BEATScore: The New Sabermetric Of Baseball

A sabrmetric endeavour by Jaden Hellmann.

As a data geek and baseball fan, the way we quantify talent is fascinating. There are so many arguments that can be made for different methods of analysis or metrics and yet there isn’t really a gold standard around this heavily discussed topic.

We have seen different metrics developed over the years and as we move into the statcast
era, every piece of data is being scrubbed for competitive edge. While this numerical voyage I set out on could be interpreted as ‘just another number’, I would like to show exactly why I think it’s rooted within the game, it’s history and belongs amongst core baseball data when discussing player value both from interchangeable perspectives pitching and hitting.

Quickly I want to start with WAR and bring up a couple things that bother me about what it
tries to accomplish. First, there isn’t a universally accepted formula that seems to be
agreeable to all. Next, the arbitrary nature of the metric bothers me. We only see 1000 WAR
represented per year (with 30 teams in the league playing a combined 2430 games) however this is skewed unevenly between pitchers and hitters (the 57-43 balance is arbitrarily aimed at encompassing defensive abilities which seems very high). All of which still detracts from the randomness that was done in implementing a 1000 game cap retroactively to make the stat work across different time periods.

Why not depict the actual weight over replacement by the games played? It would seem a multiplier of 2.43 would potentially accomplish this as this would depict an actual full season of value and not 1000 games. I have great respect for the attempts made with WAR and I credit WAR with giving me inspiration for what I have developed. Let’s get into that now.

google.com, pub-1100396528751056, DIRECT, f08c47fec0942fa0

BEATScore = bases earned and taken score.

Bases earned and taken score is a dual use metric for hitters and pitchers which focuses on the ability of the hitter to create productive plate appearances, and the pitcher to stop them.

So what is ‘productive’? Reaching base, stealing bases, moving runners with intent.
BEATScore is partially calculated using the following formula:

BEAT = Total Bases + Walks (including IBB) + HBP + SB + SH + SF – GDP – CS

This gives us a number of bases earned in total. This is an important number when looking at the historical precedent for this metric which we will take a look at. This finding is part of the reasoning for why I think this formula is valid.

Here we have 2 charts showing us the relationship between BEAT and actual Runs scored by year:

1871 – April 2022
1954 – April 2022

As we will note, there is missing data not tracked by the MLB before 1954 and this negatively impacts results. However from 1954 on, we can see that the consistency with which baseball has consistently demanded around 4 bases to score a run is both poetic and incredibly obvious when we think about it.

With BEATscore, we are accounting for each PA and it’s contributions to offensive output.

To take the formula further we can divide BEAT above by PA. This gives us a number equivalent to a ‘potency’ measure. It shows how many BEAT per PA. Next we will take the aforementioned data for the full league to find what ‘average’ is:

((BEAT/PA)^2)/(((BEAT/PA)^2)+(BEAT/PA)^2))

With our league average now scoring at .500, we can rerun the same formula again but sub in player ‘X’ and repeat for all the others.

(X^2)/((X^2)+(BEAT/PA)^2))

This gives us an effective winning% or probability that can be compared to Average at .500. For example, if ‘X’ is a .550 BEATS, then this would mean that a team of just ‘X’ hitting against a team of average would finish a full season with a .550 w%. The kicker is that these probabilities are interchangeable and can be used to compare ANY 2 players head to head. This doesn’t stop at pitching either. Pitching calculations are done using the same starting point.

BEAT = Total Bases + Walks (including IBB) + HBP + SB + SH + SF – GDP – CS

When it comes to normalizing league average data for pitchers the formula takes on one
change.

((1-BEAT/PA)^2)/((1-(BEAT/PA)^2)+(1-BEAT/PA)^2))

This time, we are taking the inverse. The result is our pitching BEATS formula. Now with both sides of the ball equally normalized, pitcher and hitter comparisons become much more realistic. It will be noticed that the ranges for hitters is much more extreme than that of pitchers, which currently speaks to the parity in pitching talent across baseball right now.

Welp, I think that about sums it up. There you have it! BEATScore. Keep an eye out for
leaderboards and more!

One response to “BEATScore: The New Sabermetric Of Baseball”

Leave a Reply

Create a website or blog at WordPress.com

%d bloggers like this: