What is your opinion of the A.J. Pierzynski signing?
MJH on accountability
A discussion was begun at the end of the Jim Thome-to-Texas Redux thread about run creation, and in particular runs per at bat. I thought about it and begun wondering whether a super runs number might be a good offensive statistic as a compliment to OPS. This is what I came up with.
How about this guys for an ultimate runs number?
1) You take RBI's and adjust it to subtract the number of times that you drove yourself in.
2) Then you take your runs number.
3) Then you take the adjusted RBI number
4) Add it to the runs number
5) Divide it by the number of at bats
6) Then you have a Runs Created Per At Bat percentage. RCAB?
How is this as a statistic to measure offensive output and to compliment OPS? It accounts for many different factors including speed on the base paths and base running abilities, and not only power, hitting, timely hitting, and OBP. It would, however, be inflated if you hit in front of awesome offensive bats. I subtract the number of times you drive yourself in from the RBI number because you would be double counting runs if you didn't.
Does anyone know where you find a breakdown of that number into the number of times you drove yourself in as opposed to driving others in? That would enable me to actually run the numbers on some guys like Hamilton, Andrus, Young, and Thome.
What sayest thou stat men? Has someone already thought of this? Is it a stupid idea?
Hamilton's number WITHOUT THE ADJUSTMENT is (anyone know where I can find the data to make the adjustment)
.376 runs per at bat and OPS of 1.044
.387 OPS 1.039
.289 OPS .774
.209 OPS .643
.232 OPS .649
Interesting how Borbon's number is so much higher than Andrus' considering their comparable OPS.
I really would like the feedback of some of the guys more knowledgeable about statistics as I have only been thinking in depth about this stuff for a short time. Is this dumb for some reason I'm not thinking of? Or is it already a commonly used statistic?
The Borbon/Andrus discrepency is even more interesting considering Elvis led the league in BA/RISP. His hitting was very timely.
You would need to subtract runs scored as a pinch runner, however, since these are essentially "free" runs (a run without an AB). Also, you shoud use Plate Appearances, not ABs. That might be why Thome's # is higher than Josh's- Thome is one of the League's best in BB%
I agree, PAs should be used instead of ABs. That will help even out the difference between guys like Nick Johnson, who tries to walk, and guys like Vlad. I'm not sure if there is an equivalent measure out there but it is interesting to try and quantify how many runs a player is responsible for.
WOW, major brain fart. Sorry guys.
The number of Home Runs is the number of times you drove yourself in.
I will repost new numbers when I have time for those guys: Hamilton, Thome, Young, Andrus, and Borbon.
Thanks, good thinking about PA's instead of AB's.
Or thanks to Scooby rather who thought of it first. And good thinking on the Pinch Runner idea. I wonder if this effected Borbon's number being significantly higher than Andrus'.
I calculated the Runs Created numbers based on dividing RBI's plus Runs (minus HR bc they are already counted in RBI's) by Plate Appearances to give a Runs Created per Plate Appearance.
I am also including the OPS number next to the RC numbers
The numbers are based on 2010 except for Gentry's which are career numbers.
Andrus .182 .643Young .235 .774Thome .241 1.039Hamilton .285 1.044Borbon .211 .649Gentry .222 .422 (career numbers)Beltre .246 .919
If we say that an average game includes 4 PA's Hamilton would create 1.14 runs a game, Thome .964, and a Gentry/Borbon platoon around .8. The difference between Thome and a CF platoon over 100 games at the plate would be about 20 runs. This is based very strongly on the number of RBI's and runs scored from last year for the players involved. So maybe the Gentry/Borbon platoon would be about average or slightly below, while a Thome bat would be significantly above average. It is just interesting to see 20 runs over 100 games as significantly above average. But think of it this way: lets say in just 5 of those 100 games Thome got one of those 20 runs, and that game was won by 1 run. That means in five games won by one run out of the 100 that Thome played in he won the game. In other words, for those five games he was the factor without which the game may easily have not been won.
What this showed me is the very strong dependence that each player in the batting lineup has on the rest of the lineup. Even someone like Hamilton only drives in or scores about one third of a run more per game than something like a Gentry/Borbon platoon in CF. Fascinating, I think looking at these numbers, superficially, it is easy to say that a big bat is worth less than one imagines, but when you consider how much those 30 runs over 100 games count (in the case of Hamilton) it changes your perspective. A game of inches indeed.
I calculated Ichiro's and it was only .151, even lower than Andrus'. I think this is because they are both leadoff hitters and leadoff hitters suffer in this category because so many of their hits, walks, sacrifices, etc. happen when no one is on base and thus their RBI totals suffer and they are usually not power hitters so they dont make up for it in HR's. Run production is a very strange thing but if you look at hitters that don't lead off there is a significant differential from hitters like Borbon and Gentry to others like Beltre, Thome, Young and Hamilton. From these numbers it does look like Thome's speed may well hinder him from scoring as many runs as Hamilton even though they have very similar OPS numbers.
I wonder how many times a pinch runner scored a run after Thome got on base late in the game in 2010? Also, there's the fact that Target Field is a pitcher's park, so he was probably driven home less often for that reason as well.
That is interesting. I couldn't figure out a way to adjust for pinch runs (which could easily effect Borbon's numbers), and you make an excellent point that he really should get some credit for the runs that score when he was pinched for as well. I just don't know how significant a difference that would be. Probably pretty minimal, it might bump Thome's percentage up to .250, and Borbon's down a bit? Hard to know. Any interest in running the numbers? I just couldn't find them.
I posted the below in the other thread, DID THOME REALLY SIGN WITH MINNESOTA FOR 3 MILLION? WHAT THE HECK?! Well, these last few conversations have me questioning Thome's overall value anyways, but not sure if these numbers make all that much sense considering how much they are dependent on how many times other guys drive you in.
The runs created number I talk about in the other forum thread puts Thome at about .245/PA and Young is at about .235. Hamilton is up at .265. So, considering Hamilton and Thome do have very similar OPS numbers it may well be that Thome's lack of speed hinders him from scoring a significant number of runs, up to .02 per plate appearance, .08 per game (at 4 PA a game), and about 10 runs over 100 games (and remember this is in spite of how many walks he gets). Thome would score about 18 (this is significant) more runs over 100 games than a Borbon/Gentry platoon and Hamilton would score about 26 more (8-10 more than Thome) runs over 100 games than a Borbon/Gentry platoon.
Do those 18 runs make you think more soberly about Thome's RC numbers, at least considering he has been compared to Hamilton's offensive output?
It seems like the fact that Andrus' RC/PA number is .03 better than Ichiro's makes him actually look pretty good? Nice? Thats an extra 12 runs per 100 games if we multiply .03 by 4 PA's per game. Of course that could be largely determined by the fact that the Rangers have such a better offensive lineup and thus Andrus will have more runs than Ichiro considering the Mariner's poor offense.
Its true, Thome signed with the Twins for 3 million. Wow.
Thome signing probably wasn't about the money. The guy has made plenty over his career. I believe he felt he fit in better with Minnisota.
I guess he just wanted to play there all along and was using the Rangers as leverage.
One of the obvious choices for RC/PA numbers I didn't do was David Murphy. He had a .227 RC/PA percentage. That means in a 4 PA game he would create .908 runs. Over 100 games Murphy would create about 90 runs while Thome would create between 95-100. Of course that is with Thome hitting in the Twins lineup, he would probably have had better RC numbers in the Rangers lineup. However, its not like Murhpy was hitting in front of Josh Hamilton as MY was last year, although he did hit in Hamilton's spot near the end of the year. All in all, I am beginning to think Thome, considering our offensive depth in Murphy and Moreland (a guy who will probably get a lot more at bats now), would not have been as much of a runs upgrade as many thought by looking at his OPS number from last year. Think about, for example, Moreland, and how he walked and walked in the playoffs, and he would have speed on the base paths unlike Thome.
Considering the additional benefit of 25th man roster flexibility as well as the potential of a Gentry/Borbon CF platoon to improve defense in Center and in Left (with already awesome defense in Right, and the big three infield positions, as well as a defensive upgrade at 1st with Moreland over MY), I am not nearly as torn up about Thome going to the Twins as I otherwise would be.
However, there is the one issue that others have brought up. Thome would have been an awesome add for big bat depth in the event of injury. Considering this I hope that Wash takes a very aggressive approach to resting Cruz and Hamilton at DH, and giving the big three infield defenders time off with MY backing up 2B and 3B and Blanco backing up Andrus for at least 10-15 games. This will give Wash the opportunity to rest Ham and Cruz for 30 games at DH, and another 20-30 games can easily be stolen from Moreland at 1st with Young filling in there defensively. It will help the rookie to develop while not having to face tough lefties.
Rest, rest, rest. Our injury prone stars need it need it need it.
Joe - essentially the problem with your idea is that it's incredibly dependent upon the surrounding talent. It's pretty much impossible to accurately compare anyone from different teams. Bill James already has a stat for runs created that's calculated as a product of OBP and total bases:
RC (runs created) - OBP * TB
You could divide that stat by PAs to get a rate to use as a comparison.
Josh Hamilton: .411(OBP) * 328(TB) = 134.808(RC)134.808(RC) / 571(PA) = .236 (RC/PA)Jim Thome: .412(OBP) * 173(TB) = 71.276(RC)71.276(RC) / 340(PA) = .209 (RC/PA)
However, you might find that using a formula to calculate runs created that's based upon weighted on-base average gives you a more accurate picture of how good the player actually was with the bat.
wRC – This is total runs created based of wOBA. It is calculated as (((wOBA – lgwOBA) / wOBAScale) + (lgR/PA)) * PA
wRC+ - Is basically a league/park adjusted version of wRC that's converted to a scale that's similar to OPS+ (100 is average).
wRC+ is an easy stat to compare players to each other and to see if a player is performing poorly or amazingly well. Looking at wRC+, here are some Rangers' wRC+ values for 2010:
Michael Young - 106Elvis Andrus - 81Vladimir Guerrero - 124Josh Hamilton - 182David Murphy - 122Julio Borbon - 76Ian Kinsler - 122Nelson Cruz - 156Matt Treanor - 61Bengie Molina - 59Andres Blanco - 80Mitch Moreland - 122
Here are the wRC+ values for the top 12 hitters by WAR (plus Jim Thome and Manny Ramirez):Josh Hamilton - 182Joey Votto - 177Albert Pujols - 169Ryan Zimmerman - 146Adrian Beltre - 143Jose Bautista - 169Evan Longoria - 140Matt Holliday - 153Carl Crawford - 142Troy Tulowitzki - 150Robinson Cano - 145Miguel Cabrera - 172Jim Thome - 180Manny Ramirez - 141
Joe - I forgot to add that Hamilton averaged 4.29 PAs per game and Thome averaged 3.148 PAs per game. If you used Bill James's RC stat, that would make Hamilton worth 1.013 runs/game and Thome worth .660 runs/game.
Thanks for this, very interesting.
Here is my observation. Bill Jame's numbers (is he at fangraphs? dont know much about it, seems like some kind of a math genius) are obviously good at measuring one player against another player, and excellent at giving you a reference point for measuring how good a player is as compared to the league average player, whether they created less or more runs than what is considered average. I have no reason to doubt the importance and accuracy of these numbers based mainly on players, and how good player a is vs player b.
Lets see if I make sense when I say this, its sort of a hard idea. It seems that in order to get a statistic that allows you to compare player x vs player y you have to abstract a great deal from the particular circumstances of any given situatoin. And this is desirable in order to reach a statistic that measures a players place in the grand scheme of baseball in general. Basically, James' statistic measures player x against all offensive players by abstracting what is considered average based on all offensive numbers, and then comparing player x's numbers to find out how good they are.
Like I said that is a great thing for finding out how good player x is as compared to all players (the average player).
But what if you want an analysis that does not have to do with all players but rather measures the impact of player x on team x in a given year, in a given park, in a given order. This is the sort of statistic that is considered so inferior by the new stat men. Am I right? And no doubt they are right that it is inferior according to what they are trying to get out of a statistic, which is the ability to compare a player against other players and to find out as compared to all players (average) how good they are, as a player abstracted from particulars (team, park, where they hit in the lineup, etc.) This is obviously valuable and I have no reason nor any desire (and not to mention no expertise) to question it.
But when I was considering these RC numbers and comparing Thome's to others such as Gentry and Borbon (two radically inferior players by any account) I found that it might be valuable to try to measure, not Thome's relative run production as compared to the average player or all other players, but rather his specified and particularized impact on the Rangers run production in the Rangers lineup with all of the many factors that go into that. Such a study, and I am not saying at all that I achieved this (I think it would take a lot more math work and thinking to really try to measure this with any value), would tell a particular team constituted by all its complex factors, the run production impact that player x would have on team x's run production considering the realities that make up team x. In other words, when team x is going after player x the question is not so much how good player x is as compared to all players, but rather, how good player x would be for team x given team x's current lineup. Further, it could be valuable to compare such a number to players that the team already has in order to measure just how much better player x is going to be for team x. Of course, a lot of this is impossible to measure considering the fact that player x has not yet played on team x, and player x's stats would have to be derived from his production on team y (his previous team). But for a player like Thome (or any veteran really) it would be perhaps valuable to plug in their career numbers as over 19 years they have amassed such a diversity of particulars into their average numbers that it might not be a completely futile exercise. In short, basically when you calculate his career number you get such a generalized number based on his abilities in so many different particulars. By plugging the career numbers in to the lineup of team x's year before numbers (in this case 2010) you can get an idea of how well player x would produce in the given lineup and how well player y (who you already have on your team) would compare. A further necessary problem might be the fact that people hit at different spots in the order nullifying any real comparability value, but if all you want is to do is compare the run production that you would get from Murphy (who will likely hit in the 7 hole if he DH's) and Thome (who would have likely hit in the 4 hole), I think you can do that by comparing their RC numbers as I conceived of them because the numbers have to do with run production on your team and in your lineup. I don't know whether this is true but it seems that Thome's impact (perhaps because of his speed problems) would have been less in our lineup than someone with a similar OPS like Hamilton. It seems. considering our offensive depth in Moreland and Murphy that Thome's bat may not have been as big as an impact than one might think when one is looking at Thome's numbers as comapred to all others players (which is the James number). I think there are a lot of variables that would need to be worked out in order to get a meaningful comparison concerning RC between Free Agent players and players already on the team. Further, such a comparison would be completely particularized and thus only useful for that particular situation, the situation in which one decides whether to sign a given player. I don't have the math skills to come up with a comprehensive formula that might be able to make such a comparison in a meaningful way in a particular situation, but I do think that this might be a worthy pursuit for some math whiz. In short, we think a great deal about how player x compares to all players (the average player) but perhaps we do not do enough thinking about how player x free agent fits on team x in team x's lineup as compared to other available players already on team x. Call it a hunch based on the very rough and rudimentary numbers that I worked out, but I think a DH of Murphy or similar in 60-80 games against RHP might not be that much worse than a DH of Thome.
So of course I concede that what I originally set out to do, which was to measure a player's RC against all other players or to give an idea of how good they are vs other players (and vs the average player or the good player), was futile and practically useless to that end as you said. However, perhaps there is something to be gained in this discussion, basically this, "it is not so much how much better player x is than the average player but rather how much run production player x would give team x as compared to other players on team x." To make that calculation one would need to make a statistic about player x very dependent on the particular circumstances that make up team x's run production, basing RC number in this scenario on something like what I came up with originally might work.
Ok, I think I got my point across in a pretty repetitive way, but hopefully it makes sense. Maybe these sorts of statistics already exist anyway because GM's probably hire stat men to create these sorts of formulas for them because the question for any given GM going after any given free agent is not so much how good a given player is as compared to the average player but rather how much will a player impact my team. This sort of deliberation goes on constantly in non-scientific ways. Might it also be done, and is it done, in more mathematical ways? Or are there just too many particular factors to really measure a RC impact on a particular team based on the contributions of a particular player who has yet to play on team x. Food for thought or trash cans if nothing else.
Any feedback from people who know more about this stuff than me would be appreciated as all I am really doing is reasoning about these things with little particular or practical knowledge about statistics (which is ironic considering the content of this post being about getting at particularized impacts).
Joe - Bill James actually works for the Redsox these days. RC is his own invention and is actually very simple to calculate (OBP * TB). On base percentage and total bases (OBP and TB respectively) are easily found on baseball-reference.com. wRC+ is readily found on Fangraphs.com and seems to be the stat that you don't find as useful for your needs. James's RC formula accounts for getting on base and advancing. It still accounts for some of the actions of teammates but isn't as dependent upon an extremely high number of variables as your version of RC.
Certainly your overall goal would require some extremely complicated statistical analysis, but I think you'll find that some of the existing metrics help. If you're in need of further assistance you could solicit more info. from BBTiA's own David Brown (firstname.lastname@example.org) or Prashanth Francis or from Baseball Prospectus's Colin Wyers (http://www.baseballprospectus.com/contact.php (Colin Wyers) or @cwyers on twitter).
Thanks Dave H.
The math whizzes loathe the use of RBIs to measure production. Juan Gonzalez won 2 MVPs based largely on his RBI numbers, which could be obscene at times in the offenses he was a part of. I saw Juan Gone more than once swing at pitches a foot outside (no exaggeration) with a man on 2B and poke the ball into right field. He absolutely refused to take the open base and miss an opportunity to add one to his RBI total.
In '96 and '98, Gonzalez led the AL in RBIs and brought home the MVP trophy, too. The math geeks will tell you Juan wasn't an MVP either yr. In fact, in '96, he finished 5th among AL right fielders (!) in WAR, and then 3rd among AL right fielders in '98. Juan was obviously staring at a lot of guys standing on 2B those 2 seasons, and you have to ask why he would get more credit for hitting an RBI double versus a bases-empty double.
If your team OBP is .338 like the Good Guys in 2010, you are probably going to get RBI opportunities more frequently than a player on another team whose OBP is .312 (Toronto). This showed to be true, as Guerrero had 100 more PAs with men on base than Jose Bautista, with nearly equal playing time. That's also a major reason why Texas outscored Toronto despite losing the HR battle by a huge margin at 257 to 162.
Baseball Reference finds 50 instances in history where a player had a game with 10 total bases, but only 1 RBI. Are you saying a guy who strikes out 4 times with runners on but finally manages a weak RBI ground out gets equal credit for his performance?
Notify me of follow-up comments via email.