Youth Baseball Stats Part 3: Mustang Kid Pitch Begins

The stats that matter most for youth baseball change dramatically as kids get older. The biggest change is the transition from coach pitch to kid pitch, which in our league happens at the age of 9.

This post is part of a series that started with Youth Baseball Stats Part 1: It’s Hard on how to gather, interpret and apply baseball data to improve youth baseball play.

Part 2 discusses stats appropriate for coach pitch.

This post discusses the stats that matter when kid pitch begins.

Kid Pitch

Kids start to pitch at ages 9 and 10 in the Mustang division of our PONY baseball league. There are no leads, balks, or dropped strike 3s, but stealing is permitted when the ball crosses the plate. There are also some mercy rules if one team leads by 10 or more runs but the mercy rules are rarely needed. It’s pretty much real baseball, albeit on a smaller field with smaller players at various stages of learning baseball skills.

Learning to pitch is hard. Many pitchers have trouble throwing strikes.

Catching wild pitchers and throwing out base stealers is hard.

Hitting is hard against strike-throwing pitchers and even harder against wild pitchers. Unlike coach pitch, the pitcher isn’t trying to make it easy for the hitter. If your umpires are older kids, inconsistent strike zones makes hitting even more challenging.

What I found, statistically, was that pitching matters most. Hitting and catching also matter. Defense for balls hit into play, base stealing, and various other baseball nuances also matter, but not as much. I haven’t done much yet with tracking catcher performance (passed balls? Throwing out stealers?). But tracking pitcher performance turns out to be very easy, while tracking hitter performance is not that much different than tracking hitting in the major leagues, with complications arising from low sample sizes and high error rates.

Baseball Scorekeeping Software

There is great scorekeeping software available for phones these days, namely Gamechanger and iScore. If your scorekeepers use either of these to track all your games, you’ll get stats compiled automatically. For some missing stats, you can download the data to a spreadsheet and calculate the stats for yourself. Once you’re regularly recording games with one of these two, you can begin to interpret the stats and take action.

As I was writing this article, I came to realize that iScore provides much richer hitting stats than Gamechanger. Given the small number of games and plate appearances in a season, small sample size is a significant issue, as I explained in part 1. This makes per pitch stats valuable.

iScore has several per pitch stats for both pitching and hitting, while Gamechanger only has per pitch stats for pitching. iScore also gives you OBP + ROE automatically, while Gamechanger requires exporting its data to a spreadsheet and calculating it yourself. iScore requires $10 to get started. It’s possible to use Gamechanger without cost, though it can actually cost more in total if several parents want to look at stats and follow games in real time.

For more detail, read my Gamechanger review.

Pitching Stats

For the beginning pitcher, it’s all about throwing strikes. Take, for example, the rec team I managed last year. WHIP is Walks and Hits per Inning Pitched—a good overall measure of pitcher performance that is less dependent on fielding performance than the better known ERA stat (Earned Run Average). Among pitchers who pitched at least 3 innings, the pitcher on my team with the highest strike percentage had the lowest WHIP. The pitcher on my team with the second highest strike percentage had the second lowest WHIP. And so on.

The main reason WHIP increases as strike % decreases is walks. Walks give the batter a 100% chance of reaching first base.

Walks are the single biggest driver of which team wins or loses at the 9-10 year old recreation level. The vast majority of the time, the team that issues the most walks loses, while the team that issues the fewest walks wins. This was true 15 of 19 games in my son’s first season of kid pitch, and was true in all 20 games of my son’s second kid pitch season. (Two of those 20 games had exactly the same number of walks for both teams. Those games had close scores.)

Another important performance measure that benefits from throwing many strikes is pitch count. The higher the pitch count per inning, the longer and more tedious the game, and the sooner the pitcher must be taken out. There is even the potential to run out of pitchers if a manager does not appropriately plan around the league’s pitch count limits. Fewer strikes almost always means higher pitch count.

A high strike percentage also allows a pitcher to further develop the other tools of pitching, which I cover in Developing a Youth Pitcher.

NOTE: Several months after writing this post I realized that WHIP and pitch count are strongly influenced by fielding at the little league level. A player whose team commits many errors may have to pitch to twice as many players per inning, resulting in twice as many hits and twice as many walks per inning if the walk and hit rates stay constant. Nonetheless, I still think WHIP and pitch count are useful. If the pitcher happens to be a great fielder, than WHIP is a joint measure of pitcher effectiveness along with how much it harms the team to take him out of a critical fielding position.

I like my pitchers to have low WHIP and lots of strikeouts but it takes many innings of pitching before the strikeout rate stabilizes, and even more for WHIP to stabilize. Sample sizes are low in youth baseball so any stat related to each pitch is great if it correlates to all the other stats you like to see in a pitcher. Strike % is that stat in 9-10 year-old rec league.

I’ve noticed that many managers get excited by high velocity pitchers. The hardest throwing pitcher is often given the start. Though I haven’t studied it formally, I would guess that there’s some positive correlation at this age with high velocity and high strike %. On average, better pitching mechanics means more velocity and increased accuracy, for a given height. However, high velocity can be achieved without good mechanics, typically from bigger players. Therefore, velocity is not perfectly correlated with strike %. Not even close. Nor is velocity strongly correlated with WHIP and other pitcher performance measures.

I couldn’t even pitch our hardest throwing guy because he rarely threw a strike and I was afraid he’d hurt a batter. Our second hardest throwing guy showed promise and was hard to hit. However, thanks to 37% strikes, he issued so many walks that he had a WHIP above 4 and the highest ERA on our team. Our third hardest throwing guy threw 54% strikes. He had our second best WHIP, 1.97. The guy who threw 61% strikes was a small guy who threw hard for his size, but not as hard as the bigger guys. He also had a good changeup he could throw for strikes. He had a team-leading WHIP of 1.25. I believe the slowest throwing guy on our team threw below 30MPH, but he threw over 45% strikes, which was enough to place him fourth on our team for both strike % and WHIP.

I don’t know if this holds true all across the nation, but in our 9-10 year old rec league, what mattered was throwing strikes. Velocity . . . not so much.

Strike % is not as great a predictor of success at the all-star level, because hitters are better. If all a player can do is throw 40MPH to 45MPH strikes right down the middle, most hitters will put the ball in play. So velocity, control, and secondary pitches (changeup) become much more important.

Velocity alone can shut down batters if it’s several standard deviations above the norm and can be thrown for over 50% strikes. I believe an all-star pitcher at this age can rely entirely on velocity if it’s at least 50MPH for 9 year-olds and 52MPH for 10 year-olds (These numbers are based on a radar gun one fellow used to measure several high velocity pitchers during a playoff game, so they can only be considered approximate, and may be a little low, based on pitch speeds listed by John Madden). For the majority of all-star players who can’t throw so hard, my sense is that changing speed and location are important. Unfortunately, I can’t back that up with data as we didn’t track pitch type or location.

Velocity does matter at the rec level, just not as much. If the strike % is the same between two pitchers but one throws 50MPH while the other throws 40MPH, the higher velocity pitcher will strike out more batters. More strikeouts means fewer balls hit into play. The high velocity pitcher can get away with issuing more walks because there will be few hits to clear the bases. In the example above, the team’s pitcher who threw 61% strikes with a 1.25 WHIP gave up about the same amount of runs per inning as the pitcher who threw 54% strikes with a 1.97 WHIP. They both achieved good pitching results but the guy with the higher WHIP threw harder, just shy of 50MPH, and could therefore get away with more walks.

The thing is, some high velocity pitchers have low strike percentages and issue a lot of walks. Those who throw both hard and accurately are dominant. But give me a choice between a 52% strike throwing 32MPH pitcher and a 32% strike throwing 52MPH pitcher and I’ll take the 52% strike thrower every time.

Hitting Stats

Hitting is tougher to measure, given the small sample size. I favor stats that are per pitch because pitches occur over four times more frequently than plate appearances, and about 6 to 7 times more frequently than at bats. Unfortunately, Gamechanger reports only one pitch related stat—the number of pitches per plate appearance. iScore reports several and is therefore a better option for getting meaningful stats at the youth baseball level.

My son’s teams have all used Gamechanger. I incorrectly assumed that contact % meant the percentage of times a hitter contacts pitches that are thrown for strikes. Contact % is actually defined as (AB – K)/AB where AB is At Bats and K is strikeouts. Being based on at bats means this stat is sampled about 1/7 as frequently as I was assuming.

Given how heavily I rely on contact %, I’m going to assume that it approximates the stat I really want, which is the percentage of times a hitter contacts a pitch that is thrown for a strike, either swinging or called. I know it opens up this post to criticism, but perhaps it can also serve to motivate Gamechanger to offer per pitch stats that are so important for youth baseball.

Taken alone, contact % can mean a few different things:

  • It may mean the obvious—that the higher the contact %, the higher the batting average and OBP (On Base Percentage).
  • A high contact % may mean the player swings at almost everything and can get the bat to contact the ball—but often for poor hits that result in outs due to swinging at bad pitches. If the walk rate is far below average, then that is the likely explanation. A player with this profile needs to stop swinging at bad pitches.
  • A low contact % may be due to taking a lot of strikes looking. We had one batter on our team who had a low contact % but the third highest OBP+ROE on the team. This was because of his Ted Williams-like approach of only swinging at the pitches that were in exactly his favorite spots. If there were a batting average measure which divided hits by number of swings, he would have easily led our team. However, this was unusual. Most players with a low contact % take too many strikes and need to swing more, usually because they’re afraid of the ball.

So unlike the handy strike % measure for pitchers, contact % by itself is not a particularly good predictor of overall batting success. In order to use contact % meaningfully, it must be combined with other measures, such as Ks and BBs (strikeouts and walks). Fortunately, as I mentioned in a previous article, Ks and BBs stabilize more quickly than most other plate appearance stats. I believe that contact %, Ks, and BBs combined with visual observation of hitting mechanics, how hard a player hits the ball, and how fast he hustles to first, can be combined to form some conclusions of how effective a batter will be at getting to first base, which is the end result that matters most.

My favorite overall measure of hitting performance is OBP + ROE (On Base Percentage + Reach On Error). This measures the percentage of the time that a player reaches first base by any means. At the major league level, OBP + ROE highly correlates with scoring runs, about the same as OBP. The reason OBP is used to track major league hitters but not OBP + ROE is that ROEs are random and rare due to great fielding, for the most part. In youth baseball, it’s quite obvious from watching that players who hit the ball very hard and/or hustle get more ROEs, sometimes many more. Since hitters can impact ROE so strongly at this age level, OBP + ROE becomes the ultimate hitting productivity measure.

As an extra bonus, OBP + ROE eliminates the issue of inconsistent judgment regarding errors. The number of times a player reaches first base is strictly objective. A first-time scorekeeper new to baseball will understandably have trouble scoring an error versus a hit, but will always mark an out when a player does not reach first base.

Looking at stats from several teams throughout the season, I came to conclude that in our 9-10 year old Mustang division, OBP + ROE for a given baseball player means:

  •   > .600:        great, will score 1-3 runs per game
  •   .550 – .599: very good, will score at least 1 run per game
  •   .500 – .549: good, will often score a run
  •   .450 – .499: will score a run in more than half of games
  •   .400 – .449: will score about 1 run every 2 games
  •   < .400:        will score less than 1 run every 2 games

These are much higher than major league numbers, because at this age, each team typically gets around 10 walks, 3 ROEs, and 10 runs per 7 inning game.

Beginning pitchers issue many walks. Most teams in our 9-10 year old division have 2-4 pitchers who consistently throw over 50% strikes, but pitch counts are limited to protect pitcher arms (a very good thing, as I explain in Pitcher Arm Care). A typical 7-inning game will feature 3 to 7 pitchers and a typical 2 game weekend will need 4-8 pitchers. Some of these pitchers will have difficulty throwing strikes and therefore will frequently walk and sometimes hit batters.

Our team last year had a team OBP + ROE of .482 (in the major leagues, league-wide OBP + ROE is typically .325 in recent years). The OBP + ROE breakdown for our team:

  •   .278 from walks and HBP (hit by pitch)
  •   .142 from hits (batting average was .198)
  •   .059 from ROE
  •   .003 from sacrifice flies

The contribution from hits was below the league average, and considerably below the MLB average. But the contribution from walks and ROEs was very high, and is what caused our OBP + ROE to be so much higher than MLB numbers. Approximately once every 3 plate appearances, a player on this team reached base on a walk, HBP, or error.

I wish I had data from other leagues or national data to benchmark against. I have no idea how typical our rec league numbers are as compared with other leagues. In a league where each team had at least 6 pitchers per team that could throw 50% or more strikes, there would be far fewer walks and therefore OBP + ROE would be much lower, perhaps approaching major league levels, given that many players struggle to hit kid pitch. Conversely, leagues without any pitchers who could throw over 50% strikes would have much higher league-wide OBP + ROE.

I also have data from our summer all-star team. The summer team had more than 6 pitchers who could throw over 50% strikes, so there were far fewer walks. Excluding games with mismatched opponents that resulted in blowouts, OBP + ROE was around .410, about half-way between major league’s .325 and rec league’s .485.

Getting extra base hits is helpful, especially with one or more runners on base. There is a measure of this called slugging percentage and an overall hitting effectiveness measure that combines OBP with slugging: OPS (On-base Plus Slugging). Given its inclusion of the slugging component, OPS is considered by some to be the ultimate measure of hitter productivity, though OBP is usually used as well. I use OPS + ROE instead of OPS for the same reasons described above for OBP + ROE.

There was one rec league player last year who had a high strikeout rate and the lowest OBP on his team, but once powered his team to victory with 6 RBI’s thanks to two extra base hits. Does this imply OPS + ROE make more sense than OBP + ROE? No.

While a 6-RBI game is certainly memorable, for this age group high OBP + ROE will contribute more runs on average than low OBP + ROE combined with high slugging percentage. The reason is that steals are so easy. An average or faster than average base runner reaching first will reach second on the first pitch to the next batter. The better base runners will reach third base on the second pitch. The really daring runners will reach home shortly thereafter, even if the ball is not put into play.

In other words, a double while the bases are empty is hardly advantageous over a bases-empty single, walk, or ROE. Doubles when the bases are empty do matter more for slow or poor base runners—but most slow runners also never hit doubles.

There were several catchers in the league who could reliably throw out all but the fastest runners at third but only occasionally did even the best catcher ever throw out a runner at second. Some teams didn’t have a single catcher who could reliably throw out a runner stealing third. Partly it was because catchers are still learning how to make that throw (and infielders learning how to field the throw). But mostly it was because runners don’t take long to run the short distance between bases. There is little time for the catcher to make the throw.

There are times when extra base hits are far more valuable than walks. That is when there are two or three base runners on with two outs. A double followed by an out will likely produce 2 runs more than a walk followed by an out. So yes—slugging and OPS + ROE do have some significance. You’ll want the player with the highest OPS + ROE to occupy the 3, 4, or 5 spot in the lineup, after players who frequently reach first base (high OBP + ROE). And of course, when two players’ OBP + ROE is identical, the one with higher OPS + ROE will contribute more offensively, on average. So while OBP + ROE is my favorite metric for this age group, OPS + ROE is my second favorite.

This isn’t much different from major league thinking, which also tends to favor high OBP players in the first 2 or 3 slots of the lineup, while favoring higher OPS players in the 3, 4, and/or 5 slots. It’s just that when average OBP is just shy of .500, then OBP becomes more important relative to OPS as compared with the majors. I expect that as field sizes get bigger and players develop more skill, OBP’s will decline and slugging will increase in importance.

So if these two are my favorite measures, why did I start with a lengthy discussion of rates of contact, walks and strikeouts? Again, it is due to sample size.

There are many random factors in baseball, most especially related to where a ball hit into play happens to land. As Russell Carleton has pointed out in two articles, it takes something like 400 plate appearances before OBP (and by extension, OBP + ROE) become stable. Youth baseball players will typically have around 60 or 70 plate appearances over the course of a spring season.

Rates of contact, walks, and strikeouts stabilize much sooner, and are therefore more reliable metrics for a 15-20 game season. The big question is whether they can be used as a shortcut to infer OPS + ROE. I’d like to think that the answer to this question is yes, and that’s what I’m assuming for the moment. But I have too little data to be sure, as you’ll see from the discussion that follows.

I examined contact %, BBs, and K’s at season’s end for all batters to see whether they correlate with OBP + ROE. I wish I had much more data to work with but unfortunately all I have at my disposal are one season of rec league and one season of summer all-stars. Rec league had about 60 plate appearances per player, while the summer all-stars had only 30 – 60 plate appearances per player, as some players missed more than a few games due to summer vacations.

My hypothesis is that as contact % increases, and PA/K increases (same as decreasing strikeout rate), then OBP + ROE increases. This assumes a walk rate that is at or a little below the team average.

The problem with this hypothesis is that I don’t have enough data to test it. For example, take the all-star data. Among our players with 29 or more plate appearances, there were 5 players who had the highest contact %s and highest PA/Ks (lowest strikeout rates). All five had contact rates above 70% and at least 4.5 PA/K.

So how was OBP + ROE for these hitters who frequently contacted the ball and infrequently struck out? 4 out of 5 of these players had the top 4 OBP + ROEs on the team, ranging from a solid .508 (the player who was #5 in both contact % and PA/K) to an astounding .686. But one of these five players had a mediocre .455 OBP + ROE, a little below the team average and seventh best on the team. This, in spite of having the second best contact % and PA/K.

Does this disprove my hypothesis? Not necessarily. That player had 33 plate appearances. Calculation of the standard error places a 95% chance that that player’s true OBP + ROE was somewhere between .285 and .624. It’s possible that with a few hundred more plate appearances (assuming his ability and his opponents’ abilities did not change), that his OBP + ROE would stabilize around .600, which would be more reflective of his frequent contact and low strikeout rates.

But perhaps I’m missing other factors. I studied this right-handed player’s spray chart and saw that mostly he hit grounders to the right half of the infield that were fielded for outs. There was no numerical statistic that captured this data but at least I could visually see that he’s swinging late and hitting the top of the ball. He obviously has a good batting eye, but likely something needs to be worked out with his mechanics.

However, even if this player’s OBP + ROE was higher than .500, my hypothesis would not have been proven. I don’t have enough data, and every player’s true OBP + ROE lies within a very large range. Nevertheless, what data I do have strongly suggests that those who frequently contact the ball and rarely strike out usually have high OBP + ROE, while those who infrequently contact the ball and strikeout often have low OBP + ROE. This is what you’d expect from common sense.

How to use pitching and hitting data?

Pitching data is much easier to interpret than hitting data. Managers will want pitchers with the highest strike % to pitch more. They are essentially forced to in order to get through a game without running out of pitchers. High velocity pitchers will be preferred when strike % is roughly the same, but nearly every player who pitches more than 50% strikes in rec league will be called on to pitch, regardless of velocity.

Unfortunately, there is nothing in traditional scorekeeping data that helps the coaching staff at the rec level understand what pitchers need to work on. Low strike % probably means issues with mechanics, which will need to be observed visually, sometimes with the aid of slow motion video. Diagnosing pitching mechanic issues is fairly involved and won’t be discussed here.

At the all-star level, players with high strike % that are getting hit a lot are probably not varying speed and location enough. This will usually be obvious by visual observation before it becomes obvious with stats.

Hitting data can be used in more interesting ways. How to arrange the batting lineup was already discussed somewhat—by the end of the season it’s really not much different than how you think about it in the major leagues, with a few high OBP + ROE batters followed by a few high OPS + ROE batters.

At the beginning of the season, with pitchers still learning to throw strikes, the lineup might be arranged differently, as players who aren’t swinging much will get a lot of walks even if they’re not very good hitters. Some managers may want to have two four person mini-lineups within the batting order, with the first person in each foursome being a good base runner who gets to first base often, even if it’s from walks.

There are a number of hitting tendencies that can be detected through data, suggesting corrective action. Here’s a few I’m aware of:

  • Player strikes out and walks much higher than team average, and has a low batting average. Player needs to swing at strikes more. If the player shows fear of the ball by standing too far from the plate or stepping away from it, this needs to be corrected.
  • Player rarely walks, has a high contact %, but has a low OBP + ROE. This player is swinging at bad pitches. This player needs further instruction on what types of pitches to swing at, which to lay off. One drill that can help with this is to have the player stand with his bat but don’t swing during a bullpen session. Have him yell “swing!” or “take!” with each pitch, before the pitch is halfway to the plate.
  • Player has high contact % and walks at close to the average rate, but most hits result in infield outs. You likely figured this out visually before enough stats were collected to tell you—this player likely has mechanical issues that need to be addressed.
  • Player hits an unusually large number of popups and/or foul tips. This player is likely swinging at high pitches. Tell him to “lay off the high stuff.” Tell him again. And again. And again.
  • Player hits an unusually large number of grounders. Do lots of tee work to get the player used to hitting line drives. Also teach the player to bend his knees when swinging at low pitches.
  • Player has a slightly worse than average strikeout rate, below average contact rate (many called strikes), yet still manages to have an average number of walks and much higher than average OBP + ROE. Many coaches would be tempted to ask this player to swing more at strikes. But talk to this player before doing so. You may find out that he is purposely waiting for “his” pitch—and that is the key to his success. If that’s the case, you should probably let him do his thing.

My guess is that I’ll be able to detect more hitting issues with stats as I get more data and more experience interpreting the data. But I don’t want to get carried away. A coach who knows a lot about hitting mechanics can often figure out many of these same tendencies with a few minutes of visual observation. That’s a lot faster than waiting for data to be collected for a month or two and then interpreted, sometimes incorrectly.

Summary

When pitching is first introduced, throwing strikes matters more than anything. Strike % is a stat that quickly stabilizes as the number of pitches mount, and is terrific at predicting most traditional measures of pitching success at the 9-10 year old rec league level.

At the all-star level, hitters are better, so pitchers need to do more than just throw strikes. Pitchers without uncommonly high velocity will need to change speed and location.

Hitting is more difficult to measure due to sampling size issues. Contact % is a useful metric, but must be combined with other stats (K, BB, pop fly rates, grounder rates) and visual observation of mechanics, hustle, and how hard the ball is hit in order to get a complete picture.

While OBP + ROE is probably the best overall measure of how much a Mustang player contributes to run scoring, it fluctuates naturally due to random factors. So be careful not to use OBP + ROE in isolation at the youth level. The sample size is too small. Ideally, youth baseball scoring software would place more of an emphasis on per pitch stats, to help with the sample size issue.

What happens as field size increases and players get older and more skilled? I don’t know. It will be another year before I get to observe my son and his teammates in the Bronco division. And then it will be time to write the next installment:

Youth Baseball Stats Part 4: Moving to a Bigger Field Changes Everything

Author: Joe Golton

I’m a dad with a son who loves baseball. Professionally, I’ve been a software developer, investor, controller, and logistics manager. I now make my living from this blog, supplemented with occasional consulting gigs.

Leave a Reply

Your email address will not be published. Required fields are marked *