Player Grades: An Explainer

One of the primary, novel features of The Baseball Board are player grades. We wanted a new way to quickly evaluate a player's abilities and performance across multiple metrics. Wins Above Replacement, wOBA, Out-Above-Average, etc. are all great innovations for evaluating players, but require context to evaluate where a player fits within the whole league. This makes them good evaluations of "a player's abilities and performance across multiple metrics", but it doesn't meet our requirement of being a "quick" evaluation. Since there will be very few people that love baseball enough to end up on an obscure stat site and don't know what an A+ represents, we see this as a good alternative to quickly understand where a player stands in pure performance and among their leaguemates.

How the Grades Work

For a quick overview of our grading system, there are three different cohorts that are graded: hitters, starting pitchers, and relief pitchers. We break them into three groups so that they can be graded on a curve. That means the top performers of each group will get an A+, the next an A, and so on until F-. The approximate share for each grade is as follows:

• A+, top 3%
• A, next 4%
• A-, next 5%
• B+, next 6%
• B, next 8%
• B-, next 8%
• C+, next 8%
• C, next 13%
• C-, next 10%
• D+, next 10%
• D, next 10%
• D-, next 8%
• F, next 4%
• F-, bottom 3%

The idea behind this is the same as when grades are curved in school; it ensures the grading system always captures the current state of the game. If power becomes even more prevalent in 10 years, it's better to have the grades scale to still let the cream of the crop of power hitters stand out from the average, rather than 10% of hitters suddenly having an A+ in power. The scale itself is calculated from qualified players, using the standard qualifying stat logic for that group, but then all players with sufficient accumulated stats are then graded. That means A+ may contain roughly 3% of qualified players, but it could be higher or lower for all players, because not everyone that is graded is also used to calculate the curve.

Each group of stats (hitters and pitchers) are broken into six categories and an overall grade. Each category is some collection of stats blended together based on weightings we calculated. For example, the power grade for hitters is a blend of ISO, barrel rate, hard-hit rate, and xSLG, with most of the emphasis on ISO and barrel rate. We select our stats based on what we determined to be reasonable isolated signals for a specific metric, with additional metrics to try and smooth out outliers and noise. The current state of the grades is by no means perfect at capturing the signal for each group yet, but based on our research thus far we feel we are at a reasonable enough starting point to make these grades public and start a more open and collaborative iterative process to improving them, rather than isolated R&D in a labratory.

Since this site is in many ways a research project, both the stats used and the weights therein are going to be constantly evolving and hopefully improving. It is our goal to produce the most accurate, helpful metrics, rather than a static metric that always has one single definition. With that in mind, the details of what stats are used for each grade could become deprecated very quickly and no longer be accurate, so please keep that in mind when reading. We will try to update this article as often as we make changes, but only elephants never forget, so forgive us if this isn't always up-to-date.

Hitter Grades

• Overall: The weighted sum of all other grades. The weight emphasis for each sub-grade is: Results, power, contact, approach, speed, and fielding
• Power: The player's raw and in-game power, represented by: ISO, barrel rate, hard-hit rate, and xSLG.
• Contact: The player's ability to make contact, the quality of that contact, and not waste swings. This grade is made with a blend of: xBA and BA, contact_rate, and strikeout rate.
• Approach: The player's ability to make the best decisions at the plate (e.g. swinging at good pitches and laying off bad pitches). This is built off a "decision quality pitch matrix" we built, where we classify pitches as good or bad locations, based on type, and swings as good and bad, based on location and type.
• Results: The player's actual, in-game results and performance. We weight this most in our overall because it's easy to forget that even if a player is getting lucky, their results are ultimately what matters most for their team, so that deserves extra emphasis in their grading. We use RV/PA, wOBA, ISO, OBP, leverage-adjusted RV/PA, and a small emphasis on GIDP% and K% as "needle pushers".
• Speed: The player's raw and in-game speed using: sprint speed, 90-foot splits, bolt rate, baserunning RV, and base-stealing RV.
• Fielding: The player's overall fielding, this includes: Fielding RV, outs-above-average, arm strength/value, and, depending on position, some catcher specific and outfielder specific range, blocking, framing, etc. metrics.

Pitcher Grades

Starters and relievers share the same categories of grades, but their grades are curved against their respective cohorts:

• Overall: The weighted sum of all other grades. The weight emphasis for each sub-grade is: Results, stuff, control, in-play, location, and arsenal
• Stuff: The raw quality and traits of each of the pitcher's pitches. This metric focuses on stuff+ ranked by percentile for that given cohort.
• Location: The quality of the placement of their pitches, based on performance against similarly shaped pitches. This uses the same pitch location matrix described for the hitter's approach grade, but it focuses on the location itself, rather than the hitter swing decisions.
• Control: The pitcher's ability to throw strikes and manage their count. Based on: walk rate, zone rate, strike rate, and chase rate. This is one of the stats that's going it get adjusted the quickest, since multiple of these stats feed into each other, making them somewhat redundant.
• Balls In-play: A pitcher's ability to manage balls in-play via weak contact, groundballs, etc. We use: barrel rate, hard-hit rate, ISO, and BABIP.
• Arsenal: The depth and usability of a pitcher's total pitch mix. It's easy to confuse this with the "stuff" grade; the key difference is that the arsenal grade pays special focus to how deep a pitcher's array of high quality pitches goes, not just the quality of the pitches' shape and velocity. We calculate this by measuring the pitches quality, weighting them by usage rate, and rewarding pitchers by the more high quality, high-usage, pitches they have. Think of the stuff grade as the sum and the arsenal grade as the weighted sum of a pitcher's pitch quality.
• Results: The pitcher's actual, in-game results and performance, based on: ERA, wOBA, K%, BB%, RV/PA, and leverage-adjusted RV/PA.

What's Next?

Like mentioned previously, expect a lot of changes to the grading system in the future. Whenever we do make a substantial change to the grading system, know that we will implement that change retroactively across previous seasons as well, so two grades will always be comparable to one another. We have ideas on how to improve our existing grades, make weighting distributions more accurate, and introduce whole new grades that capture a separate part of a player's game, so expect a lot of grading discussions in the future. On top of that, we would love your feedback on our grading system. Do you feel like the grades accurately represent what they claim they to? Do you have a great idea for an improvement that would take our grades to the next level? Let us know!