08 March 2016

Why Swinging More Won't Help Chris Davis

Previously, I showed how Chris Davis propensity to strike out decreases his overall offensive production despite a decent walk rate. I made two questionable assumptions in that article. I assumed that players will hit the same proportion of pitches in the strike zone regardless of swing rate. I also assumed that production per pitch put into play would remain constant regardless of swing rate. These assumptions seem to be counterintuitive. Batters that swing less frequently are expected to swing at only the best pitches. It makes sense to take a closer look to see how batters perform.

In order to measure this, I used a pitchf/x dataset with data from 2013, 2014 and 2015 and determined all batters that faced 1,000 pitches or more in a given season. For each batter, I determined their monthly performance in a given season for all months from April to September while discarding players that didn’t play in each month from the dataset. Then, I sorted each player’s monthly performance in a season by swing rate. Finally, I grouped the players by that ranking, so that the month when a player had his lowest swing rate in a season was ranked 1 and the month with their highest swing rate was ranked 6.

For example, look at Adam Jones in 2013. He had his lowest swing rate in April, so that month was ranked “1”. His next lowest swing rate was in July, and therefore that month was ranked “2”. The next lowest was September, so that month was ranked “3” and etc. His monthly performance in a given year is placed in a different category based on how his swing rate in that month compares to his swing rate in other months.

For Chris Davis, I looked solely at 2013 and 2015 to avoid being biased by poor performance in 2014 and ranked his performance by month into two groups based on swing rate with six months in a low swing group and six months in a high swing group. This methodology allows one to compare players to themselves and measure the impact that reducing or increasing swing rate has on the results.

This first chart shows there’s considerable variation in a batters’ swing rate over a given year as measured in units of months. Every batter is in each group once and only once, but Swing% goes from 42.77% in the group where batters swing least frequently to 51.43% in the group where batters swing most frequently. As swing rate decreases, the chance of a called ball increases while the chance of a strike slightly decreases. The ratio of called balls to called strikes remains constant. As swing rate decreases, called ball rate increases and strike rate stays static.

This second chart shows the frequency of where pitchers throw the ball and when batters swing and put the ball into play. As swing rate goes down, batters swing less at pitches in all areas of the strike zone. Batters do put a larger proportion of pitches in the strike zone into play as swing rate decreases but the absolute impact is minimal. Batters put few pitches out of the strike zone into play regardless of swing rate and therefore it’s going to have only a small impact on actual production. Players that swing less will put a higher proportion of pitches in the strike zone into play than pitches outside of the strike zone, but the impact is minimal in absolute terms.

This third chart shows production based on grouping. As swing rate decreases, walk rate increases while strikeout rate remains constant leading to large increases in “wOBA Not In Play”.  This makes sense because the called ball rate is increasing considerably while in play rate is decreasing as swing rate goes down resulting in more walks. Meanwhile, strike/foul rate decreasing suggests there should be fewer strikeouts while in play rate decreasing, suggests there should be more strikeouts. It would seem these two factors cancel each other out.

Also note how the increase in both BB% and “wOBA Not In Play” is reasonably linear and is split reasonably evenly among all categories. This is what one would expect to see if swing rate has an impact on wOBA Not in Play. It should come as no surprise that there’s a moderate correlation of -.5 between Swing% and wOBA Not in Play.

There’s only a small increase in wOBA in play. It’s constant when pitches aren’t thrown in the strike zone and there is only a slight change when pitches are thrown in the strike zone. These differences aren’t particularly linear or split evenly among all categories. It should come as no surprise that even the correlation between Swing% and wOBA In Play for strikes is .03. In all likelihood, the difference in wOBA in play between these groups is probably nothing more than luck. This indicates that production of pitches put into play is largely independent of swing rate.

One might think the two assumptions made above could have huge implications, but it turns out that one has a minimal effect and the other has no effect.

Chris Davis’ stats tell a similar but unique story. He has a well above average called ball/called strike ratio regardless of whether he’s being aggressive or not. But his contact tool is so poor that being aggressive does little to help him put pitches into play. The Hardball Times noted that Chris Davis' "True Contact" rate is one of the worst in the majors (462 out of 498). In fact, his contact tool can be described by this:

Despite a 4% increase in swing rate between his most aggressive months and most selective months, he only had a .27% increase in in play rate. This suggests that he only put 6.8% of these extra pitches into play, which is less than half of his base rate.

This next chart shows that pitchers are only throwing roughly 34-35% of pitches to him in the strike zone compared to a standard of 38-40%. He does do a good job swinging at strikes rather than balls suggesting that he has an above-average eye.

But if pitchers are going to give him relatively few pitches to hit, then he needs to swing at relatively few pitches. Otherwise, his aggression just results in a lot of fouled pitches, swinging strikes and ultimately strike outs. He should probably be swinging closer to 37% of the time instead of 47%. This would likely result in more walks, fewer strikeouts and being thrown a larger proportion of strikes.

This third chart shows his actual results. He swings so frequently and has such a poor contact tool that a reduction in swinging resulted in fewer strikeouts and more walks. While players typically want to put pitches into play, an increase of BB% by 3.25 and a decrease of K% by 2.14 is an extremely favorable trade for every batter that will ever play baseball.

The data indicate that he has better results when putting pitches into play when he swings often. This is counter to the results in the original data set and could be a reason for him to swing frequently. Alternatively, it could be a function of small sample size as he only put roughly 360 pitches into play total in both sets of six months.

Swinging more often isn’t likely to help him improve by much. Suppose we go back into time and make the following change. For each at bat that he doesn’t put the ball into play, we decide that he will swing at every pitch in the strike zone that he didn’t actually swing at in real life. Given his inability to make contact, such a change would only have a limited impact on his overall production. His walk rate would drop from about 10% to 9%, his strikeout rate would drop from roughly 31% to 25% and his wOBA would increase by only 6%. This shows that a higher swing rate, even presuming it would occur in only potentially beneficial situations, would have minimal impact on his production.

The above analysis shows that production per pitch put into play is largely independent of swing rate for the average batter. It is possible that hitters that swing at an extremely high or low percent of pitches do better or worse than the average. The assumption that I made in my article that Adam Jones would have a constant wOBA In Play regardless of swing rate was close enough to accurate.

This analysis also illustrates how poor Chris Davis is at making contact. Such a weakness makes it unlikely that he’ll ever be able to make average levels of contact even if he swings at extremely high levels. His aggressiveness pretty much solely results in more swinging strikes and fouls rather than pitches put into play. Therefore, he should consider swinging less frequently in order to increase his walk rate and likely decrease his strikeout rates. The average player may do better when they put the ball into play rather than when they don’t, but contact simply isn’t in Chris Davis’ toolbox. Players with extreme power but limited plate discipline would seem to be more successful with an extremely selective approach. And, like Mr. Burns, at Camden Depot we like to play the percentages.


Roger said...

Even without reading this analysis I think one could surmise the same result from watching some of the gawdawful swing and misses he has. IF you're correct and he does have a good eye then the answer is certainly to not just swing at strikes but to swing at strikes in his hot zone. He should reduce his effective vision of the "strike zone" to his hot zone. The impact I think you're telling us is that we will see more Ks on takes, more walks, and fewer swinging Ks.

I think this is what Barry Bonds did when he had every umpire convinced that if he didn't swing them it must have been a ball...... And it didn't impact his HR rate very much.

Matt Perez said...

He can't just swing at strikes only in his hot zone because he's a real person. He will get fooled by pitches that he thinks are in his hot zone but aren't and will think that some pitches aren't in his hot zone but they are. He can pretty much control the rate at which he swings. Hopefully if he swings less, he'll increase the proportion of good pitches that he swings at compared to bad pitches.

The impact for him by swinging less has been an increase in walks and a decrease in strikeouts. His increase in called strikes by swinging less has been more than offset by the decrease in foul balls and swinging strikes. The real question is whether this will continue linearly if he swings 37% of the time instead of 47%. If so, he could easily have an 18% walk rate and a 26% strikeout rate with limited impact on the number of balls he puts into play. That will help the old OBP and wOBA.

Roger said...

Can someone tell him?