Bringing the infield in is a high risk high reward tactic to prevent a base runner scoring from third base. Having the infielders play in or close to the grass makes it much harder for the runner to score on a ground ball, but the strategy opens more holes in the field for bloop singles. The fielders are also closer to the plate so they have less time to react to a ball hit at them. Because teams allow so many extra hits in this defensive alignment, it is most often employed late in a game with the winning or tying run on third. Exactly how big is the tradeoff between more hits and fewer runs scored on balls fielded? Perhaps teams aren’t calling on this strategy optimally.
Collecting Data
I watched 205 clips from last season to see what happens when a team has their infielders play in near the grass. I used Baseball Savant’s play search to pick which plays I viewed. I set the parameters as: balls in play, in the 8th inning or later, runner on third, 2019 season, 0 or 1 outs, tying or go ahead runner on base, and strategic infield alignment. "Strategic Infield Alignment" is counted as any alignment that is not standard or a normal infield shift with 3 defenders to one side of second base. After watching the clips, I charted results from the 183 plays where the infield was in.
I soon decided under 200 plate appearances is nowhere enough to draw a solid conclusion. I was most interested in finding the BABIP, which needs at least 800 at bats to stabilize. Instead of watching every play since 2015 that met the criteria mentioned above, I just took the results from baseball savant. About 10% of these plays (10.7% in 2019) are probably against a normal defense, so the results aren’t exact. After charting these at bats, I had 899 to analyze.
For 2019 I kept track of the lead runner and where the ball was hit. When I stopped watching videos, I stopped differentiating between the runner being thrown out at home and staying put at 3rd.
I also used Baseball Reference and Baseball Savant to see how likely a plate appearance ended in a specific result and how often the run scored from 3rd. I used those results and the .362 BABIP from above to adjust for when the infield plays in. I assumed all extra hits because of the shift were RBI singles. To keep the math simple, I assumed half of the extra hits came from outs that would’ve scored a run and half from outs that would not score a run. That allowed me to make the following charts:
I assumed runs never scored on strikeouts, but there may have been a couple that scored on a wild pitch K.
I used numbers from the first chart to calculate how often a run scored on an out when the infield played in (644 runs minus 381 hits divided by 657 outs yields 40%). I used percentages instead of total number of plays because this chart is an estimate based on the other chart with a standard defense. Hits in play increased by 4.77% because the BABIP times percent of balls in play was 4.77 higher with the infield in.
This table summarizes the two tables above.
This chart shows the frequency of individual hit types. Singles are responsible for 62% of all hits, and 74% of hits on those that are in play.
Next, I used a RE24 chart and the information from the tables above to see which defense is better when there is only a runner on third, and less than two outs. If you are unfamiliar with run expectancy, it gives the average number of runs scored for the rest of the inning. There are 24 possible situations (3 out possibilities times 8 different runner combinations). A higher number means the offense is expected to score more runs.
I took the chance of each play happening and multiplied it by the run expectancy change if that particular result occurred. RE change is the new RE minus .95 which is the starting run expectancy.
For an infield-in alignment, I used data from the first chart of this article to determine how often the out came at home (18 outs at home out of 106 total outs is 17%) . This is important because it is much more valuable to the defense.
Same story, but now with 0 outs in the inning
If I assumed all outs were made by the batter instead of the runner, the total change in RE would be 3.305 instead of 1.465. Either way, it is significantly lower than when there is a normal defense set.
I expected the total RE changes to be closer to 0, but I think that comes from the RE24 chart and play data coming from different seasons. More runs are being scored now than 5 years ago, so it makes sense the totals add up to above 0.
Discussion
The final BABIP (after accounting for sac bunts and homeruns) with the infeld in was .362, which is pretty high. If the defense was standard, we would expect the number to be around .300. This means over the past 5 seasons there were an extra 55 hits, or 11 per season (.062 more hits from BABIP times 899 PA’s).
Despite having a higher batting average with infield in, teams scored 3.03% less often. This is because the defense did a much better job of holding the runner on third when the ball was fielded.
For the first two charts using run expectancy, the total sum was negative. This means after having a plate appearance with a runner on third and one out, the batting team was expected to score slightly fewer runs. I could not find data specifying which runner made the out against a normal defense, so I assumed it came from the batter instead of the runner on 3rd. This made the math simpler, but resulted in the total number being a little closer to 0. While the difference between the two defenses with 1 out already in the inning is .956, it is probably slightly lower in reality. The total was greater with the infield in, so it was the better defense in this situation.
With 0 outs in the inning, both total sums were positive. This means teams saw their run expectancy increase when batting in this situation. The number was closer to 0 with the infield in, so this defensive alignment did a better job preventing runs. Because the RE24 chart I used was not from the same year I pulled stats and frequency of individual hit types, the results are not exact. That being said, the infield-in defense was better by 3.509 runs (or .035 runs per plate appearance). In 600 plate appearances (about a full season for most players) this becomes over 21 runs which is the difference between an average and above average player.
MLB teams should play the infield in with a runner on third regardless of the inning. The opposing team will score less often and fewer total runs. One difference when bringing the infielders in early in a game is the runner on third will likely be less aggressive. His run will be less important earlier in the game, so he will take fewer risks. One scenario a team might choose to keep a standard defense is if they are already up by a lot. When winning big, the runner on third is nearly meaningless. All focus should be on limiting a big inning, which would be more common with the infield in. If a team is up by a lot, they would be more worried about giving up extra hits than allowing the runner to score on an out.
Conclusions
A run is 3.03% less likely to score in a plate appearance if the infield is playing in.
A runner is 13.09% more likely to stay at 3rd or be thrown out at home on an out when the infield is playing in.
A batter is 4.77% more likely to get a hit when the infield is playing in.
Teams are expected to score fewer runs in an inning when the infield plays in with a runner at third.
TLDR: If you want to win, play the infield in