The post High-Peak vs. High-Consistency Players and Winning appeared first on Baseball POP.

]]>Some players compile strong stats over the long season because they are *extremely *hot during certain periods, and colder in others. Over a large enough sample size, they average out to a strong season-long performance. We’ll call these `High-Peak` players.

Then, there are players who never make headlines for their hot streaks, and also never slump. They can deliver the same volume of season-long production as High-Peak players, but they do so in a smoother way. These are `High-Consistency`players.

I posit that of the two players in our hypothetical situation above, the more consistent (High-Consistency) player will be worth more to his team than the streakier (High-Peak) player.

To test this hypothesis, I did a quick analysis of team production and winning. I’m not trying to investigate *total* production, just consistency in delivering however much production a hitter delivers. Thus, I used the standard deviation of runs scored by game as a proxy for High-Peak vs. High-Consistency offense.

Also since I’m not concerned with *total* offense, I’m not concerned with *total* wins either. Instead, I’m looking at the difference between a team’s actual win % and their Pythagorean win %. Since I am investigating the difference between High-Peak and High-Consistency players in generating the same amount of total offense, I believe these differences will explain the difference between actual win % and the projected win % derived from total runs scored using the Pythagorean method. In other words, High-Peak teams / players will more consistently under-perform against their Pythagorean win %, while High-Consistency teams / players will perform truer to their Pythagorean win %.

The chart below shows the results from 2017 and 2018. Each dot is a team and season, with that teams standard deviation of runs scored by game on the X-axis, and the difference between the actual win % and their Pythagorean win % (relative performance) on the Y-axis. The results show a strong correlation (r-squared = 0.22) between consistency of scoring and relative performance, with High-Consistency teams performing better than High-Peak teams compared to expected Pythagorean wins.

Logically, this makes sense. The team that scores 800 runs in a season by putting up 5 runs every game (assuming a decent pitching staff), should win more often than a team that scores 10 runs half of their games and 0 runs the other half. Consistency of production matters.

But how much? Can we put a value on a more consistent player? Yes, we can. Using the simple linear regression above, we can determine at the team level that each incremental run-scored in standard deviation results in a 4% drop in win %. This is equal to 6.5 wins over the course of 162 games. In other words, had the 2018 National (3.82 Runs Scored SD) had the consistency of the 2018 Tigers (2.84 Runs Scored SD), then they would’ve won 6 or 7 additional games with the same total production.

Why is this so? I think it relates to the investment strategy of portfolio diversification, which I will explore in future posts, as well as the practical strategy for applying this to baseball.

The post High-Peak vs. High-Consistency Players and Winning appeared first on Baseball POP.

]]>The post Judge and Altuve: A Tale of 2 Strike Zone Oddities appeared first on Baseball POP.

]]>Here’s a line of reasoning using the transitive property about short hitters and small strike zones:

- If a player is shorter, then his shoulders are closer to his knees.
- If his shoulders are closer to his knees, his strike zone should be smaller.
- If his strike zone is smaller, then it should be harder for the pitcher to throw strikes.
- If it is harder to throw strikes, he should get on-base more.
- So, shorter players should have higher on-base percentages.

But this isn’t how it actually works. Why not

Aaron Judge is tall and Jose Altuve is short. That’s analysis. They both are very productive at the plate. That’s deeper analysis. However, Aaron Judge gives pitchers a 25% larger strike zone to target due to his 6′ 7″ height, so he must be doing something different *and better *than Altuve to offset the larger area he gifts to pitchers with every pitch.

Have you ever asked yourself how weird the Judge/Altuve co-existence is? To me, it’s like quantum mechanics and general relativity – we don’t understand how these two things can be true at the same time, but they do. Aaron Judge is general relativity and the tallest every-day player in MLB and Jose Altuve is quantum mechanics and the shortest. They are opposites in stature, yet both are legitimate MVP candidates (or past winners in Altuve’s case). This is not common.

It’s not comparable to the NBA, where you can have a 6′ 11″ Giannis Antetokounmpo and 6’0″ Allen Iverson both as HOF players – they affect the games in different ways that accommodate their heights. Giannis uses his size like Thor’s hammer to dominate opponents, while Iverson played like a torpedo fired down the exhaust port of a Death Star. It *is* comparable to identifying the best Left Tackle in the NFL. Question – who am I describing when I `At 6' 5" and 330 lbs., he has the balance of a ballerina, reactions of a mongoose, and power of a silverback gorilla`

I looked at the distribution of strike zone heights and mid-points across all hitters who saw at least 400 pitches in 2018, and compared that to their season wOBA. The results below show a huge distribution in heights, but no strong correlation with wOBA. This is counter-intuitive and also shows how astonishing it is that Judge and Altuve are on extremes in both dimensions.

How uncommon are Judge and Altuve? They are both members of what we call `2 & 1 Club`

`Strike Zone Height`

(though obviously, they deviate in opposite directions), and 1 standard deviation above league average wOBA. In fact – This is so rare, they are the *only* members.

The odds of a player being >= 2 standard deviations away from the strike zone size is about 5%, and being 1 standard deviation above the wOBA average is about 17%. If we assume these are independent variables, then the probability of a single player being both of these groups is less than 1%. I suspect that is even over-stating the probability because tall players can fake-short and create small strike zones, but short players can’t do the opposite. This would be Judge an even rarer bird due to his tall zone.

Altuve is unique in this respect, but Judge is the true marvel to behold. First, he’s not only the only player in `2 & 1 Club`

`2 & 1 Large`

– he’s the only player in the `1 & 1 Large Club`

(1 standard deviation above the mean in both strike zone height and wOBA)! And it’s not even close. Amongst players who are at least 1 standard deviation over the average, Judge is still over .050 wOBA points ahead of the next-best hitters (Cody Bellinger and Jose Ramirez). Altuve has players who are at least almost-peers (Adam Eaton and the surprisingly-squatty Bryce Harper), while Judge stands tall and alone.

This begs the question – how does Judge do it? How does he offset the enormous liability of a strike zone 25% bigger than average well enough to be one of the best hitters in the league? I believe it has something to do with how he capitalizes on pitches in the zone while smaller hitters gain an advantage by avoiding pitches outside of the zone. That is Judge smashes hittable pitches while Altuve is *judicious* about what he swings at.

The post Judge and Altuve: A Tale of 2 Strike Zone Oddities appeared first on Baseball POP.

]]>The post The Best and Worst by Ball-Strike Count appeared first on Baseball POP.

]]>Using our `Expected wOBA`

by at-bat, we tabulated the results for both batters and pitchers across every count for the 2018 season. In other words, for all at-bats in which Mike Trout ever saw a 0 balls, 1 strike count, his wOBA was `0.518`

for those at-bats, which was the 5th best for all players after going 0-1. To reduce noise and outliers, we only included players in each count with at least 20 at-bats going through that count. So if Craig Kimbrel only went 0-3 six or seven times in 2018, then he wouldn’t show up in this analysis.

We analyzed pitchers and batters in the same way, but separated their results in the interactive tools below.

**Mookie Betts is comparatively very good,****except****for 3-0 counts.**His wOBA in these situations is a mere 0.456, compared to highs reaching almost 0.800.**The secret to getting Mike Trout out is starting with an 0-2 count**. He is very good, except only`above average`

in 2-strike counts.**Albert Pujols can barely get on-base even if you spot him 3 balls.**Let me put it this way – Mookie Betts has a 0.100 wOBA*advantage*at the start of his at-bats over Albert Pujols’ at a 3-0 count.**Joey Gallo is very, very**If he gets ahead in the count, he’s a very good bet to get on-base. If he gets behind in strikes, then he’s heading towards an out.`Three-True-Outcomes-y`

.**Howie Kendrick doesn’t waste 2- and 3-ball counts.**In these situations, he’s one of the best. With a 0.752 wOBA in full counts, he’s almost a lock to get on-base.

**The Top 5 positions are dominated by relievers.**This could be because they have lower sample sizes, which helps them resist reversion to the mean; as well as specialization in their repertoire which is more repeatable in specific situations.**The Bottom 5 positions are also dominated by relievers.**I suspect this because most of these guys are flash-in-the-pans and they fall out utilization and/or get demoted/released before they have a chance to improve their ratings. Starters typically have more opportunities to normalize their results (a few starts for a starter vs. a few innings for a reliever).**Closers close.**They jump ahead in the count and they don’t relent. Aroldis Chapman and Wade Davis were among the best in two-strike counts, and Craig Kimbrel didn’t even qualify for the leaderboards in most two and three-ball counts.**David Price is one of the best after being really bad.**In three-ball counts, he is consistentlybest or one of the best in keeping hitters off base. I don’t play, but is anything in this situation analogous to Fortnite? That might explain it.*the***Jacob Degrom is surprisingly bad in 3-0 counts.**He’s great everywhere else, but allowed a 0.674 wOBA after reaching a three-balls, no-strikes count.

The post The Best and Worst by Ball-Strike Count appeared first on Baseball POP.

]]>The post Predictive Value of Pitch Combinations vs. Single Pitches appeared first on Baseball POP.

]]>Our hypothesis is that ** pitches don’t happen in vacuums** and the pitches a hitter has seen historically affect their ability to hit pitches in the present. However, we need to test this hypothesis to see if the data

If that hypothesis were true, then we should see higher predictive value in pitch combinations or sequences of pitches than in single, isolated pitch evaluations. That is, the outcomes of pitch combos should be more consistent than the outcomes of single pitches.

To test this, I looked at the standard deviations for wOBA Added for both discrete`pitch types`

by `pitch combos`

`pitch type`

`pitch combo`

For each of these data sets, I was focused on `standard deviation`

`wOBA Added`

outcomes observed. A larger standard deviation would indicate lower predictive value of the input variable (pitch type and pitch combo), and a smaller standard deviation of outcomes would indicate more predictive value.

LargerStandard Deviation of wOBA Added Outcomes =LessPredictive Value

SmallerStandard Deviation of wOBA Added Outcomes =MorePredictive Value

The chart below shows the results from the 2018 season. I filtered to only include pitches (as defined by pitch type and location) or pitch combinations with at least 40 observed samples to try to remove as much noise as possible. I then charted every sample by average wOBA Added with reference bands showing the Standard Deviation for wOBA Added.

The results are subtle but telling. As we look at pitch types and pitch combos with a minimum sample size of 30 observations to eliminate noise from the analysis, we see that **standard deviation of wOBA Added outcomes can be 10-33% lower for pitch combos compared to single pitches**. Let’s dive in.

If we set the minimum sample size to 40, then the typical Standard Deviation size for wOBA Added on Single Pitches is 0.300. However, the same measure for Pitch Combos is 0.270. This 0.030 represents a 10% improvement in predictive value for the effectiveness of pitch selection.

I even think this 10% improvement is underestimating the potential advantage of Pitch Combos over Single Pitches. As we increase the minimum sample size threshold, we see the Pitch Combo advantage increase, getting as high as a `33% advantage`

at `n = 50`. This analysis DOES come with a serious qualification as there are only 3 Pitch Combos that have at least 50 observations, and they are all Justin Verlander Four Seam Fastball / Four Seam Fastball combos. However, the trend holds and intuitively it makes sense.

Any analysis or pitch decision that only looks at a Single Pitches effectiveness is only looking at half of the picture; not all Four Seam Fastballs low and away are created equal. If we want to look at effective pitch selection, we have to look at the compounding effects of sequencing to get better predictions about what each incremental pitch will add to the pitcher’s chance of success.

The post Predictive Value of Pitch Combinations vs. Single Pitches appeared first on Baseball POP.

]]>The post Repertoire and Results: The Leading Indicator for Good Pitching appeared first on Baseball POP.

]]>Anecdotally, we saw a correlation between ‘effective repertoire usage’ and ‘good pitchers’. I wanted to quantify that relationship better to see if there really was a statistical relationship between pitchers who use their repertoire effectively and objectively ‘good’ pitchers.

To do this, I calculated the actual wOBA allowed for each pitcher and the `Combo Repertoire wOBA Added`

from the 2018 seasons. My hypothesis was that there should be a positive correlation between these two measures of effectiveness because the `Combo Repertoire wOBA Added`

would be an input or leading indicator of overall wOBA Allowed.

The chart showing these results is below. To try and eliminate as much noise as possible from low-innings pitchers, I filtered to only pitchers with combo sample sizes >= 300. The size of the dots also represents sample size (larger = higher *n*). The X-axis `Combo Repertoire wOBA Added`

`2018 wOBA Allowed`

`wOBA Allowed`

.

The resulting `r-squared`

`Combo Repertoire wOBA Added`

`2018 wOBA Allowed`

Effective pitchers use their pitch combo repertoire effectively. Pitchers who are good at getting hitters out do so because they use the 1-2 pitch combos that give them the most success more often; and they avoid the combos that hitters hit well. Less effective pitchers either lack any effective pitch combo to go to, or they are underuse them.

The post Repertoire and Results: The Leading Indicator for Good Pitching appeared first on Baseball POP.

]]>The post Most Effective Pitch Repertoires appeared first on Baseball POP.

]]>In our last post, we looked which pitch combos are effective and being *used* effectively. This study showed us that there are definitely 1-2 pitch combos that are creating the most value for pitchers, and some that are letting hitters eat their probabilistic lunch.

So, given that we know:

- Some pitch combos are far more effective than others, and
- Pitchers use ineffective pitch combos far more than they should,

The next natural question is: **who is using their entire repertoire of pitch combos the best?**

This is best answered by looking at each pitcher’s observed pitch combos and packaging them as a portfolio of combos, with each combo weighted by how much it is used relative to others. Thus, the weighted average output of this calculation gives us a pitcher’s Combo wOBA Added across their entire portfolio of pitch combos. The better the pitcher’s combos, and the *more often* he uses those combos, then the lower the `Combo Repertoire wOBA Added`

The chart below displays where individual pitchers sit on this measure of effectiveness. The X-Axis is `Combo Repertoire wOBA Added`

In looking at the insights from the tool above, we can learn a few things:

The greater the sample size of combos observed in our sample, which has a direct relationship with pitches thrown, the more likely pitchers are to be closer to 0.000 on their Repertoire wOBA Added (see the blue trend line in the graphs).

I’ve cherry-picked a few examples below of pitchers who have negative `Combo Repertoire wOBA Added`

. Yes, these guys are cherry-picked; no, this is not a robust correlation analysis; and yes, they are *close* to zero. However, the quality of the names demonstrates that there could be a there there; and they are closer to zero because they were workhorses for their teams (see analysis above).

This is an iterative step in understanding the dynamics at play in pitch sequencing. Coaching a pitcher to “have a better Combo Repertoire wOBA Added” isn’t helpful. However, understanding that this is a critical determinant in pitcher success *is* helpful so that we can seek to understand the deeper dynamics better and use those in the pitch sequencing decisions.

So, more to come.

The post Most Effective Pitch Repertoires appeared first on Baseball POP.

]]>The post The Pitch Combinations Destroying Hitters appeared first on Baseball POP.

]]>We discussed in a prior post the most and least potent pitch combos from the 2018 season. It’s one thing for a pitcher to have a potent 1-2 pitch combination in their repertoire, but it’s of no value if the pitcher doesn’t *use* that often enough in their pitch sequencing. In other words, pitchers only benefit from these pitch combinations if they are using the ones that fool hitters the most; and they are actively reducing their effectiveness if the continue going back to the pitch combinations that hitters capitalize on.

We put pitch combos into four generalized buckets:

**Pitcher Work Horses:**Combos that have a negative wOBA Added (reducing the hitter’s prospects of success) AND get used more than the median usage rate for all qualifying combos, which is about a 6% usage rate. These are the combos that pitchers should and*do*use the most often.**Pitcher Secret Weapons:**Similar to the Work Horses, these are combos that have a negative wOBAAdded, but are relatively*underused*relative to other Combos. It’s unclear if their effectiveness comes from this low usage rate, and if they would lose their effectiveness if pitchers have the opportunity to see them more. Currently, they function as surprise combos that hitters shouldn’t be expecting.**Hitter Feasts:**Pitch combos that have a positive wOBA Added (benefiting the hitters) and get used more than the median. These combos are over-used by pitchers, destroy value for the defense, should be anticipated by the hitter, and should be capitalized on by the hitter when seen.**Hitter**Combos that are seldom seen but benefit the hitter when they arrive. These are typically easy-to-adjust to combos that hitters probably don’t need to actively look for to be able to handle well.Treats:

Simply put, **Pitcher Work Horses** are the 1-2 Combos that pitchers should-and-are using the most, while **Hitter Feasts** are the Combos that are ill-advised for defenses.

As an example, let’s look at Chris Sale’s repertoire of Pitch Combos in 2018 against these four categories. In the scatter plot below, we map each 1-2 Combo against the `wOBA Added`

`Usage Rate`

on the Y-axis.

It shouldn’t be a surprise that a premier pitcher like Sale is good at using his most effective Combos the most. Inferior pitchers have many more dots gravitating towards the upper-right.

To look at one specific example, Sale’s `Four-Seam Fastball, Four-Seam Fastball`

sequence that moves 1 zone more inside to the hitter is a potent combination that had a wOBA Added of `-0.196`

in 2018. This was a dangerous sequence for hitters and a valuable one for the Red Sox. What’s more, Sale used this combo 10% of the time, which is a relatively high usage rate relative to all other possible combos. This is a perfect example of **Pitcher Work Horse**.

Investigate which pitch combos are being used effectively yourself with the tool below. Filter on individual pitchers and hover over each dot to see detail on that specific Pitch Combo.

The post The Pitch Combinations Destroying Hitters appeared first on Baseball POP.

]]>The post Most- and Least- Potent Pitch Combos appeared first on Baseball POP.

]]>The approach I took was to evaluate every pitch as the second pitch in a 1-2 combo (forcing us to exclude at-bat first pitches). I defined these pitch combos using the pitcher, the pitch types of both the 1st and 2nd pitches (e.g. ‘Four-Seam Fastball followed by a Curveball’), and the pitch location change from the 1st to the 2nd pitch (e.g. ‘The 2nd pitch was further down and more inside than the first pitch’). I then gauged the effectiveness or value of these pitch combinations using the sum of the

The chart showing every pitch combo is below:

So how can we use this Pitch Combo Effectiveness Tool and what can we learn? The depth of detail is great and patterns are surprisingly strong and insightful. For each pitch combo, we also report `average pitch speed change`

and `difference in breaking action`

between the 1st and 2nd pitches. We also measure the `Usage Rate`

of that combo for the pitcher.

The results show clear patterns and trends in what is most and least effective for individual pitchers. As an example, we’ve captured `Max Scherzer`

`Four-Seam Fastball, Four-Seam Fastball`

`more inside`

`further up`

`further outside`

`lower in the zone`

.

Visit the Pitch Combo Effectiveness Tool permanent page, or do your own investigation and exploration in the embedded version below.

The post Most- and Least- Potent Pitch Combos appeared first on Baseball POP.

]]>The post Pitch Zone Change Effectiveness appeared first on Baseball POP.

]]>If the pitcher’s objective is to deceive the hitter, then, of course, they must have good pitches that are hard to identify by the hitter. This means they move in pronounced or unanticipated ways and they look similar enough to other pitches in the repertoire that they are hard to distinguish.

However, I believe that the pitches

In this study, I’ll examine the effects of location change on Pitcher success. Location is measured on the X (inside/outside) and Z (high/low) axes, so we will look into how changes from pitch-to-pitch along these axes affects Pitcher success.

To gauge Pitch Success, we will use a simple proxy for hitter confusion & deception: swing-and-miss (aka whiff) rates. For all possible directions and magnitudes of location change, we’ll measure the relative frequency of swings, contact, and whiffs.

The interactive heatmap below shows the swing, contact, and whiff rates *after* a pitcher changes locations from the immediately preceding pitch.

- The X-Axis represents the change of position relative to the hitter across the plate, with positive numbers indicating movement
*away*from the hitter / outside, and negative numbers indicating moving*towards*thehitter / inside . For scale, the strike zone is 3 zones wide. - The Z-Axis represents the change of pitch height when it crosses the plate. For scale, the strike zone is 3 zones high.

The most immediate take-aways from the heatmap above are:

- Staying in the same place will induce the most swings
- Staying in the same place will also induce higher contact rates
- Moving
*down*in location will induce higher whiff rates

Depending on the pitcher’s at-bat level strategy for a hitter (e.g. pitch around, avoid contact, induce weak contact), they can construct 1-2 pitch combinations that will increase their chances of success.

- If a pitcher is trying to
**avoid contact**, then constructing pitch sequences that start high and progressively**step-down in the zone**should create more whiffs as the hitter is less capable of adjusting their swing down. - If a pitcher is trying to induce
**weak contact**, keep pitches around the same location to induce more swings and contact, but use other tactics to diminish the quality of contact.

The post Pitch Zone Change Effectiveness appeared first on Baseball POP.

]]>The post wOBA Added – A New Metric for Measuring Single Pitch Effectiveness appeared first on Baseball POP.

]]>Pitches that terminate an at-bat are relatively easy to assign a value to – did at-bat end in an out or hit. However, in 2018, almost 500k pitches (or 75%) of pitches thrown in Major League Baseball game didn’t end an at-bat. So how are to determine which pitch is good and which isn’t? Is moving from no strikes to one strike just as valuable as moving from two to three? What if there are two balls? three? How does the hitter’s change in approach in these counts change the marginal value of each new strike or ball?

And is there more going on? Are all 1 ball, 2 strike counts created equal? Is the hitter in better or worse shape if they got there after seeing three fastballs; or a fastball, breaking ball, and change-up? My hypothesis is the movement from one count to the next has a unique value to the hitter and pitcher, and the journey to get to those counts matters as much as the count itself.

To begin this test, I must first have an objective measure of success for each marginal pitch, *especially* if they don’t terminate an at-bat. I need something that is objective,

By looking at the occurrences of every possible count and aggregating the wOBA outcome of those at-bats, we see a clear and predictable relationship between pitch count seen during an at-bat and ultimate at-bat outcome. For example, the chart below shows that *all* at-bats that passed through a `1 ball, 2 strike`

count had an average wOBA `.219`

`wOBA Added`

would just be the improvement or decline of the hitter’s prospects as the count advances. A ball that moves the count from 0 balls, 1 strike to 1

- Most pitches have an amplified impact on the eventual outcome later in the count
- Marginal balls help the hitter significantly later in the at-bat than earlier. Going
from `0-0`

to `1-0`

is almost completely inconsequential. However, goingfrom `2`

`-0`

to `3-0`

is enormous. - Strikes are slightly less important later in the count, and oftentimes are more important early in the count

So what does an example at-bat look like as the count advances and wOBA added gets tabulated? Here is a set of examples from a 2018 game:

**In Sum: **pitch count matters greatly to the pitcher and hitter’s chances of eventual success. Also, **strikes matter more early** and **balls matter more late**.

The post wOBA Added – A New Metric for Measuring Single Pitch Effectiveness appeared first on Baseball POP.

]]>