The Sabremetric revolution in baseball has adhered to the idea that OBP is a prime measure of an offensive players worth. Certainly not the end only, but a major statistic to consider. In time this led to folks combining this stat with slugging percentage to convey an importance to both hitting with power and getting on base. And as flawed as it is OPS has taken root as a major stat.
I've learned to appreciate OPS - it does give a sense of how valuable a player is offensively without the complexities of wOBA, but the non-stats orientated fans I've tried to explain it to have usually been confused. So, without delving deep my basic qualms with OPS are twofold -
First of all, it combines two measures that are ratios with different denominators (plate appearances for OBP and official at bats for SLG). So in it's initial conception this stat is adding apples and oranges.
Second of all, I feel that both SLG and OBP both are an extension of the concept of batting average, So OPS is counting traditional batting average twice.
With this in mind It seems what we are really trying to look for here is a a way to combine OBP and ISO. Based on the traditional slashline, I'd suggest that OPS-BA would logically appear to be close to this.
After mulling it over in my head for a while, and playing occasionally with the idea, I have come up with a different way to combine these with some basic measures. In this formula I will use the idea from slugging percentage and look at total bases, and in essence count walks and HBP's as a base. I'll call this the "Adjusted Slugging Percentage" and use SLG* as the symbol...
SLG* = (TB + BB + HBP)/PA
Yes, I know this is not a 100% accurate representation of OBP, but I'm not only looking for a better statistical model to look at a player's offensive worth, I'm looking to not clutter things up. So yes, if a player has a lot of sacrifices and other infrequent results this formula will misrepresent things a little. My original though of subtracting BA from OPS to account for the double counting of BA in both OBP and SLG will be called "Est SLG*"
When I look at the Astro's stats from up to yesterday, I find that there is a 99.9 correlation between my SLG* formula and OPS-BA. Perhaps some of the difference is the components of OBP I'm overlooking, maybe some of it is due to the PA versus AB in the denominators. Maybe it's just to be expected from design?
Another thing about OPS that bothers me is it really does not account for a major part of the game - scoring runs. Runs scored has always seemed to me to a neglected stat. But for a runner on first who scores on a walk-off double, it also seems that the batter should get some credit for that important play. In traditional stats the batter would get an RBI and the runner a run scored.
So I come up with the concept of Total Runs - basically runs scored plus RBI (as flawed as that stat is). But in order to not double count a home run, we'll subtract those:
TR = (R + RBI - HR)
(Ironic that the rally killer actually is a negative in this formulation...)
From here it's an easy step to divide total runs by plate appearances (giving a measure of how often a batter participates in a scoring play), and you have two measures of important, but different, parts of the offensive picture that are based on the same denominator. If I call this the "Adjusted Slugging plus Total Runs per Plate Appearance" it might be a mouthful, so maybe we can just call it SRP?
SRP = SLG* + (TR/PA)
Now of course, it's possible for a pinch runner to score and total runs per plate appearance to infinite. I'd say just call that zero...
Now, how does this contraption I've dreamed up compare to OPS? After all, since I'm wanting to purge the mathematically orientated baseball fan/stats geek to give up OPS for this new measure, can I show it's worth a crap?
I'll take the Rays next opponent, the Astros...and look at the data from baseball reference for games through 6/30/2013. Please Note - I am only using players with plate appearances - Josh Fields has scored a run without a PA, and I am using the RBI's in the team number when I really feel that looking at team numbers runs/PA is the better number. And when computing the correlation coefficient I am excluding pitchers.
Overall the SRP is usually a lower number than OPS. It varies among players, but for the Astro's only one person has a higher SRP than OPS, utility player Brett Wallace. He only has 46 plate appearances. For the entire team, the OPS is .676 versus an SRP of .596 (SRP is 8% less). Breaking up the team SRP, we have a SLG* for the team of .423 (compared to an "estimated" SLG* of .440 using OPS - BA). The Total Runs per plate appearance is .173, which means it takes about 5 3/4's plate appearances for a run or RBI to happen. Again, for the team as a whole I think that runs per plate appearance is
a better number - for the team it's .103, or almost 9 3/4 plate appearances to make a run. For comparison, the Rays (since this is very easy to look up and calculate) are at .136 runs per plate appearance (about 7 1/3 plate appearances per run).
There are no real surprises between the OPS and SRP, despite the fact that players will get ordered slightly differently. While going strictly by OPS the Astro's 2 catchers,Carlos Corporan and Jason Castro, lead the pack, but the SRP indicates that the DH, Chris Carter is the most productive hitter. I can't see immediately why the catchers take a hit in my system...except that between the 2 of them they have almost 100 more plate appearances than Carter with pretty similar overall run production.
My SRP indicates what we all know about Carlos Pena - despite a slash-line that give traditionalists heartburn, he's still a serviceable major league player, or at least on a team like the Astro's. That is to say, with respect to OPS, SLG and my measures, he's still above average on that team.
When I look at the list of players sorted by plate appearances, I find that everyone with over 200 PA's so far this season has an OPS that at or above the team average, but one (Jose Altuve - I presume their normal lead off guy based on him having the most PA's) is lower than the team average. It seems that SRP might be a more stringent measure than OPS, but with OPS having traditional batting average in it twice, maybe that should not be surprising.
Anyway, Altuve - if he is the lead off guy - intrigues me that his SRP is .571 compared to the team's overall .596. It makes me think that he's either not getting on base enough or not scoring a lot of runs. When I look at his slash line and run scored I see this:
0.292/0.327/0.375, 32 runs scored
The 0.327 OBP is better than the team's overall 0.296, but there are 3 guys with more runs scored than him. The power numbers between these 4 players vary, but Altuve has got the lowest total runs per plate appearance between the 4 (but very close to Jason Castro).
The OPS/SLG* correlation is 91%. So I would guess that maybe close to 10% of the players will see their ranking change when looking at OPS versus my SRP.
The link to my spreadsheet (Open Office format) is:
Thanks for considering this...although I know it's not going change the world, but maybe it can generate some discussion.