February 17, 200521 yr I know you're trying to say that the statistical approach is accurate and true... But sometimes statistics don't justify a player's value on his team..
February 17, 200521 yr If it's a representation of what he ads to his team, how can it not be? Unless the statistic or team is so poorly created...
February 17, 200521 yr While I too at first strongly disagreed with what the analysis was saying, I have thought through it a bit and they may be on to something. Ok, on the Matsui vs Castillo debate, last year was right in line with Castillo's career averages, maybe his average was on the lower end, but the difference between .290 and .300 is about 6 hits, so it is basically neglible. Castillo being 29 is most likely to repeat his performance, generally 27 is the peak season for players with their careers slowing tapering off after, and speed players tend to taper off quicker than power players. Matsui, despite being 29, is more likely to improve as this will be his 2nd go around in the majors and he will be better adjusted. Castillo is definitely better at D, their offense now favors Castillo but Matsui has a higher upside, so maybe Castillo should be a few VORP runs ahead. However, I think an even greater atrocity occurred at SS, where Gonzalez has a higher VORP than Reyes. Now all of this is predicated on Reyes staying healthy, a big if, but if he does he could be one of the best SS in the NL, possible 30+ VORP, gonzalez is not likely to change his stat lines. At OF, they probably should have used Conine instead of Encarnacion, but it is a toss up and depends on who is healthier. As for the defensive argument, I do not thing the Marlins are that much better than the Mets. It really boils down to our IF is superior to their IF, but basically at 2 positions - 2B and SS, Wright and Lowell are both solid defenders and neither Delgado or Phillips/Piazza are good at 1B. However, in the OF the Mets have a huge advantage with Beltran and Cameron, two of the best CFs manning the OF. This offsets Floyd and is surely better than our non-descript defensive OF, especially when COnine is in the game. Pitching though is the real wildcard in the analysis. Burnett and Beckett both could dominate and be much better than Pedro, or they could mess around like they have been doing and be only good. On the other hand, Pedro could dominate like he should, Peterson could impact Zambrano and Benson, and the MEts could have a feared rotation, but all in all the Fish definitely have the better rotation, i would even say much better. The bullpen is erratic at best for both teams and they both have lots of relievers that could be great or awful, you never know (see Hammond, Chris). The Mets have huge injury risks all across the board, which I am not sure how PECOTA captures, and thus the Marlins are thought to be the better team. HOwever, if all the Mets stay healthy, a huge if, then they are a very dangerous team, which the aforementioned analysis captures.
February 17, 200521 yr If it's a representation of what he ads to his team, how can it not be? Unless the statistic or team is so poorly created... 687947[/snapback] Of course, if you could measure everything that is relevant (meaning it actually affects output) then statistics tell you a whole lot about a player. The point is that what they measure and how they measure is not enough to have a very high degree of confidence in predicting outcomes. That's the point. Is there a statistic that measures Louie working the count and making the pitcher throw 10 pitches just to get an out or, even worse, give up a walk or a base hit? No, all that will show up is that he walked, got a base hit, or made an out. Working the count impacts a game big time, but it just doens't show up in the stats you use. Statistics are only as good as how well they predict outcomes... and in this case, it's not very good.
February 17, 200521 yr Problem is that comparison had bump as one of our bullpen guys and did not include conine instead of enc
February 17, 200521 yr Example A: Jeter, 2004. Example B: Palmeiro, 1999. Case closed. Jury rules in favor of Shamrock. 687813[/snapback] "If the glove don't fit you must acquit."
February 17, 200521 yr I think statistical evaluations like this need to be taken with a grain of salt. In a logical sense, a statistical analysis like this one seems right on. While you may disagree a bit with the way some guys are ranked, for the most part the rankings are pretty much in line with what you would expect. A huge problem with an analysis like this, however, is the fact that the players are given an individual ranking. Like has been mentioned previously, this analysis doesn't take into account the roles of the players. Almost any GM in baseball would tell you we have a better offense than the Mets. Why? Not because of the summation of the individual talents on our team, but because of the way our lineup collectively shakes up. We have speed guys that get on base at the top. Both are perfect top of the lineup guys; they work the count, know how to bunt, move runners over, and create havoc for the opposing pitcher when on base. The value of guys like that in any statistical analysis are greatly underappreciated. A team of Juan Pierre's would probably be pretty bad, but his and Luis' value at the top of our lineup is immeasurable. Then of course we have the meat of our lineup that consists of good hitters who are not big strikeout guys. They know what to do with guys on base. In a nutshell, the reason this analysis is not that accurate is the same reason the Yankee's don't with the World Series every year. Individually, they could represent over half of the AL All-Star team, but you just can't throw every MVP on a team. Every team must have role players, guys that get on base and do things other than just jack the ball out every 15th plate appearance.
February 17, 200521 yr Very true, prin, and that's very apparent when you look at attempts to define a player's defensive value, but a team's goal on offense is to move men around the bases to score runs. That can easily be calculated from the simpliest of statistics (the more compliated statistics put the individual circumstances into explaining how runs are created, and/or how a player fairs against his peers). The so-called little things are outliers, so infrequent and insignificant that they are not even worth the attention. Not that there isn't something to be learned from examining those little things, maybe even something you can turn into a winning strategy or might be the difference between two equal teams, but the chances are likely that it will have not have as significant as many baserunners and extra-base hits.
February 17, 200521 yr Shamrock, enlight me to why Matsui may be a little better than Castillo? 687750[/snapback] Age 28 seasons for both with Matsui posting a slightly better OPS in his first ML season. Conventional wisdom says Matsui improves (and so VORP says) on his season while Castillo would do well just to repeat. Castillo has the edge in terms of SB and defense (although GGs are overrated) as it stands. 687757[/snapback] What about expierience and durability
February 17, 200521 yr That s*** is complete bull!!!! Excuse me while i laugh but Lowell and Wright tied? what the hell has wright done in the MLB, he played a couple of games toward the end of the year and now he is tied with Lowell thats Bull. so now what they are saying is that Wright has an incredible upside and this and that. so if we are going by potential then lets just put Hermida in there and put him with 30 pts. Also AJ around 24, again if we go with potential like they did with Wright then AJ should top all pitchers, and Perdo at 50 LOL!!! Also if Castillo is better than Matsui IMO but if they say Matsui is better it aint by that much of a difference that they have. Also Bump is not in our rotation
February 17, 200521 yr I think statistical evaluations like this need to be taken with a grain of salt. In a logical sense, a statistical analysis like this one seems right on. While you may disagree a bit with the way some guys are ranked, for the most part the rankings are pretty much in line with what you would expect. A huge problem with an analysis like this, however, is the fact that the players are given an individual ranking. Like has been mentioned previously, this analysis doesn't take into account the roles of the players. Almost any GM in baseball would tell you we have a better offense than the Mets. Why? Not because of the summation of the individual talents on our team, but because of the way our lineup collectively shakes up. We have speed guys that get on base at the top. Both are perfect top of the lineup guys; they work the count, know how to bunt, move runners over, and create havoc for the opposing pitcher when on base. The value of guys like that in any statistical analysis are greatly underappreciated. A team of Juan Pierre's would probably be pretty bad, but his and Luis' value at the top of our lineup is immeasurable. Then of course we have the meat of our lineup that consists of good hitters who are not big strikeout guys. They know what to do with guys on base. In a nutshell, the reason this analysis is not that accurate is the same reason the Yankee's don't with the World Series every year. Individually, they could represent over half of the AL All-Star team, but you just can't throw every MVP on a team. Every team must have role players, guys that get on base and do things other than just jack the ball out every 15th plate appearance. 688101[/snapback] Very true. When discussing a team, or an aspect of a team you have to look at its sum parts. There are many strategies to win. Not every team needs Pierre or Castillo (who has given the Marlins the best leadoff on-base percentage between 2000-04) to be an offensive juggernaught. A team that can take walks and avoid outs to make that 15th plate appearances more frequent. A team that uses platoons to increase their chances to win every game no matter the opposing starting pitcher. A team that takes extra bases liberally. As long as you have the personnel and the circumstances are right there's no wrong strategy.
February 17, 200521 yr Very true, prin, and that's very apparent when you look at attempts to define a player's defensive value, but a team's goal on offense is to move men around the bases to score runs. That can easily be calculated from the simpliest of statistics (the more compliated statistics put the individual circumstances into explaining how runs are created, and/or how a player fairs against his peers). The so-called little things are outliers, so infrequent and insignificant that they are not even worth the attention. Not that there isn't something to be learned from examining those little things, maybe even something you can turn into a winning strategy or might be the difference between two equal teams, but the chances are likely that it will have not have as significant as many baserunners and extra-base hits. 688109[/snapback] I'd like to see a regression done of all the stats that BP uses to things such as runs produced, wins, etc., and see how large their error range is. I imagine it is large enough to make them statistically insignificant. Almost guaranteed to be >.05.
February 17, 200521 yr That s*** is complete bull!!!! Excuse me while i laugh but Lowell and Wright tied? what the hell has wright done in the MLB, he played a couple of games toward the end of the year and now he is tied with Lowell thats Bull. so now what they are saying is that Wright has an incredible upside and this and that. so if we are going by potential then lets just put Hermida in there and put him with 30 pts. Also AJ around 24, again if we go with potential like they did with Wright then AJ should top all pitchers, and Perdo at 50 LOL!!! Also if Castillo is better than Matsui IMO but if they say Matsui is better it aint by that much of a difference that they have. Also Bump is not in our rotation 688128[/snapback] Wright was a stud in the minors and very good in the majors. Lowell slipped a bit from a career 2003 year and is past 30. As has Castillo from his earlier career numbers and style of game does not translate well into future success in their system as well as Matsui's. A.J. is hurt by lost playing time. Pedro helped by the move to the NL. Why as to why Bump was included for the bulllpen instead of Alfonseca or Spooneybarger, I don't know. Probably because of Alfonseca's unpredictable career track and Spooney's injury, I doubt Mets fans expect Bottalico or DeJean to do better than Orber Moreno. You can learn more about it here.
February 17, 200521 yr I'd like to see a regression done of all the stats that BP uses to things such as runs produced, wins, etc., and see how large their error range is.? I imagine it is large enough to make them statistically insignificant.? Almost guaranteed to be >.05. 688133[/snapback] Many of BP's stats are more niche statistics that examine trends in existing statistics, but VORP, like all runs created stats, is based on run expectancy drawn from real major league games. It is a measurement of run production, the amount of runs a player contributes over a replacement-level player. As you may know, run differential (runs scored- runs minus) as a strong correlation with winning percentage. Almost as much as OPS FOr - OPS Against.
February 17, 200521 yr I'd like to see a regression done of all the stats that BP uses to things such as runs produced, wins, etc., and see how large their error range is.? I imagine it is large enough to make them statistically insignificant.? Almost guaranteed to be >.05. 688133[/snapback] Many of BP's stats are more niche statistics that examine trends in existing statistics, but VORP, like all runs created stats, is based on run expectancy drawn from real major league games. It is a measurement of run production, the amount of runs a player contributes over a replacement-level player. As you may know, run differential (runs scored- runs minus) as a strong correlation with winning percentage. Almost as much as OPS FOr - OPS Against. 688174[/snapback] Yes, I know all of that. But how well do these stats predict things? It's easy to look back at things and see a strong correlation - like runs scored v. runs given up. Those are easy. The question is how well can they predict actual outcomes. Based on the PECOTA stats I just saw (and that you just referred us to), not very well at all. The 25th to 75th percentile range for most players is huge and for almost all the players I saw they were way off (in fact, I can't see a single one where this wasn't the case). For example, Miguel Cabrera's predicted 25th percentile numbers for 2004: .243 12 HR .394 SLG. 50th percentile: .259 14 HR .428 SLG. 75th percentile: .283 20HR .482 SLG. This is WAY off. Basically, Miguel's performance last year was like 99th percentile of his expected production. I would say that all the stats that predict players' performances do not have a high degree of confidence. I doubt they're statistically significant. In fact, I would be shocked if it were. Does it give you a general ballpark range but with a high "but it's possible that this might happen"? Yes, but you can do that without the stats.
February 17, 200521 yr How many projections had Cabrera hitting .300, 30 homers and 100 RBIs at age 21?
February 17, 200521 yr You can't judge BP stats on how they did with Miguel Cabrera because Miguel Cabrera is such a rare player, few have done what he did at such an early age, I dont think anybody saw him having that good of a year. Regardless, here are the predictions in 2004 for the Marlins regulars Predicted vs Actual Mike Lowell .276/.351/.476 vs .293/.365/.505 Jeff Conine .263/.331/.407 vs .280/.340/.432 Luis Castillo .285/.357/.369 vs .291/.373/.348 Alex Gonzalez .256/.315/.403 vs .232/.270/.419 Juan Pierre .299/.355/.380 vs .326/.374/.407 Paul Lo Duca .271/.331/.400 vs .286/.338/.421 Difference -0.017/ -0.014/ -0.029 -0.017/ -0.009/ -0.025 -0.006/ -0.016/ 0.021 0.024/ 0.045/ -0.016 -0.027/ -0.019/ -0.027 -0.015/ -0.007/ -0.021 % Different -5.80%/ -3.84%/ -5.74% -6.07%/ -2.65%/ -5.79% -2.06%/ -4.29%/ 6.03% 10.34%/ 16.67%/ -3.82% -8.28%/ -5.08%/ -6.63% -5.24%/ -2.07%/ -4.99% Predicted vs Actual Cumulative (average) .275/.340/.406 vs .285/.343/.422 So I think the predictions were fairly close, i mean only .003 off on cumulative OBP, so are the predictions perfect? No, but they are as good as anything out there.
February 17, 200521 yr That s*** is complete bull!!!! Excuse me while i laugh but Lowell and Wright tied? what the hell has wright done in the MLB, he played a couple of games toward the end of the year and now he is tied with Lowell thats Bull. so now what they are saying is that Wright has an incredible upside and this and that. so if we are going by potential then lets just put Hermida in there and put him with 30 pts. Also AJ around 24, again if we go with potential like they did with Wright then AJ should top all pitchers, and Perdo at 50 LOL!!! Also if Castillo is better than Matsui IMO but if they say Matsui is better it aint by that much of a difference that they have. Also Bump is not in our rotation 688128[/snapback] Wright was a stud in the minors and very good in the majors. Lowell slipped a bit from a career 2003 year and is past 30. As has Castillo from his earlier career numbers and style of game does not translate well into future success in their system as well as Matsui's. A.J. is hurt by lost playing time. Pedro helped by the move to the NL. Why as to why Bump was included for the bulllpen instead of Alfonseca or Spooneybarger, I don't know. Probably because of Alfonseca's unpredictable career track and Spooney's injury, I doubt Mets fans expect Bottalico or DeJean to do better than Orber Moreno. You can learn more about it here. 688147[/snapback] [/quot i what did pecota had to said about the mets last year?? and the year before and the 2002 with mo,jr ,cedeno and burnitz,glavine etc
February 17, 200521 yr How many projections had Cabrera hitting .300, 30 homers and 100 RBIs at age 21? 688193[/snapback] That's exactly my point. The projections are simply not within the margin of error... they're statistically insignificant. They do a decent job of predicting, but that's it. Let's not look at these statistics and pretend they're somehow more valuable than using common sense (i.e. looking at numbers) and watching the players perform, etc. It's better to use these statistics in combination with observation and knowledge of the game. Simple as that. Also, if you look at the PECOTA, it was wrong about a lot of people, not just Cabrera. Cabrera is not an outlier.
February 17, 200521 yr Don't deny the facts sabermetrics spit out, even as dumb as they might sound. Trust the numbers and not your eyes. Numbers don't lie, your eyes tend to. BWAHAHAHAHAHAHAHA!
February 17, 200521 yr Don't deny the facts sabermetrics spit out, even as dumb as they might sound. Trust the numbers and not your eyes. Numbers don't lie, your eyes tend to. BWAHAHAHAHAHAHAHA! 688411[/snapback] I think your acting like a moron, honestly. Nobody here is saying that. What we are saying is that relying on your subjective view of player performance does not provide a full picture and neither is it completely irrelevant (although, I wasnt aware of your scouting background). A combination of the 2 approaches is the best we can do at this point.
February 17, 200521 yr Don't deny the facts sabermetrics spit out, even as dumb as they might sound. Trust the numbers and not your eyes. Numbers don't lie, your eyes tend to. BWAHAHAHAHAHAHAHA! 688411[/snapback] I think your acting like a moron, honestly. Nobody here is saying that. What we are saying is that relying on your subjective view of player performance does not provide a full picture and neither is it completely irrelevant (although, I wasnt aware of your scouting background). A combination of the 2 approaches is the best we can do at this point. 688425[/snapback] I was being sarcastic. :plain
February 17, 200521 yr i dont believe this "article", if it deserves being called that, is very reliable. i question some things about what it says.
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.