixeira 1/7 - 14%
Jeter 1/10 - 10%
Cano 0/1
Gardner 0/2
As a team, the runner at first has been stopping at third on doubles at a very high rate 93%. They lead the league in doubles with a runner on first are worst in the league in scoring that runner on that play.
Its very early. Could be lots of things going on in these small samples. Jeter's number will regress back up quite a bit. His XBT% has been in the 30s the past few years (it was in the 40s and 50s when he was in his 20s).
April 27th, 2011 at 11:27 am
@3
The fact that Toronto is on this list is surprising because they are near the top of the league in stolen bases.
John Farrell had promised a more aggressive team on the bases this year but it hasn't really materialized.
I wonder how much a manager's "personality" impacts a baserunning list like this?
April 27th, 2011 at 11:33 am
Note that XBT% does not do a great job of measuring the gain or loss of a more aggressive base running strategy. There are 2 ways to fail: only advance the minimum of bases on the hit, or to be thrown out trying to gain the extra base. XBT% treats these failures equally, while getting thrown out is obviously much much worse than not advancing. To better rank each team: use RE24. (Ideally adjusted for ballpark, pitcher(s), and succeeding batters.) I think that WAR uses something like that to figure Rbas.
April 27th, 2011 at 11:54 am
Out of curiosity, I checked stats on the Cardinals of 1982-87:
1982
SB -- 1st with 200, +51 vs. league avg.
CS -- 1st with 91, +22 vs. league avg.
SB% -- 6th at 69%, +1% vs. league avg.
XBT% -- 5th (tie) at 46%, +1% vs. league avg.
R/G -- 5th at 4.23, +0.14 vs. league avg.
HR -- 12th with 67, -41 vs. league avg.
1983
SB -- 1st with 207, +58 vs. league avg.
CS -- 2nd with 89, +16 vs. league avg.
SB% -- 3rd at 70%, +3% vs. league avg.
XBT% -- 5th (tie) at 45%, equal to league avg.
R/G -- 5th at 4.19, +0.09 vs. league avg.
HR -- 12th with 83, -34 vs. league avg.
1984
SB -- 1st with 220, +76 vs. league avg.
CS -- 3rd with 71, +7 vs. league avg.
SB% -- 2nd at 76%, +7% vs. league avg.
XBT% -- 1st at 51%, +4% vs. league avg.
R/G -- 5th (tie) at 4.02, -0.04 vs. league avg.
HR -- 12th with 75, -32 vs. league avg.
1985
SB -- 1st with 314, +178 vs. league avg.
CS -- 1st with 96, +36 vs. league avg.
SB% -- 2nd at 77%, +7% vs. league avg.
XBT% -- 3rd at 47%, +4% vs. league avg.
R/G -- 1st at 4.61, +0.54 vs. league avg.
HR -- 11th with 87, -32 vs. league avg.
1986
SB -- 1st with 262, +108 vs. league avg.
CS -- 4th with 78, +6 vs. league avg.
SB% -- 1st at 77%, +9% vs. league avg.
XBT% -- 3rd at 47%, +2% vs. league avg.
R/G -- 12th at 3.73, -0.45 vs. league avg.
HR -- 12th with 58, -69 vs. league avg.
1987
SB -- 1st with 248, +94 vs. league avg.
CS -- 4th with 72, +9 vs. league avg.
SB% -- 2nd at 78%, +7% vs. league avg.
XBT% -- 7th at 45%, -1% vs. league avg.
R/G -- 2nd at 4.93, +0.41 vs. league avg.
HR -- 12th with 94, -58 vs. league avg.
What does it all add up to? I'll let the smart guys explain it.
All I can tell is ... Those Cards teams, as a group, stole far more bases than any other team, and at a good percentage. But their XBT% overall does not immediately mark them as an aggressive baserunning team.
April 27th, 2011 at 11:58 am
@5, John said: "Teams that don't generate a ton of offense are going to be a lot more aggressive on the basepaths, to squeak out more and more runs."
Sounds good in theory, but has it actually been proven? I haven't seen the evidence. Anecdotally, I've seen plenty of low-scoring teams that are also passive on the bases, simply because they don't have the speed or skill to be aggressive.
April 27th, 2011 at 1:02 pm
It'd be a tough thing to prove; I guess my thinking on the matter goes: if you're not stocking your team with guys who are good at the big things (reaching base, slugging), those guys might as well be good at the little things (or they're defensive studs, but defensive studs tend to have some speed by and large, outside of catcher and corner infielders).
April 27th, 2011 at 1:03 pm
"Note that XBT% does not do a great job of measuring the gain or loss of a more aggressive base running strategy. There are 2 ways to fail: only advance the minimum of bases on the hit, or to be thrown out trying to gain the extra base. XBT% treats these failures equally, while getting thrown out is obviously much much worse than not advancing."
It also means that a team could be highly aggressive and just bad at it. A team could try for the extra base every time and only succeed 25% of the time. This would be far worse than a team that tries 30% of the time and succeeds 50% of the time when it does try (they'd have an XBT% of 15% but far fewer outs on the basepaths). Ideally, we could break this down to opportunities, attempts, successes, and failures.
April 27th, 2011 at 1:17 pm
Right - it's the same principle as the stolen base. A team that steals 200 bags isn't doing itself any favors if it gets caught 140 times.
April 27th, 2011 at 1:46 pm
You also have to account for the Yankees low % as an EARL WEAVER-type thought process; coaching for the 3 run homer.
The Yankees have 6 guys who can hit 30 HRs.
I'd say caution works to their advantage.
For the small market/small ball theory, I think it makes sense but mostly depends on the manager. Ozzie Guillen has plenty of big bats and a big budget, but the CHISOX steal and sacrifice a ton. I also think the more faith an individual manager places in their bullpen, the less likely they are to try and stretch a 1 or 2 run lead by playing small ball. Ergo, the Yankees may feel less pressure in the eighth inning, up a run, knowing they have Mo.
Who knows.
April 27th, 2011 at 3:06 pm
Is there anything to the old saw about putting pressure on the defense or is it just a myth? JA's data on the '80's Cardinals suggest that their XB taken was merely average but they had a bunch of greyhounds at the top of the order.
My brain must be cramping but why is there the odd number missing from the baserunning events? For example why did Oakland have 48 runners on 1B situations but only a total of 47 outcomes? Somebody scored or was doubled at second? I must be tired at work, sneaking on this site......looking over my shoulder!
April 27th, 2011 at 4:00 pm
What XBT% may correlate well to is hitting behind the runner. All alse being equal, a runner is most likely to take an extra base (especially 3rd base) on a hit to RF, and progressively less likely to do so on hits to CF and LF.
No idea how to try to measure that but, intuitively, my hunch would be that teams that hit more to right field should also have higher XBT%.
April 27th, 2011 at 5:03 pm
Neil L. @15. When they don't add up it means runner(s) were lost trying to take the extra base. The Twins have lost a triplet of base runners trying to score from 2nd on a single. Part of what we are measuring here is not base running skill but 3rd base coaching skill.
Doug @16. Then we should be able to see a correlation between XBT% and LHB PA%. Maybe the yankees are just too old ,slow and HR oriented for it to show up here.
April 27th, 2011 at 5:23 pm
[...] What does XBT% really tell us?: Andy K. of B-R wonders about the significance of XBT%, the percentage of extra bases taken. [...]