Wichita State Pitching Can't Strand Baserunners

Monday, March 28, 2016

Quick Blog:

Wichita State pitchers are struggling. They lost All-American and potential first round draft pick Sam Tewes to injury and the staff has felt that load. As far as I can tell the staff - minus Tewes - are regressing to the numbers they put up last year. Their BABIP has come back down under .350 (It was .342 last year) and their K% and BB% are now identical to what they were last year. The K% is 19.6% and the BB% is at 11.3% which is honestly not terrible. For comparison, last year their K% and BB% was 19.7% and 11.7% respectively.

The line drive percentage (LD%) is also nearly what it was last year. LD% gives us an estimate of how many hard hit balls the pitching staff is giving up per at-bat. It is not exact, but it is a close approximate. The LD% for this year is 38% and last season it was 37%. So it is within the margin or error.

With luck becoming less and less of a factor we see regression to the mean in full force. That being said the Fielding Independent Pitching (FIP) stat is actually considerably lower this year for the entire staff - 4.10 this year compared to 4.55 last year. The same can be said for ERA - 5.10 this year and 6.10 last year. What gives?

The answer lies with runners on base starting with the percentage of runs scored per batters faced. This ratio is high to say the least. Last season the Shockers ended with a 14.4% Runs/Per Batter Faced metric. This year that number has ballooned to 17.2%. They are allowing 17 of every 100 opposing batters to score!! The main culprit of this increase is how the pitchers have performed with runners on base.

Left-on-base percentage is a stat that tells us how well pitchers are stranding runners. The higher the number the better they are at doing just this. Unfortunately, WSU is not doing well in this department. Check out the table below comparing Wichita State pitching LOB% over the last three years including 2016.


Wichita state cannot keep opposing runners from scoring once they reach base. Overall the pitching staff is doing poorly in this category and it doesn't get any better if you look at individual players. Willie Schwanke's LOB% is down more than 10% from last year. That is insane! McGinnis, Jones, and Heuer numbers for this metric are sub 50% and they have a good deal of innings pitched and batters faced.

A team cannot compete if they cannot find some way to increase this ratio. A place to start would be looking at bringing in high strikeout guys with runners in scoring position especially later in the game. Strikeout percentage is the most important stat to look at when analyzing relief pitchers. The Shockers have some guys that might be able to do this, but some are being used as starters - and rightly so. However, there are some younger freshman that have a K% - albeit with limited batters faced. At this point though all resources should be considered.

Wichita State Baseball: Comparison of Each Batter's Contribution This Season

Tuesday, March 15, 2016

I finished the first iteration of parsing the NCAA D1 baseball play-by-play data today. Now I want to take a look at some advanced metrics that can be used from this information. The first one I am looking at is the run contribution each batter has made to the team this season (2016).

Let me explain briefly what I mean by run contribution and how it fits into evaluating baseball players. For any given plate appearance there is a potential to score a run. We also know not every batting play is equal. An obvious one is that a home run is more valuable than a hit with no runners on. However a hit with the bases loaded could be more valuable than a home run. The play-by-play data allows us to create a run expectancy matrix for all of the possible runners on base states which is 8 and the possible number of outs which is 3 (0,1, and 2). This matrix showed me the run potential for each of those possible scenarios.

Note: This matrix is not 100 percent accurate. It is based on the my first attempt to parse the play-by-play data and seems to be slightly underestimating each state.

For example with a runner on first and no outs the average runs scored by all NCAA D1 teams for the rest of the inning where this is the scenario is - by my calculation - +.86 or about +1. If a player gets a lead-off single, that team should score on average around one run in that inning. Why does this matter? If we know the run expectancy of each state we can determine the run contribution of each batter based on how they performed in each one of those states. Knowing the run contribution of each player allows us to better evaluate the players. It let's us understand how well each batter performed in every scenario from no runners on and no outs to bases loaded with two outs. Total all of those scenarios together and you can see the total contribution of that player. Runs Created is not RBI's or Hits it is a new value generated by estimating the run contribution of each batting play of the season.

Without further ado....

Run Contribution Chart For Players With More Than 10 AB's

No surprise Troutwine tops the list with nearly 6 Runs Created. To put things into perspective. Corey Ray from Louisville University's - one of the best players in college baseball - has a Runs Created stat of around 7. The players who have negative values should not be discouraged because it is early in the season. A solid player will end with a Runs Created (RC) around 15 based on the averages I calculated from the play-by-play data from the 2015 season.

As you can see from the chart, a high or low OPS (OBPct + SLGPct) does not necessarily mean that player is contributing more or less than another player. For example, a player who hit ten triples in ten at-bats with the bases empty would have a very high OPS and a decent Runs Created value. On the other hand, a player who hit ten doubles in ten at-bats with the bases loaded would have a lower OPS than the previous player, but a higher Runs Created. Obviously this is an extreme hypothetical, but I was trying to make a point. Runs Created goes beyond the triple slash to paint a more accurate picture of the contribution the team is getting from each batter.

As I mentioned above, the RC stat can be used to see how well batters performed in each type of situation such as with the bases loaded and so on. Below are visualizations of this exact measure for all of the players listed in the chart above. Notice the line at the zero mark in each chart. Points above the line represent positive run values contributed for each base state. Points below represent negative run values such as when the batter makes an out or performs some other type of low level outcome like being hit by the pitch or sac bunt. The more points above the line the better. Please note you may not be able to see all the points as some overlap each other. This can make a few of the charts looked skewed.

In the next post I will use this research along with other baseball statistics and research to create an optimized lineup for Wichita State. Preliminary analysis looks like Shocker baseball could be slotting guys in the order differently to achieve better results over the course of the season. As it stands guys are being placed in the batting order where it is not maximizing their individual benefits.











Is Wichita State Baseball Suffering From Bad Luck?

Tuesday, March 8, 2016

Wichita State baseball is eleven games into their 2016 season and things have not started off well. It is very early and WSU can still turn it around, but that process needs to start now. The schedule from here on out is going to get tough to say the least including one-offs with rivals Oklahoma and Oklahoma State as well as a three game series with Cal. State Fullerton and Nebraska.

By this point we have a decent idea as to what type of performance the team is receiving from the everyday starters and even some of the most-used replacements. I scraped play-by-play data as well as team statistics from the NCAA website in order to take a closer look as to what has gone wrong thus far and to see if there is anything that can be done to correct this trend.

A good starting point is to look at the team's runs scored and runs allowed. We can use these numbers to find the teams run differential. This metric is important because it a fairly solid predictor of a teams' win-loss record. The thought process is if a team scores lots of runs and allows very few runs they will win a majority of their games. As I am writing this - through the first eleven games - Wichita State has scored 78 runs and allowed 77 runs. Intuitively one would come to the conclusion that WSU should be a .500 team and that would be correct. Bill James - famed sabermetrician and a consultant to the Boston Red Sox - came up with a neat way of using a teams run scored and runs allowed to predict a teams winning percentage and the math is usually relatively accurate. His formula is very similar to the Pythagorean theorem and looks like this:

Winning percentage = Runs Scored ^ 2 / (Runs Scored ^ 2 + Runs Allowed ^ 2)

Squaring the variables provides a close estimate for Major League Baseball but to get a more accurate figure the exponent needs to be calculated for the college run environment. I will spare you the gory math and just say that I did this using the the NCAA play-by-play data and came up with an exponent of 1.8. Using this exponent in Bill James' formula I came up with a predicted win total - for WSU - of 5 to 6 games for this season through 11 games.

There is an easy explanation for why WSU has under-performed their Pythagorean estimate. The run differential of games they won is much higher than the ones they lost. Their four wins came by an average margin of 7.25 runs. Of the games they lost the average run differential was 3.5. So what does this mean? Basically this means that WSU can score runs, but they also allow a ton of runs. They need to find a way to keep games closer. The 3.5 run differential in their losses is not great, but it is better than last year. In 2015, the team had a 4.2 run differential in their losses. Much like this year, last year the team under-performed their predictive win total. They scored 342 runs and allowed 337 but came away with a 44% winning percentage when our formula had them pegged at 51%. This year they are on pace to score more runs than last year and last year they scored a lot. To put things in perspective, in the 2012 season the team had a record of 35-25 with an average runs per game of 5.4. Last year their runs per games stat was 5.7 - higher than there winning season in 2012 - and they had a losing record. This season they are scoring runs at a clip of 7.1 runs per game yet they continue to struggle to win.

Logically one might come to the conclusion that pitching is the culprit and that reasoning would be partially correct. There is no doubt that the pitching and defensive side of the game has failed WSU thus far, but is it the pitchers fault or has it been partly due to bad luck?

Take a look at the chart below showing the batting average on balls in play (BABIP) for this year and last year. The chart shows the stats for the pitchers who have the most batters faced this season and who we had stats for from last season.

* Sam Tewes was injured most of 2015 so 2014 stats were also used

For those of you unfamiliar with BABIP, it is a stat that gives us an idea as to how batters have performed against pitchers once they put the ball in play. In a sense, it is isolating only the batted balls. Every pitcher has an average BABIP that they hover around year in year out. If the pitcher has a higher than normal BABIP it could mean they are due for a regression and could start to see an increase in performance. Those experiencing a lower than normal BABIP could be candidates for a decline in performance as they are likely to see more balls in play go for hits against them.

By looking at the chart above we can see that the Shockers have seen an abnormal amount of balls in play go for hits. Is this because they are just giving up harder hit balls or are opponents simply finding holes.

To answer this question we need to look at batted ball metrics and I will use these same four pitchers as an example. Remember these players represent the majority of batters faced for the pitching staff. The next chart compares the batted ball data (ground ball %, fly ball %, and line drive %) from this season to last.

Not good all the way around. It would seem that the high BABIP's are in fact due to an increase in the number of hard hit balls - aka line drives. I will admit this batted ball data is not perfect, but I did pull the information from the NCAA website so I think it gives us reasonable estimate.

I want to take a look at one more stat to see if we can't more accurately pinpoint the problem. This final table shows you the left-on-base percentage (LOB%) for our four pitchers as well as the team from 2016 and 2015.

This could be the problem. Three of the four pitchers have seen a significant decrease in this percentage. This means that they are leaving less runners stranded on base. The hits are coming with runners on base and this spells problems for any pitcher. In the Shocker's case, three of their four most relied on pitchers are allowing poorly timed hits all at the same time of the season. It is no wonder why they are struggling. Can this be corrected? Most definitely! In fact, it has been researched and proven by many sabermetricians that LOB% regresses to the mean more often than not. For are four pitchers this means good times are coming. Let's just hope that when these times come the offense is still firing on all cylinders so that the increase in performance can translate into wins.