In my previous piece, on the intro to infield outs above average (OAA), I quickly compared Marcus Semien and Javier Baez. I showed you that, in 2019, Baez had 19 OAA and Semien has -4 OAA. This analytically tells us that Baez was a better shortstop than Semien. But what if I told you they were both as useful as each other to their teams in 2019. Doesn’t seem right, does it?

You see, both were successful in getting the out on the play on 88% of the time, when they were the infielder responsible for the play. This happened because Baez was only expected to be successful 83% of the time on his plays where as Semien was expected to be successful 88% of the time on his. Baez’s ability brought him in line with Semien’s output. But what could be the main differences behind Semien and Baez having such a large gap in expected success?

**Estimating Success**

If we think about the breakdown of probability of a play as created in the OAA model, there are three action points which make up the overall probability. These can be simplified to: getting to the ball, fielding it cleanly & having the time to throw it to get the out.

The probabilities on the final part are independent of the initial two with only the runner’s speed being a variable that could be different. I would be very much surprised if there was any difference here on average for Baez and Semien across there 400 and 500 plays respectively. So, I will ignore this part in determining the difference.

For the other two action points, the two main factors are the speed and direction of the batted ball and the starting location of the infielder. Unsurprisingly, these are the two things which will differentiate Baez’s and Semien’s expected success rates are where they were positioned and how hard the balls were hit at them.

So, if we can say that the types of ball faced by both players were similar then we can suggest that difference in estimated success rate is down to where the player has been positioned.

Thanks to the search functionality on Baseball Savant we can get a high level view of the hitting data. I searched for ground ball and line drives (with a launch angle below 10 degrees) which went to the shortstop area (in between second base and 20 degrees to the left of it). I took the data for the league, the Cubs and the Athletics to determine if there was any difference.

Although this includes some data from other players filling in at shortstop, it should give a good overview as both Baez and Semien were the main starters and took the majority of reps at the position.

As you can see, the Cubs saw slightly above average exit velocities and the Athletics saw slightly below. I would be amazed if this difference led to any significant difference in the OAA model and therefore based off this data point, I believe it is reasonable to say that most of the difference in estimated success rate is down to where the player has been positioned.

**The cost of
misplacement**

That means that either the Cubs or Baez himself have basically negated his ability by starting in a comparatively ‘bad’ position. If we compare Baez to the rest of the league’s shortstops who had 250+ plays in 2019, he is 27^{th} out of 27 for estimated success rate. A full 2% off 2^{nd} worst, which is significant given that the rest of 26 players are separated by just 4%.

Here is the top 5 and bottom 5 for estimated success rate.

And the top 5 and bottom 5 for actual success rate.

Here you can see that Andrelton Simmons comes to the front, his estimated success of 87% plus his immense abilit,y means he is getting outs on 93% of the plays he is responsible for, which is 10% more than the worst shortstops. Given a shortstop can easily see 400 plays (some get to 500) in a season by positioning and ability, Simmons and the Angels could get 40 more outs than Jorge Polanco and the Twins.

Given Baez had 397 plays for which he was responsible, and that the average estimated success rate for all shortstops is in 2019 was 87%. You could say that the positioning of Baez cost the Cubs 16 outs, which is roughly 12.5 runs and 1.2 wins. All of which he made back by being an amazing defender…

**Baez is still great though**

But to Baez’s credit this shows what an outstanding shortstop he is. If we compare him to the others in the OAA Top 5 SS for 2019 you can see that where he generates his OAA is much different to the others – by making those ‘highlight plays’ which have become his trademark.

Baez is seeing more low probability plays than any of these other guys and has made significantly more OAA from these plays. Baez makes some fantastic plays and makes very few mistakes on the easy plays (96% success rate on plays over 85%, expected 94%).

Let’s compare him to Nick Ahmed, who is basically average on plays below 85%, but makes even less mistakes than Baez on the easy plays (98% success rate on plays over 85%, expected 94%). If you were to give Baez plays to Ahmed, given the same success rate, we would expect Ahmed to have roughly 11 OAA (8 worse than Baez). Whereas we would expect Baez to still have 19 OAA with from Ahmed’s plays (3 better than Ahmed).

So, Ahmed has a high OAA as the Cardinals managed for him to be in a good position before the play started, but Baez was great even when the Cubs’ positioning handicapped him.

**Positioning is the
Key**

If we work with this hypothesis, that the difference between a team’s estimated success rate and the league average is down to positioning, we can approximate how well or badly teams have been setting up. This is unlikely to be as true as the specific example above as teams see some variance in the groundballs and line drives faced – using the same filter as before the average exit velocity of teams ranged from 84.9 mph (Mets) to 88.1 mph (Royals).

Nevertheless, here are the teams estimated success rates for each of the infield positions.

You can see that for quite a few teams the expected success rates are either all better or all worse than average. With this we can look at the number of outs which teams are expected to get compared to average. These are based from estimated totals of plays at each position.

Before the infielders do anything, this estimates that Yankees would have gotten 19 more outs than average, if they had average infielders in every position. Which is equivalent to approximately 15 runs and 1.5 wins. Whereas the Brewers are -36 OAA and nearly 3 wins worse than average.

Now, as I have said before there are quite a lot of assumptions that have been made here but given the Yankees and Royals are not teams which get soft contact on the grounders they give up compared to average and the Dodgers are it’s likely there’s minimal impact here.

If these assumptions are true, it would be fascinating to see Statcast split the data into ‘shift’ or ‘no shift’ to help us estimate the impact of these set ups in position on overall outcomes.

I will continue to investigate this but for now we have an interesting initial conclusion that positioning of players correctly/incorrectly could get/cost teams the all-important win they need to get into the playoffs.