Analytics and its effects on the MLB – The Bunt

Each year I see another ex-professional baseball player come out and say analytics is ‘killing the game’, this year it was the ex-Nationals and Phillies outfielders Jayson Werth. In August he said “I think it’s killing the game. It’s to the point where [we could] just put computers out there. Just put laptops and what have you, just put them out there and let them play. We don’t even need to go out there anymore. It’s a joke.” If you have read any of my previous articles you get that this isn’t a position I agree with.

This the typical statement coming from the ‘old school’ player who doesn’t like what is happening to the game but Werth isn’t that old school. He retired mid-way through this season after 15 years of service in MLB, Jayson would have played through the changes analytics has brought about and obviously he didn’t like what he was seeing happening around him.

All sports develop overtime and in baseball the flux between the game being in the favour of the hitters or the pitchers has changed many times over the MLBs lifespan since 1871.  Multiple times in MLBs past has the organisation made rather drastic rule changes to counter act a perceived bias, the strike zone and the height of the pitching mound have been reduced historically to counter the dominance of pitchers.

Usually the changes happen slowly over decades, eventually the league decides enough is enough and introduces some new rule. What the increase use of analytics by the MLB teams has done is speed up that process.  They can far more accurately and in a shorter time determine if a new playing style is working or not, also other teams can work this out as well and can choose to adopt this sooner.  This gives players/managers less chance to adapt to the changes and therefore these changes can be perceived to amend the game rapidly.

So what is perceived to be ‘wrong’ with baseball that analytics is at fault for? Well pretty much any major change of style of play in the last 20 years you can probably subscribe to analytics having some impact but for the last 5-10 years it definitely is responsible. I have highlighted the 5 below as the main things I have seen as ‘killing’ baseball.

Less Bunting

Less Stolen Bases

More Shifting

Less Balls In Play & More Strikeouts

Too Many Pitching Changes & Games Are Too Long

A lot of these can be thought of as they don’t play the game like they used to back in my day, which can be a valid complaint for older generations but in most sports it doesn’t effect the game as there are newer fans coming through who like this version of the game as it is what they know.  The issue baseball has as it hasn’t had the growth in the younger demographics that it would like, so the voice of the older generation is louder and appeasing this fans is something MLB has to look into.

I will be writing articles about all of these over the course of the off season, with the analytical reason behind the changes and why it may upset people. I will also postulate a potential change which may improve the game for each of them but will also show the negative effects of making any change.

But here is a teaser for the series to come. Why do we see less bunting?

Firstly lets cover the basics, what is a bunt?  A bunt is when the batter loosely holds the bat in front of the plate and intentionally taps the ball into play. The aim can be two fold either for the batter to get on base themselves with the hit or allow a baserunner to advance in exchange to the batter being thrown out.  The latter of those is called a sacrifice bunt.

Why do people like the bunt? I think this comes down to it being different and adding a bit of variety to a game. It makes it look like a team is thinking more because they have shown different playing options on the field. It also tasks the infielders, pitchers and catchers to do something different so we can see the full range of their abilities.

What is the purpose of the bunt? The sacrifice bunt has used in MLB because there was an idea that advancing a runner and having 1 more out puts the team in a better position than before. But analytics has shown this not to be true for almost all scenarios when the batter isn’t a pitcher.  The piece of analysis which was created that destroyed the case for the sacrifice bunt is the run expectancy table, this looks at a situation (no. of outs and position of base runners) to determine how many runs you should expect to get in this scenario based of what has happened historically.

Above is the Run Expectancy table based off the 2010-2015 seasons, I will just go through a couple of scenarios so you understand what is says.

Example 1 – Start of an Innings (Green) – With 0 on and 0 outs, the average number runs scored would be 0.481 but only 26.8% of teams will score any runs. This difference between the two being some teams will score more than 1 run.

Example 2 – Bases Loaded, No Outs (Orange) – With 3 on and 0 outs, the average number runs scored would be 2.292 and 86.1% of teams would score any runs. That means that roughly 1 in 7 teams fails to score any runs when they load the bases up with 0 outs.

So lets look at our typical sacrifice bunt scenarios when we have a runner on first/second/both with 0 outs.

Scenario 1 – Man on First, No Outs (Yellow) – The batting team would score on average .859 runs with them scoring 41.6% of the time. If I was ‘successful’ with my sacrifice bunt and the runner is on second and we now have 1 out, the team would score on average .664 runs with them scoring 39.7% of the time. So the sacrifice bunt has brought down the chance of scoring a run by 1.9% and decreased the average about of runs scored by .195.

Scenario 2 – Man on Second, No Outs (Blue) – The batting team would score on average 1.1 runs with them scoring 61.4% of the time. If I was ‘successful’ with my sacrifice bunt and the runner is on second and we now have 1 out, the team would score on average .950 runs with them scoring 66% of the time. So the sacrifice bunt has increased the chance of scoring a run by 4.6% but decreased the average about of runs scored by .15.

Scenario 3 – Men on First and Second, No Outs (Red) – The batting team would score on average 1.437 runs with them scoring 61% of the time. If I was ‘successful’ with my sacrifice bunt and the runner is on second and we now have 1 out, the team would score on average 1.376 runs with them scoring 67.6% of the time. So the sacrifice bunt has increased the chance of scoring a run by 6.6% but decreased the average about of runs scored by .061.

Both scenario 2 & 3 increase your likelihood of scoring a single run but bring down your chances for more so they should only be used when the team only needs to score 1 run. The perfect timing for this would be in the bottom of the ninth in a tied game but I can imagine a few players would look at a manager funny if I told them to go up to the plate and bunt Kenley Jansen’s slider or Aroldis Chapman’s 100+ mph fastballs.

All of those scenarios are based on being at 0 outs before the sacrifice but if you were to look at the numbers for when you had 1 out then there is no scenario which give positive results. This holds true for expected run charts going back to the 1950s and is for reason we have seen sacrifice bunts decrease over the last few decades. This was first wrote about by George Lindsay in the 1963, it took a long time for the baseball world to accept his ideas.

So the sacrifice bunt, in itself, isn’t really a viable option for batters to use in the modern game as in most scenarios you making your team worse off by doing it. But what about attempting to bunt for a single? If we aren’t just going for the sacrifice and really attempting the single as well that adds a little extra up lift to the bunt which we should consider when evaluating its worth.

This leaves us three scenarios when looking to complete a bunt, get a single and all runners safely advance 1 base, ending up with a sacrifice bunt or failing to advance the runners by for example striking out or bunting into fielder’s choice. I have excluded the potential for a double/triple play as this don’t happen that often. So let’s look at the impact on run expectancy for each of these scenarios.

So, the table above shows the changes to run expectancy based on our 3 scenarios, plus a 4th of 0 on, 0 outs, based on there being no outs or 1 out before the bunt attempt. Also it shows the batting average you would have to better when bunting for this to be worthwhile over a season.

Without anyone on base you have to have a bunt batting average of better than .375 for it to be a worthwhile endeavour. Let’s walk through these tables for our previous scenario 1 (Yellow) so you can understand it better. With 0 outs and a runner on 1st, if you are successful at getting the runner across to second you have to get to 1st safely more than 25.2% of the time for it to be worthwhile overall.

But for scenario 3 (red) you only have to successfully bunt for a single batter than 6.7% of the time (2 out of 30) for it to be a positive endeavour overall, if they managed to advance the runners. That seems like an entirely plausible amount to expect off MLB but it is it, what is the ‘success’ rate of a non-pitcher bunting in the MLB.

According to Fangraphs, in the 2018 season there was 1,327 attempted bunts. Of which 880 were against a shifted infield and 447 against no shift. This gives us some rates for the likelihood of singles, sacrifices and outs from MLB players attempting bunts. I used these rates and multiplied them against the expected change from above to see if we get any overall positive outcomes. The reason I have split the bunts for the shift is that you have two very different scenarios as the shifted batters are bunting for a gap whereas the non-shifted bunters are attempting the more traditional infield bunt.

I will be honest I was surprised by these results, it shows that bunting in certain scenarios it has a positive impact, it still shows that you should really only be looking to do this in a no outs situation but there are runs to be gained.  What really jumps from this is the numbers for non-shifted batters which suggests that at the 2018 success rate bunt is a viable option across all scenarios.

Bunting against no shift is viable in 2018 but has that always been the case? Over the last 4 years, yes but for 2014 and before it wasn’t because the sacrifice rate was too high. Since the start of the 2015 season 38.9% of bunts against a non-shifted infield have become a hit where as for the previous 5 seasons on 25.7% did.

The reason behind this increase is two-fold, a decrease in sacrifice bunts and the fact that the players who are still bunting on the whole have an above MLB average sprint speed. Of the top 30 players who have bunted against no shift by plate appearances only 2 of them have a sprint speed below the league average of 27ft/s and the average for this top 30 is 28.6ft/s. The man who tops this list is Cesar Hernandez, who has 22 singles from 39 bunts, registered an average sprint speed of 29.2ft/s in 2018. This includes a bunt this year which he managed to score on thanks to two errors from the fielding team, video here.

These top speedsters have bunt to hit rates of 50+% so this should be something more of the fast players should be considering.  I am uncertain if we will see an increase in the number of attempted bunts from these players but it should become part of their repertoire. Adding anything to a batters repertoire is good from them as the opposing manager/catcher/pitcher has to think about it as part of their analysis. There may be a surprise factor at play here as well

In conclusion historical analysis showed, and current analysis still shows, that a sacrifice bunt wasn’t worth it, so slowly over time managers have stopped telling their players to sacrifice bunt. In the last few years the speedsters in the MLB have showed that for them a infield bunt is a viable option for getting a single. Here is hoping to seeing more of these next year but actually suspect we might see more bunts overall next year due to ones against the shift.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.