What is your opinion of the A.J. Pierzynski signing?
MJH on accountability
Dude in UK,
As answer to your question: of course not. However, the thing I am wondering about has to do with the real impact that a given player has on the run production of a given team. I have no reason to question the way fangraphs is measuring runs created as a way of comparing one player to another player and deciding based on the average player who is above average and who is below average.
The question that I have is a bit more complex. It seems that if you look at statistics that are adjusted to compare every player to the average player that it could be very easy to get an inflated understanding of the REAL IMPACT of one player on the run production of a given team. You might say look at Thome's OPS and his RC numbers, they are unbelievably above average. But when you actually compare the real impact that he had on actual runs scored and RBI's, the REAL times when he was actually involved in the scoring of a run, it seems like the impact of a big bat lessens significantly depending on what sort of a lineup he plays in. I found that Julio Borbon scored and batted in about .200 runs per at bat while Thome scored and batted in about .24. Over 100 games this is a difference of only 15-20 real runs. It is definitely legitimate to say that who knows how many runs Thome would have scored and driven in, his real impact, on the run production of the Rangers? It is hard to know considering the very real difference in the Twins lineup and the Rangers lineup. I just wonder whether it is always the best thing to look at a statistic that compares a player to the average player as opposed to a particularized statistic that could measure better the actual reasonably expected impact that a Thome (or other) type bat might have on run production of a given team. The formula would have to be a lot more complex than what I talk about here, but basing such a number on the real runs impact seems to be a first step to arriving at this sort of number. Further, this sort of number would only be useful for a particular situation and would not give you a good idea of how good a given player is as compared to the average player. Those sort of statistics are invaluable and have already been created. Maybe my question ultimately would make no real difference in how one might measure the expected impact of a given player but I think it is interesting to think about. Could it be that a more particularized number might shed some light on a players impact that a number that compares a player to the average player is not capable of? Maybe in the end the answer is no, and there is no real worth in trying to come up with such a number. Regardless, I think the question is worthy of being asked.
I meant PA's and not AB's, sorry.
You're placing way too much emphasis on the RBI statistic. The real Runs Created statistic was invented to quantify the contribution every player makes to adding to his team's scoring. What you want is already out there. I just want to know wOBA or wRC if I need to know the offensive performance of a given player. A guy who gets on base often is contributing greatly to his team's runs scored. A guy hit gets on base equally well, but with power, is creating even more runs.
The player who is standing on 3rd with less than 2 out is probably going to be Elvis a lot more often than Benji, which means one of the guys hitting behind him therefore will get RBIs for simply putting the ball in play most of the time. The guy hitting behind Benji has got to do something a lot more spectacular to chase that fat butt home.
Vlad came to the plate last yr 370 times with men on base. Bautista only got about 270 such opportunities. Yet Bautista had a little better result in the RBI column (124 to 115). Its not hard to reason that Bautista could have helped the Rangers score more runs than Vlad if the players were switched. Is that what you are trying to measure?
Josh (who missed 29 games) had nearly equal opportunities with men on as Bautista, and they were both a lot fewer than Vlad. That's because Vlad was staring at Josh standing on base ahead of him about 40% of the time. Bautista had no one getting on base ahead of him at a decent clip. And yet, he drove in 124 runs. He had a pretty good season despite that because he had an outstanding .422 wOBA and a 169 wRC+.
I understand all of your arguments dude, I've acknowledged them all. I am not questioning the statistics that are already out there and I am not trying to measure what they measure (and I'm sure they do a great job at measuring what they were created for). I am suggesting that a very different sort of statistic might help. Considering you are repeating your arguments as if I disagree with them when I've already acknowledged their validity I very much doubt whether you have read what I wrote. I am talking about particularized statistical analysis that would be something totally different than a statistic that measures a player's abilities as they relate to the average player.
This thread makes my brain hurt.
Joe - As far as some of these stats go, I think you might be just a little bit mixed up. wRC+ is the only stat here that's intended for comparison to league average. The original RC is a statistical measure of how many runs you were worth to your lineup.
I think what you really need is a wRC created with a wOBA that's created from team/lineup specific weights. The formula to actually create new weights is EXTREMELY complicated (trust me, you don't want to see it); however, the formula for wOBA isn't that bad:
(0.72xNIBB + 0.75xHBP + 0.90x1B + 0.92xRBOE + 1.24x2B + 1.56x3B + 1.95xHR) / PA
Now if you determined that your lineup has a lot of speed, you might want to simply increase the value of getting on base which could make walks HBP, and 1B a little more valuable and devalue XBHs. If you adjusted the weights and ran the formula, you could then plug it into the wRC formula (a little more complicated than I’d like to manipulate right now). Essentially you could have wOBA and wRC formulas tailored specifically for the Rangers run scoring tendencies and then run it for all available players, and you’d have some sort of an idea of who would create the most runs in the lineup. Really you would also have to recalculate the linear weights for each spot in the lineup (OBP and speed would be more important than power in front of Hamilton). Really the problem with your goal here is that lineups change so frequently (amongst other things) that you’d never find a big enough sample size to make any incredibly specific stat we could create statistically significant or reliable.
Essentially wRC (not wRC+) is the stat that best fills your need with any sort of statistical significance. We could make adjustments to it, but then the accuracy/reliability goes to hell.
Interesting reflections. Thanks for the correction concerning the original RC not measuring against the average player. It seems to me that the original RC, my understanding is that it is based not on runs that were actually scored but rather on OBP and total bases, may inflate the perception of RC numbers for a certain player in a given lineup. I realize that such a project would have very limited scope and would not be widely usable, maybe the GM's could have stat men that just sit around and run these particularized formulas for them to try to discern which player would optimize RC in a given spot in the lineup. The formulas might shed some light on who might be best but they would only be useful for a very short time because, as you said, lineups change so readily. If I were a math man I would want to explore the possibility of marketing the idea of particularized formulas to GM's. I guess you would have to do an analysis of FA's and run a bunch of formulas on them to show who might be the best for your lineup. It would be difficult to measure any decision making success based on the formulas however, because the other FA's that you didn't sign in a given situation would end up not playing for you, and there would be no basis for comparison. But such a mathematical assessment that could project which FA might optimize RC in your llineup could still be useful. I think I'm going to run this by my brother in law whose a math genius and see what he says, only problem is he's not too terribly into baseball.
Thanks for your reflections Dave and I hope I'm not annoying you too much by not gutting the idea just yet. I definitely lean in that direction but it is still an interesting question to explore, even from a purely philosophical perspective as it relates to an assessment of abstract vs. more particularized ways of looking at reality.
Even the old RC number is sort of abstract considering it is based not on runs (actual runs scored or batted in) but rather heavily on bases. I understand the need to do this in order to be able to compare one player's stats to the next guy's, and that's great, I just wonder if more particular assessments are possible and whether they could be useful for decisions making in the present moment (ie when a GM is considering signing this as opposed to that FA).
Thinking about this last night I thought maybe such taylor-made formulas could work as ways of advising managers on their batting order. Perhaps it could be used as a way of optimizing RC by rearranging your lineup based on different hitters strengths.
The goal, to me, on offense is to score runs. Hypothetically, a hitter who has an over .400 OBP or an obscene number of total bases is, presumably, going to score and drive in more runs than someone with lessor numbers. However, if that player ACTUALLY only scored 85 runs and drove in 90 runs, with those numbers, RC would overinflate his ACTUAL run production which is the ONLY goal on offense.
Now, maybe he gets stranded often on 2nd or 3rd because he bats 6th in a crappy lineup or he bats behind slow runners who can't score from 1st on a double, whereas on a better team his runs and rbis would increase, but the point is that his batting only produced 175 runs.
So, for my money, a statistic that uses runs+rbis (subtract out HRs)/PAs shows how many runs the player produced per PA. What would then be useful, to compare players from different teams, is to weight that runs produced average against the runs the player's team actually scored. This will not account for where in a lineup a player hits, but I think we could agree that a better hitter will be hitting 2, 3, 4, or 5 in the lineup and there's is little reason to expect that a number 8 hitter would become Babe Ruth just by moving him to 3rd in the lineup. So, I do not believe accounting for where in the lineup a batter hits is very useful because it is unlikely that there are too many undiscovered Ruth's hitting 8th in a lineup whose numbers are artificially kept down because of his spot in the order.
My days as a stats guru are behind me, so I am not espousing anything scientific here, just trying to come up with an idea that promotes a useful statistic that measures a player's ACTUAL contributions to the only goal on offense. Thoughts?
Qbert - Your idea would measure how many runs the player actually created; however, it has virtually no predictive value.
Notify me of follow-up comments via email.