Quantifying Player Development in the Minor Leagues

Over the past few years, the baseball landscape has seen a growing increase in the usage of technology, giving teams across all levels insights into their players, creating a host of new possibilities within player development. Add on top of this the seemingly ever growing focus on value and ROI (which a good PD staff can be great for) in Major League Baseball, and focus on player development within baseball in the public and private realms has unsurprisingly grown significantly. Success of notable PD stalwarts such as the Dodgers, Astros, and Rays in recent years hasn’t hurt either.

Just about two years ago, I published a piece at The Hardball Times on the subject of quantifying player development skills within an organization. This research focused primarily on player development at the major league level by using player projections produced by Steamer as a value for a player’s baseline performance. If a player over-performed or under-performed their projection, those deltas were subsequently attributed to their organization they were with.

The research idea for this project was spawned a few years ago thanks to a tweet by Kyle Boddy, someone who has been very involved, externally and internally with the pro ball-sphere, in this increased focus on player development within the sport. He was gracious to provide some further insight on the subject of player development for this research.

“True organizational player development doesn’t care about a prospect list or so-called ‘priority players,’ especially for pitchers. Arms that can contribute to a major league roster can come from any source – the amateur draft, Rule 5 draft, pro free agency, international signings, and many other avenues. Getting specific characteristics in the door that your team isn’t very good at developing and tolerating deficiencies in areas where your player development department can meaningfully improve is a recipe for success. These are some of the concepts I learned from nearly a decade in professional baseball, and a vision I sought to bring to the Cincinnati Reds from 2019-2021. My coaches were on board and the leadership that brought me into the organization put their full weight behind it. The results speak for themselves, and I’m proud of my coaches who are responsible for the bulk of the improvement – they were in the trenches. I merely laid out the system and helped facilitate it.

-Kyle Boddy

Putting together a player development department that consistently produces results is no small undertaking. It’s no coincidence that teams that have seemingly been able to achieve that have seen success at the highest level.


While the process in my previous research was surely an imperfect measurement, as all player development within baseball doesn’t come from the 30 major league baseball organizations, and each over/under-performance by a player can’t be assumed to be the workings of a developmental process, I felt that with a large enough sample, everything would at least come close to evening out and I’d have a quantitative measurement of which organizations were good at developing players and which ones weren’t. At the time, I felt that the results mostly matched up with intuition acquired from information made public (i.e. the Rays are good at developing players!), which was good to see.

As time has past since I did this project, more questions arose in my head and I wanted to reopen research into this subject. To start, while developing players at the major league level is surely very important (look at what the Giants did last season), most player development efforts and resources within a major league organization are allocated to players in the minor leagues, probably rightfully so. Players in the minors are younger, therefore leaving more room for the development of their skills, therefore leaving more room for actual player development to take place. The average 20th round starting pitcher pitching in a complex league in their age 21 season is going to have more areas to work on in their game than the average 30-year old starting pitcher with 500 major league innings under their belt.

So in this updated research, rather than looking at an organization’s ability to develop players at the major league level using projections, I wanted to quantify player development at the minor league level, using mostly the same process. As you can probably guess, using this process at the minor league level is a bit less straightforward and would require some new steps.

To start, there is no projection system available to the public for minor league players. For this research, I would need to create historical projections for each minor league player season. The projection system is pretty simple and is built off a weighting of past seasons, regression to the mean, and applying an aging curve. It surely isn’t built for maximum accuracy, but it should be good enough for the purpose of this research.

  1. Acquire minor league data (via FanGraphs) on a season-level for both hitters and pitchers at each level for every season since 2007. For hitters, I’ll use wRC+ and for pitchers I’ll use FIP+ (FIP scaled for league averages). I’m not the biggest FIP fan, but it was either that or ERA.
  2. Find major league equivalencies (MLE) for each level to normalize for strength-of-opponent for each player. In a nutshell, a player who hits for a 130 wRC+ in Triple-A is better than a player who hits for a 130 wRC+ in Low-A, at that current time. MLEs add context for this and put each player on the same level. As an example, a player jumping from Triple-A to the major leagues is expected to see a 20% reduction in performance, based off historical data. With a few more calculations, you’d find that a player jumping from Low-A to the major leagues would expect to see a 60% reduction in performance. Therefore, a 130 wRC+ in Triple-A would be adjusted to a 110 MLE wRC+, while a 130 wRC+ in Low-A would be adjusted to a 70 MLE wRC+ (H/T to Clay Davenport for the MLE numbers).
  3. Group performances together for players who played at multiple levels in one season. A player who posts a 70 MLE wRC+ in Triple-A across 200 plate appearances and 80 MLE wRC+ in Double-A across 200 plate appearances would have a 75 MLE wRC+ across 400 total plate appearances.
  4. Regress MLE wRC+ and MLE FIP+ to the mean. This will help control for varying sample sizes.
  5. Calculate MLE wRC+ and MLE FIP+ aging curve.

With all the minor league data acquired, let’s start out with a snapshot into what the best minor league seasons since 2007 are with no context adjustments added.

With MLEs and quality-of-level accounted for, here’s what the top 20 seasons now look like. Unsurprisingly, these are all in the upper minors.

Projections

With all the historical data lined up and adjustments made, here’s where get into the key area of this research: establishing a baseline expectation for each minor leaguer heading into a respective season. As I mentioned above, not all over/under-performances by a player can be directly attributed to an organization’s player development, but establishing an expected level of performance for each player, we can determine how often an organization has a player that exceeds expectations, which is one of the more obvious signs of a good PD system.

The first step in this process is a necessity for pretty much any projection system, implementing an aging curve. By finding the weighted mean for change in production from age to age + 1 (23 to 24, 29 to 30, etc.), smoothing the line of best fit, and then applying that expected change to each player season, context for the effects of aging, which should be independent of an organization’s PD skills, can be accounted for.

Much research on the aging curve of Major League Baseball players has been done in the past. According to research done by Chet Gutwein at FanGraphs, hitters usually see slight improvement year-to-year up until just past their mid-20s, plateau in performance around the age of 30, and then see growing decreases in their performance the further they get into their 30s.

For the minor league aging curve, the dynamics are obviously different. For reasons such as…

  • Most hitters and pitchers in the lowest levels of the minor leagues are still quite far way from fully developing their bodies
  • There’s far less variance in age across every level of the minor leagues compared to the majors.

For hitters, the minor league aging curve shows a rapidly increasing gain in performance each year from the age of 17 until 24, where strides performance-wise for a hitter peak. Hitters still make decreasing year-over-year gains from ages 26 to 30, followed by a plateau and decline into their 30s.

For pitchers, the minor league aging curve shows the same rapid improvement in results on the mound from the same ages, 17 to 24, where the curve peaks. And just like hitters, a shrinking increase year-over-year is seen to late-20s/early-30s, followed by a plateau and peak. With pitchers, there is one very noticeable difference. The peak in performance gains is more pronounced.

With the aging curve set up and the performance measurements adjusted for sample size and level, the projections are ready. There’s not really a whole lot to get into with this part. It’s a simple weighting system that rates recent seasons more heavily, less recent seasons less-so.

Though the goal here wasn’t to build the most accurate projection system possible, I did want to make sure what I was doing made sense.

R^2RMSE
Adjusted wRC+ (n) vs. Adjusted wRC+ (n-1)0.51914.69
Final wRC+ Projection (n) vs. Adjusted wRC+ (n-1)0.5545.85
Adjusted FIP- (n) vs. Adjusted FIP- (n-1)0.64311.26
Final FIP- Projection (n) vs. Adjusted FIP- (n-1)0.64410.04

With the historical projections now in place, we can now start the evaluation on an organizational-basis. I’ll be weighting the delta between actual and projected performance by PA/IP and find the mean difference. I then scaled the average to 100 and adjusted the numbers to add in seasonal fluctuations (2021 saw a lot more variance than past years, for obvious reasons). For recency, let’s start with 2021. At the top, there’s a good mixture of some surprises and non-surprises. On average, Royals minor leaguers out-performed their baseline expectations more than any other organization’s minor leaguers. This is a worst-to-first story, as this aligned with a significant overhaul in their player development after the 2019 season, a year that saw them rank dead last in this measurement.

Filling out the rest of the top five is a list of teams associated with player development success: Yankees, Blue Jays, Dodgers, Astros.

A big litmus test for this project was seeing where the Rays ranked with pitching development. In 2021, their pitchers out-performed their baseline expectations the second most of any organization, only trailing the Yankees, so that was good to see. Looking at the opposite end of the spectrum, the three organizations the under-performed their projections the most were the Angels, Tigers, and Rockies.

SeasonTeamHitters ScoreHitters RankPitchers ScorePitchers RankOverall ScoreOverall Rank
2021KCR103.83102.93103.31
2021NYY102.26103.81103.02
2021TOR104.02101.46102.73
2021LAD105.3199.616102.54
2021HOU102.94101.18102.05
2021PIT100.512102.94101.76
2021TBR100.016103.12101.67
2021CLE102.35100.713101.58
2021CIN101.09102.05101.59
2021CHC100.610101.010100.810
2021PHI99.918101.37100.611
2021BOS100.016100.714100.312
2021CHW99.918100.615100.213
2021SFG100.41399.41899.914
2021NYM100.51199.12099.815
2021ARI101.9797.72699.816
2021MIL99.71999.21999.517
2021BAL99.22199.61799.418
2021SEA100.31498.52399.419
2021OAK101.1897.52799.320
2021STL97.526101.0999.321
2021TEX97.525100.91299.222
2021SDP99.52098.32498.923
2021MIN99.22298.22598.724
2021MIA98.12499.12198.625
2021ATL95.330101.01198.226
2021WSN95.82998.92297.427
2021LAA98.92395.83097.328
2021DET96.92797.42897.229
2021COL95.92897.22996.630

I’m not sure if it is wholly fair to judge a player development department off of one season though. There’s still a lot of external factors that are outside of any organizations control that can influence these results. Increasing the sample size would help with this, so let’s see how organizations have done since 2018.

The Blue Jays now take the top stop. Since 2018, their hitters have out-performed their baseline expectation more than any organization’s hitters, while in the pitching department they rank seventh.

Taking the three top spots in pitching development are the Astros, Rays, and Yankees. For hitting development, it’s the Blue Jays, Dodgers, and Reds.

TeamHitters ScoreHitters ScorePitchers ScorePitchers RankOverall ScoreOverall Rank
TOR103.51101.37102.41
HOU102.24102.41102.32
TBR101.28102.42101.83
PIT101.37102.04101.64
NYY101.110102.13101.65
PHI101.29101.76101.46
LAD102.8299.518101.27
CLE100.912100.88100.88
ARI101.8599.519100.79
SDP101.7699.323100.510
CIN102.4398.428100.411
ATL98.821101.95100.412
MIL100.21599.815100.013
CHW100.81399.12699.914
LAA100.61499.22599.915
NYM99.81799.71799.716
BOS99.61899.71699.717
KCR98.722100.6999.618
STL98.920100.41099.619
MIN98.524100.31199.420
DET99.21999.42199.321
OAK100.91197.22999.022
SFG98.62399.42099.023
MIA97.828100.11299.024
CHC97.82799.91498.925
WSN98.22599.22498.726
SEA98.02699.42298.727
TEX96.930100.01398.528
COL99.91696.83098.429
BAL97.72998.52798.130

And since the player development landscape in baseball is ever-changing, it might be useful to look at the trends across organizations.


The landscape of player development within baseball will continue change, and with that the performances of each organization to a degree. Innovative ideas and technology have changed the sport and will continue to do so. Add in the increasingly clear importance of organizational skill in developing players, and a purpose for evaluating these organizational performance trends exists.

Featured Image Credit: Tim Campbell

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Create your website with WordPress.com
Get started
%d bloggers like this: