The text in the images above shows the 19 metrics which I’ll explore in more detail below, scattered across several of Baseball Savant’s colorfully intriguing baseball chart graphics
About Me
Hello, I’m Josh Goldberg! I’m a proven sports, technology and healthcare data strategist and thought leader. Experienced in Analytics, Strategic leadership, Governance and Infrastructure, I am also expert in SQL, PowerBI, R, ChatGPT, Python, and an award-winning Tableau professional. Moreover, I’m currently completing a five-year New York State STEM-Scholarship. I have accrued 35+ professional learning licenses and certifications and I’m an actively engaged leader for numerous social organizations across the New York City area.
Course Context
Over the past several months, the Society for American Baseball Research (SABR) has released two Lessons within their SABR Analytics Certification Course. The class is designed to enhance the baseball minds of Sabermetrics fans worldwide. There are four Lessons within the overall certification, including:
Lesson 1 — Conversational Analytics and Critical Thinking in Baseball
Lesson 2 — Advanced Analytics in Baseball
Lesson 3 — Advanced Baseball Programming (SQL, R, Python)
Lesson 4 — Using Data Science to Create and Replicate Work
Each section dives progressively deeper into the comprehensive universe of baseball and data, with a strong emphasis towards Sabermetrics and analyzing baseball player data. To this point, only Lessons 1 and 2 have been released, though both are extensive, valuable and indispensable for avid baseball aficionados looking to sharpen their understanding of the vast and ever-expanding industry of baseball data.
This spring, I completed Lesson 1 and I’m actively working through Lesson 2. Lessons 3 and 4 will likely be developed and released later this year as well as into 2024. In the meantime, it’s a perfect time to work through the first two Lessons and leverage critical thinking and problem-solving skills to become more in tune with the advanced metrics that currently exist across Major League Baseball. The future Lessons will apply this knowledge within a more technical lens, which is why it’s critical to become fluent in these concepts and understand their situational importance before parlaying them into more detailed scenarios.
Lesson 1 & Case Study
The first Lesson is an immersive overview into MLB’s most common statistical metrics. There’s also detailed explanations of practical applications for when these metrics are used for player comparisons across MLB. The Lesson concludes with a case study, which highlights baseball statistics for two active MLB players.
At the completion of each Lesson, there is a hands-on project where you’re able to demonstrate your new baseball analytics insights within a brief case study. For this project, you are presented with player data for two types of players: hitters and pitchers. For each of these two positions, there is data on two (unspecified) players. For each position, your job is to act as the GM. In this article, I’ll refer to my selections as those by Josh the GM. You will decide which of the two players you would pick for your team for each position and make a statistically-backed case (considering both basic and advanced metrics) demonstrating your player advocacies.
The first Lesson is a fun way to start becoming more familiar with how to use these metrics in the context of real players. One of the best parts about the case study is that once it’s submitted, a staff member from Baseball Prospectus will review your analysis and share feedback. How’s that for some instant feedback from high-profile baseball data experts?! I received feedback on my project within a week, and it was detailed, thorough and highly analytical. For reference, it appears that there will be a case study in each of the remaining Lessons, as well.
Motivation
Given the depth of analysis that I put into my player comparisons for Lesson 1, I enjoyed the ability to think critically about these two players and figured that I could share several learnings from this experience with fellow stat-heads. That I am an enthusiastic writer translated well into this context. It also furthered my motivation to turn some of my findings into a short, extended public piece for others to utilize as they consider taking or reflecting on the course (and especially Lesson 1).
This is the point where I will share the disclaimer that if you plan to take this course and do not want to become influenced by my analysis, you should pause here, go complete the Lesson, and then come back once you’ve had the chance to put your critical thinking skills to the test yourself first.
MLB Advanced Metrics
Before exploring the data underlying the Lesson 1 case study, I want to share a brief overview about the 19 main advanced metrics that the Lesson emphasizes. This chart below is a high-level summary about the stats, though there are numerous other sources you can peruse to understand the deeper logic and rationale behind them, including several which I’m listing out here below:
As I delineated within the chart, you’ll notice that 12 of the metrics are hitter — specific, 4 are pitcher — specific, and 3 apply to both player groups.
Name / Abbreviation | Abbreviation Explained | Description | Formula | Applies Directly to Offensive Metrics | Applies Directly to Defensive Metrics | Applies Directly to Pitching Metrics | Leveraged for Comparison Analysis |
BABIP | Batting Average on Balls in Play | Measures how often a batter gets a hit when they put the ball in play. | (H - HR) / (AB - K - HR + SF) | X | X | X | |
DRC+ | Defensive Runs Created Plus | An all-encompassing metric for batters which focuses on each hitter’s expected contribution rather than merely averaging the result of hitter PAs. | Complex to explain and additional resources have more information. | X | X | ||
DRS | Defensive Runs Saved | Measures how many runs a defender saved. It takes into account: errors and range and outfield arm and double-play ability. | Complex to explain and additional resources have more information. | X | X | ||
ERA+ | Adjusted Earned Run Average | A metric which normalizes a player's ERA to across the League. | League ERA (adjusted for park factors) x 100 / ERA. | X | X | ||
FRAA | Fielding Runs Above Average | Defensive metric created using play-by-play data with adjustments made based on plays made and the expected numbers of plays per position and the handedness of the batter and the park and base-out states. | Complex to explain and additional resources have more information. | X | X | ||
FIP | Fielding Independent Pitching | Similar to ERA though it focuses solely on the events a pitcher has the most control over including strikeouts and unintentional walks and hit-by-pitches and home runs. It entirely removes results on balls hit into the field of play. | ((HR x 13) + (3 x (BB + HBP)) - (2 x K)) / IP + FIP constant | X | X | ||
fWAR | Fangraphs' Wins Above Replacement | Calculates the Wins Above Replacement value according to Fangraphs' calculations. They leverage wOBA as their baseline hitting stat and IP as their baseline pitching stat. | For Hitters: fWAR = (Batting Runs + Base Running Runs + Fielding Runs + Positional Adjustment + League Adjustment + Replacement Runs) / (Runs Per Win). For Pitchers: fWAR = [[([(League “FIP” – “FIP”) / Pitcher Specific Runs Per Win] + Replacement Level) * (IP/9)] * Leverage Multiplier for Relievers] + League Correction | X | X | X | |
OPS+ | On-base Plus Slugging Plus | Combines a player’s on-base percentage and slugging percentage and adjusts it for league and ballpark factors. | (OPS / League OPS) x 100 | X | X | ||
OAA | Outs Above Average | The cumulative effect of all individual plays a fielder has been credited or debited with. This makes the metric a range-based metric of fielding skill that accounts for the number of plays made and their difficulty. | Total value of all fielder play probability outcomes. | X | X |
Table Continued
Name / Abbreviation | Abbreviation Explained | Description | Formula | Applies Directly to Offensive Metrics | Applies Directly to Defensive Metrics | Applies Directly to Pitching Metrics | Leveraged for Comparison Analysis |
rWAR(or bWAR) | Baseball-Reference's Wins Above Replacement | Calculates the Wins Above Replacement value according to Baseball-Reference's calculations. They leverage the replacement level calculation of 1000 WAR per MLB seasion as their baseline hitting stat and runs allowed as their baseline pitching stat. | Complex to explain and additional resources have more information. | X | X | X | |
RF | Range Factor | Measures a player’s defensive contribution by looking at how many plays they made per opportunities as expressed by assists and putouts divided by total games played. | (Fielder's putouts + assists) / defensive games played | X | X | ||
SIERA | Skill-Interactive Earned Run Average | Quantifies a pitcher's performance by trying to eliminate factors the pitcher can't control by himself. But unlike a stat such as xFIP - SIERA considers balls in play and adjusts for the type of ball in play. | 6.145 - 16.986(SO/PA) + 11.434(BB/PA) - 1.858((GB-FB-PU)/PA) + 7.653((SO/PA)^2) +/- 6.664(((GB-FB-PU)/PA)^2) + 10.130(SO/PA)((GB-FB-PU)/PA) - 5.195(BB/PA)*((GB-FB-PU)/PA). | X | X | ||
UZR | Ultimate Zone Rating | measures a player’s defensive contribution based on the number of runs they saved or cost their team. | Complex to explain and additional resources have more information. | X | X | ||
WARP | Baseball Prospectus' Wins Above Replacement | Calculates the Wins Above Replacement value according to Baseball Prospectus' calculations. They leverage the replacement level calculation of 1000 WAR per MLB seasion as their baseline hitting stat and Deserved Runs Average (DRA) as their baseline pitching stat. | Complex to explain and additional resources have more information. | X | X | X | |
wOBA | Weighted On-base Average | Metric that measures a player’s overall offensive contribution by considering not just their batting average but also their ability to get on base and hit for power. This is done by giving “weight” to hits based on how many bases. So a double counts with more weight than a single. | (unintentional BB factor x unintentional BB + HBP factor x HBP + 1B factor x 1B + 2B factor x 2B + 3B factor x 3B + HR factor x HR)/(AB + unintentional BB + SF + HBP). | X | X | ||
wRC+ | Weighted Runs Created Plus | Measures a player’s total offensive contribution and adjusts it for league and ballpark factors. | (((wRAA per PA + League runs per PA) + (League runs per PA - ballpark factor x league runs per PA) / league wRC per plate appearance)) x 100 | X | X | ||
xFIP | Expected Fielding Independent Pitching | xFIP finds a pitcher's FIP but it uses projected home-run rate instead of actual home runs allowed. The home run rate is determined by that season's league average HR/FB rate. | ((Fly balls / League average rate of HR per fly ball x 13) + (3 x (BB + HBP)) - (2 x K)) / IP + FIP constant. | X | X | ||
xSLG | Expected Slugging Percentage | Calculates slugging percentage using the expected: singles - doubles - triples - home runs and outs instead of the actual outcomes. | (1B + 2Bx2 + 3Bx3 + HRx4)/AB) | X | X | ||
xwOBA | Expected Weighted On-Base Average | Measures offensive value of players by weighting outcomes like HR - BB - 1B etc.) by their run value. | (xwOBAcon + wBB x (BB-IBB) + wHBP x HBP)/(AB + BB — IBB + SF + HBP) | X | X | X |
With a somewhat robust understanding of these metrics, the data and evaluation for these players will start to make more sense. In the four tables below, you will note the player data tables with which I was presented for the Lesson 1 case study. The data involves looking across four years of playing data for each player, though not necessarily the same four years of experience, by age, for each player. From a quick read through each of these metrics, you’ll notice that the data include offense, defense and pitching statistics.
Hitter Data
Player A Statistics
Player B Statistics
Pitcher Data
Player A Statistics
Player B Statistics
Trend Highlights
Using these player data, we evaluate performance for hitter A and B as well as for pitcher A and B (for simplicity, going forward I will generally refer to Player A as A and Player B as B). In the six bar and line charts below, I focused on capturing trends or insights that are especially relevant within the context of my case study analysis to provide a quantitative perspective. For example, the first chart shows how hitters A & B differed in Deserved Runs Created Plus (DRC+). The other charts demonstrate data comparisons for Fielding Independent Pitching (FIP) and the three Wins Above Replacement (WAR) metrics (fWAR, bWAR/rWAR, and WARP). These charts were developed using Python, although they could easily be reproduced by other data tools of choice, including R, Tableau, etc.
While the charts are not complex to create through some code exploration, you can also leverage advanced AI tools to assist in developing data visualizations. For example, here are two instances highlighting how to use ChatGPT to create code for a basic line chart and ChartGPT to create the visualization for a simple line chart. While both of these resources have distinct pros and cons (speed, efficacy, etc.), each is a valuable resource to peruse in more detail.
ChatGPT
Chat GPT basic line chart example
ChartGPT
Chart GPT basic line chart example
Hitter Trends
Hitter Deserved Runs Created Plus by Hitting Season (Bar Charts)
Hitter WAR Totals by Player Season (Line Charts)
Pitcher Trends
Pitcher Fielding Independent Pitching by Pitching Season (Bar Charts)
Pitcher Fielding Independent Pitching by Player Season (Line Charts)
Now, with a slightly more informed visual awareness about how these players compare in metrics, let’s evaluate these concepts from a more contextualized and nuanced viewpoint. In the sections below, significant points of discrepancy are highlighted for each pair (hitters and pitchers), along with an impact summary, which denotes the strong performer across numerous stats. After outlining these findings, we’ll analyze how to select one hitter and one pitcher to join Josh’s team.
Evaluating Hitters A & B
After analyzing this data, we can discern that Player A is both the better hitter and defender than Player B. How can we tell? Below are some of the most valuable factors to consider for these two players.
Age & Development
Player A is younger and in the prime development of his career. We can identify this because of three main reasons, including:
He’s earlier on in his playing career
He’s a young player (i.e. early to mid-20's)
He’s currently malleable as a player. It’s easier to facilitate player performance strategy changes and modify existing approaches based on the recommendations of others.
However, Player B is in his early-30s and beginning to decline in production. Similarly, we can see this because:
Gradually decreasing WAR metrics performance
Gradually decreasing Defensive metrics performance
This general trend towards reduced positive output across this period could be due to numerous reasons, for example lessened production in these areas, among others:
Running Speed
Strength
Reaction Time
Physical ability
This observation is important to note because it’s unusual for players above the age of 30 to significantly enhance their production or to (successfully) make drastic hitting adjustments. To understand player decline in more detail, check out several stories from Fangraphs on the topic here and here.
Additionally, this player cohort typically declines in production rapidly. Essentially, a player like B here could hit a “wall” in his performance randomly (i.e. in 6 to 12 months), and then eventually stop playing in the MLB altogether. Conversely, the future potential for A is more promising, less risky and generally positively trending, as noticed through progressively-increasing production during these four years across several key metrics (i.e. wOBA, DRC+).
Offensive Production
In addition, we can tell that Player A is a much better hitter than B, with the biggest points of differentiation stemming from A’s better plate discipline (more walks and fewer strikeouts). From a standard statistical perspective, the pair’s offensive production is relatively similar (i.e. HRs, RBIs, R — runs, H — hits). However, the statistical divide is more pronounced with respect to walks (A has accumulated nearly 150+ more) and strikeouts (A has had 50 fewer). This area of stronger output in turn drives home better values for numerous metrics for A, for example, including: OBP, OPS, DRC+, the three WAR metrics).
Exploring adjusted offensive metrics, A showed a significant advantage in statistics like OPS+, wRC+ and DRC+, further emphasizing that he is a better overall hitter. The performance differential in WRC+ in particular is a great sign for A in terms of adjusted performance, because this metric is generally regarded as the best well-rounded metric for measuring offensive performance. There are a couple components underlying the metric’s reputational regard, namely:
It’s an adjusted, benchmarking stat, which takes into account player performance above/below a 100 scale and factors in league and park adjustments
It combines the value from wOBA, OPS and other offensive metrics (through not defense) to measure a player’s offensive values based on runs
While the three metrics highlighted here are top contenders for tracking offensive production, you can explore these other Sabermetrics investigative analyses from Bleacher Report and the Baseball Bible to evaluate them in more detail.
Additionally, while both players had a comparable stolen base success rate, A has three times as many stolen bases as B. If we’re considering these players within the current moment, A might have greater incentive to steal more bases given MLB’s recent rule changes around the base sizes. This could contribute to driving greater WAR values, as well, as compared to B.
Defensive Production
Moreover, A is also a superior defender. He has achieved better defensive metrics across the board. Most notably, he has saved 48 more runs than B across these four years. He has also accrued a DRS of 10 or more in 3 of 4 seasons, where a value of 15 or larger would potentially fall within Gold Glove range, according to the scale for this metric emphasized in Lesson 1. This is excellent defensive work by A, and significantly better than the same metric output for every season in this data set for B.
Evaluating Pitchers A & B
As with the prior example. Player A is the better player as compared to Player B. Here are the supporting highlights for my case below.
Age & Development
Leveraging a similar logic with player development in mind as for our hitters example, A is several years earlier on in his career. While both pitchers have similar levels of experience in games played and innings pitched, B is up against the “career clock” already and could begin to decline in production quickly, as aforementioned in the hitters example.
Pitching Effectiveness
For those still becoming more acclimated to Sabermetrics across MLB, it might be simple to notice that A’s ERA is significantly better generally and overall than it is for B. However, this factor is actually not a sharp differentiator, because when looking at ERA+ (adjusted ERA), we can see that both players are similar in performance.
Moreover, A’s strong performance in advanced metrics, including FIP and the WAR stats, highlights that he is a better pitcher. One driving reason for A’s higher total here derives from allowing fewer HRs, which again relates back to having better opposing hitter metrics than B.
Further reviewing the WAR comparison for these players, A has a higher total across the three WAR metrics. In two of the three calculations for the metric, B is generally decreasing sharply across this time span, while A either stays relatively consistent or worsens only marginally. This WAR spread is most pronounced in the two most recent years, where A is averaging nearly double B’s value (3.5 to 2.0). With an eye towards future development, A has a clear development advantage.
Evaluating these players’ performance within the current times, MLB’s new rules could also impact these players’ outputs. For example, B struggled with respect to walks and wild pitches. These two areas could become more challenging to navigate given MLB’s new pitch clock, potentially providing an advantage for A.
Recap
As a quick overview from the main points emphasized above, Josh the GM would select both “Player A” for the hitter spot and “Player A” for the pitcher spot on my team.
While the above evaluation sections detail these decisions, here are the strongest points of consideration in each case, for quick reference.
Hitters — Player A (vs. Player B)
Younger & better development potential
Better plate discipline (strikeouts and walks)
Significantly stronger defensive production
Pitchers — Player A (vs. Player B)
Younger & better development potential
Better Park discipline (less allowed HR’s, weaker contact induced and fewer allowed baserunners)
Greater collective player value
Contextual Implications
With respect to my role as Josh the GM in this case study, it’s also valuable to keep in mind several other factors which would be meaningful to consider if I were making these player selection decisions on behalf of an actual team. Several of these areas of focus include:
Team salary
Team league & division
Team’s propensity for leveraging Sabermetrics to drive results
Current team strategy (i.e. more offensive vs. defensive-minded, youth vs. veteran leadership, etc.)
Player desire to join the team
Player salary allocation capacity
Player camaraderie with other members of the organization
Current MLB rules/season
Comments