|
New Fangled Statistic Talk
|
|
|
Posted: 12/7/2012 7:32 AM
New Fangled Statistic Talk
Thought per other threads that we might have a thread on new statistics. I think they are interesting especially when evaluating relative value of one player to another. I know that Cardinals use these statistics extensively to evaluate prospects and I assume trades.
|
|
Reply |
Quote |
|
|
Posted: 12/7/2012 7:43 AM
Re: New Fangled Statistic Talk
Quote is from Fangraphs Glossery.
Many people try to look at defense and end up saying all flawed, but I saw Player x blow and easy play, so they suck. That seems flawed also. What do people think about UZR?
Ultimate Zone Rating (UZR) is one of the most widely used, publicly available defensive statistics. The theory behind UZR is tougher to intuitively grasp than Defensive Runs Saved (DRS), but the simplified version is that UZR puts a run value to defense, attempting to quantify how many runs a player saved or gave up through their fielding prowess (or lack thereof). There are a couple different components to UZR, including: ● Outfield Arm Runs (ARM) – The amount of runs above average an outfielder saves with their arm by preventing runners to advance.
● Double-Play Runs (DPR) – The amount of runs above average an infielder is in turning double-plays.
● Range Runs (RngR) – Is the player an Ozzie Smith or an Adam Dunn? Do they get to more balls than average or not?
● Error Runs (ErrR) – Does the player commit more or fewer errors compared with a league-average player at their position? The run values in each of these categories are then compiled into one overall defensive score, UZR. Since UZR is measured in runs, it can be compared easily with a player’s offensive contributions (wRAA). For the details on how UZR is calculated — i.e. how we can attach a run value to defensive events — see the FanGraphs UZR Primer. |
|
Reply |
Quote |
|
|
Posted: 12/7/2012 11:32 AM
Re: New Fangled Statistic Talk
As of this time, I'm not too comfortable with them. Offensive or defensive. I don't believe in magic bullets, if you will.
I remember the warning that went with using a coeffecient of determination (R²) (applied to correlation, r too) was the more independent variables you introduce to the equation, the larger the R²/lrl. Regardless of relevance to the expression. i.e., you get a number that "looks better than it is".
|
|
Reply |
Quote |
|
|
Posted: 12/7/2012 11:50 AM
Re: New Fangled Statistic Talk
MagnoliaCardFan wrote:
stlcards5129 wrote:
MagnoliaCardFan wrote: ...
...There's literally tons of literature discussing WAR. If you're really that interested in it go here.
.... Let me approach you this way: In 2012 Matt Holliday had 177H/599AB. Hence he had a BA of .295.
He also had a WAR of 3.9. Would you please show me the calculation for that? And then I can ask my next question... (33.7 - 10.5 + 0 - 7.3 + 22.9)/10 = 3.88EDIT: that's the simple calculation, I prefer Fangraphs calculation, but that's what you asked for. 33.7? 10.5? 0? 22.9? 10? |
|
Reply |
Quote |
|
|
Posted: 12/7/2012 12:42 PM
Re: New Fangled Statistic Talk
I'm hoping to understand them better. I don't believe that any stat is going to give you an answer Player A better than Player B, but I wanted to better understand what each one means and how they can be used to evaluate a player. I'm old so I'm still dependent on BA/RBI/HR and this guy has so many errors. MagnoliaCardFan wrote: As of this time, I'm not too comfortable with them. Offensive or defensive. I don't believe in magic bullets, if you will.
I remember the warning that went with using a coeffecient of determination (R²) (applied to correlation, r too) was the more independent variables you introduce to the equation, the larger the R²/lrl. Regardless of relevance to the expression. i.e., you get a number that "looks better than it is". |
|
Reply |
Quote |
|
|
Posted: 12/7/2012 12:53 PM
Re: New Fangled Statistic Talk
MagnoliaCardFan wrote: As of this time, I'm not too comfortable with them. Offensive or defensive. I don't believe in magic bullets, if you will.
I remember the warning that went with using a coeffecient of determination (R²) (applied to correlation, r too) was the more independent variables you introduce to the equation, the larger the R²/lrl. Regardless of relevance to the expression. i.e., you get a number that "looks better than it is". How is linear regression relevant when discussing baseball metrics? I'm not aware of any often used metrics that use linear regression... Really? The more variables you regress the less sure you're going to be about your regression? NO WAY! UNBELIEVABLE!
Last edited 12/7/2012 1:00 PM by stlcards5129
|
|
Reply |
Quote |
|
|
Posted: 12/7/2012 12:59 PM
Re: New Fangled Statistic Talk
MagnoliaCardFan wrote:
MagnoliaCardFan wrote:
stlcards5129 wrote:
MagnoliaCardFan wrote: ...
...There's literally tons of literature discussing WAR. If you're really that interested in it go here.
.... Let me approach you this way: In 2012 Matt Holliday had 177H/599AB. Hence he had a BA of .295.
He also had a WAR of 3.9. Would you please show me the calculation for that? And then I can ask my next question... (33.7 - 10.5 + 0 - 7.3 + 22.9)/10 = 3.88EDIT: that's the simple calculation, I prefer Fangraphs calculation, but that's what you asked for. 33.7? Batting-10.5? Fielding0? Baserunning
-7.3 Position22.9? Replacement10? Runs to Wins divider Here's my problem. I've literally given you a goldmine of information for the question's you've asked. If you had any real interest in understanding the metric you would have already gone and read up on it...years ago. I think your agenda is to try to discredit it. |
|
Reply |
Quote |
|
|
Posted: 12/7/2012 1:41 PM
Re: New Fangled Statistic Talk
stlcards5129 wrote:
MagnoliaCardFan wrote: As of this time, I'm not too comfortable with them. Offensive or defensive. I don't believe in magic bullets, if you will.
I remember the warning that went with using a coeffecient of determination (R²) (applied to correlation, r too) was the more independent variables you introduce to the equation, the larger the R²/lrl. Regardless of relevance to the expression. i.e., you get a number that "looks better than it is". How is linear regression relevant when discussing baseball metrics? I'm not aware of any often used metrics that use linear regression...
Really? The more variables you regress the less sure you're going to be about your regression? NO WAY! UNBELIEVABLE! If you've got numbers (w/rank/bias: eg, NOT the registration numbers of race entrants) ...then you have linear relationships. Baseball, hydraulics, sales figures, time relationships, etc.
Way. Note that the addition of independent variables NEVER decreases the value of R².
Now we're getting off on a tangent. Of a tangent.
|
|
Reply |
Quote |
|
|
Posted: 12/7/2012 1:44 PM
Re: New Fangled Statistic Talk
I'm a long time SABR guy and have read many articles using advanced metrics in comparing and evaluating players, teams, etc.
I like the research that's been done and think that many, if not most, of the latter day created statistical methods can be insightful. However, most of the time when you get down to it, on offense you usually come to the same conclusion on a player's value by using the old BA-HR-RBI numbers as you do OPS and other things. With pitchers you can look at IP, Hits allowed, K, BB and ERA and have just about all you need to assess effectiveness.
Defensively is where the advanced metrics really help in my opinion.
At the same time, the game is still played by humans, and the human element will always have a degree of randomness. In other words, we can dig for a cause-effect relationship based on numbers, and sometimes it won't work.
So I think that as long as users of data keep in mind that we'll never be able to break the sport down into a perfect numerical based analytical process, then the more research a person wants to do attempting to find correlations, the better.
|
|
Reply |
Quote |
|
|
Posted: 12/7/2012 1:45 PM
Re: New Fangled Statistic Talk
stlcards5129 wrote:
MagnoliaCardFan wrote:
MagnoliaCardFan wrote:
stlcards5129 wrote:
MagnoliaCardFan wrote: ...
...There's literally tons of literature discussing WAR. If you're really that interested in it go here.
.... ... (33.7 - 10.5 + 0 - 7.3 + 22.9)/10 = 3.88EDIT: that's the simple calculation, I prefer Fangraphs calculation, but that's what you asked for. 33.7? Batting-10.5? Fielding0? Baserunning
-7.3 Position22.9? Replacement10? Runs to Wins divider Here's my problem. I've literally given you a goldmine of information for the question's you've asked. If you had any real interest in understanding the metric you would have already gone and read up on it...years ago. I think your agenda is to try to discredit it. I think I have a feel for the concept. eg: I want to know how you calculated "33.7". |
|
Reply |
Quote |
|
|
Posted: 12/7/2012 3:01 PM
Re: New Fangled Statistic Talk
MagnoliaCardFan wrote:
stlcards5129 wrote:
MagnoliaCardFan wrote: As of this time, I'm not too comfortable with them. Offensive or defensive. I don't believe in magic bullets, if you will.
I remember the warning that went with using a coeffecient of determination (R²) (applied to correlation, r too) was the more independent variables you introduce to the equation, the larger the R²/lrl. Regardless of relevance to the expression. i.e., you get a number that "looks better than it is". How is linear regression relevant when discussing baseball metrics? I'm not aware of any often used metrics that use linear regression...
Really? The more variables you regress the less sure you're going to be about your regression? NO WAY! UNBELIEVABLE! If you've got numbers (w/rank/bias: eg, NOT the registration numbers of race entrants) ...then you have linear relationships. Baseball, hydraulics, sales figures, time relationships, etc.
Way. Note that the addition of independent variables NEVER decreases the value of R².
Now we're getting off on a tangent. Of a tangent.
You clearly have no clue as to what you're talking about. A coefficient of determination is used in statistical models when you're attempting to predict future outcomes based on other related data. The types of metrics we're talking about aren't attempting to do that. Using a coefficient of determination in any sort of talk about baseball statistics is completely irrelevant.
Last edited 12/7/2012 3:04 PM by stlcards5129
|
|
Reply |
Quote |
|
|
Posted: 12/7/2012 3:03 PM
Re: New Fangled Statistic Talk
MagnoliaCardFan wrote:
stlcards5129 wrote:
MagnoliaCardFan wrote:
MagnoliaCardFan wrote:
stlcards5129 wrote:
MagnoliaCardFan wrote: ...
...There's literally tons of literature discussing WAR. If you're really that interested in it go here.
.... ... (33.7 - 10.5 + 0 - 7.3 + 22.9)/10 = 3.88EDIT: that's the simple calculation, I prefer Fangraphs calculation, but that's what you asked for. 33.7? Batting-10.5? Fielding0? Baserunning
-7.3 Position22.9? Replacement10? Runs to Wins divider Here's my problem. I've literally given you a goldmine of information for the question's you've asked. If you had any real interest in understanding the metric you would have already gone and read up on it...years ago. I think your agenda is to try to discredit it. I think I have a feel for the concept. eg: I want to know how you calculated "33.7". And if you really wanted to know how to calculate any of this you'd have figured it for yourself...years ago. This stuff has been around for a while, you obviously know how to use the internet. Why don't you research it yourself. |
|
Reply |
Quote |
|
|
Posted: 12/7/2012 4:29 PM
Re: New Fangled Statistic Talk
I like the new statistics fine, although many of them are just a little too esoteric to really resonate with your average fan. One of the things I always liked about Game Score was its relative ease of calculation and - more importantly - its relative ease of understandability.
The key, to my mind, is to use as many of the stats as you have time for. If you don't have much time, then even using BA and OBA and SLG is at least something. If you have time to dig a little further and get into FIP and OPS+ and some of the others, then also fine.
They all have value. Some have more value than others, although again, the most valuable ones are the ones you have the time and understanding to use. I think there are a few people out there who'll throw out one of the new stats without really understanding them.
|
|
Reply |
Quote |
|
|
Posted: 12/7/2012 4:48 PM
Re: New Fangled Statistic Talk
I use the new metrics all of the time. I prefer them over the old time stats because they wittle things down to the nitty gritty. For instance FIP (Fielding Independent Pitching) I think is better than ERA because it takes the effect of the defense behind the pitcher out of the equation wOBA adjusts for the greater value of one type of hit over another that BA doesn't (HRs are worth more than singles). It also adjusts OBP for the different ways of getting on base.
These are the stats that the majority if not all of MLB front offices use. Those who still rely on the old time stats are going to get left behind eventually. If you have at least a basic knowledge of these metrics, it will be a lot easier to understand why teams do what they do in terms of player evaluation. Just looking at BA and ERA is not going to cut it.
|
|
Reply |
Quote |
|
|
Posted: 12/7/2012 6:44 PM
Re: New Fangled Statistic Talk
Just looking at *any* one stat isn't going to make it. You have to look at as many as you have time to look at, that was my point.
I should mention, too, that's one reason why I'm not too keen on such one-number-fits-all stats like Win Shares; there are just too many variables out there for one number to give a clear picture of what you are trying to find.
|
|
Reply |
Quote |
|
|
Posted: 12/7/2012 6:59 PM
Re: New Fangled Statistic Talk
Jmodene1 wrote: Just looking at *any* one stat isn't going to make it. You have to look at as many as you have time to look at, that was my point.
I should mention, too, that's one reason why I'm not too keen on such one-number-fits-all stats like Win Shares; there are just too many variables out there for one number to give a clear picture of what you are trying to find. I wasn't advocating just looking at one or two stats. I picked out two to use as examples. There are a whole raft of stats I look at, both offensive and defensive. I look at the whole picture. |
|
Reply |
Quote |
|
|
Posted: 12/7/2012 9:44 PM
Re: New Fangled Statistic Talk
And that's the way it should be.
That said, I don't have a problem with someone who, in a quick conversation, chooses to cite somebody's batting average when talking about him - as long as that person is willing to recognize that there's more to the big picture than just the batting average.
|
|
Reply |
Quote |
|
|
Posted: 12/8/2012 11:29 AM
Re: New Fangled Statistic Talk
Jmodene1 wrote: And that's the way it should be.
That said, I don't have a problem with someone who, in a quick conversation, chooses to cite somebody's batting average when talking about him - as long as that person is willing to recognize that there's more to the big picture than just the batting average. And that BA indicates nothing more than quantity of hits per at bat and that there are better stats for evaluating offensive production.
Last edited 12/8/2012 2:30 PM by crdswmn
|
|
Reply |
Quote |
|
|
Posted: 12/8/2012 1:27 PM
Re: New Fangled Statistic Talk
The reason i am not a big backer of these metrics or baseball card stats in general is that people tend to place too much weight on them. These numbers are only indicators of what might happen but the actual outcome always remains variable. Statistics in baseball are a bit of a red herring. There are more stats in BB than any other sport but they actually matter the least of other sports. If you think about it Baseball is the one team sport were the addition of one player will not dramatically change the prospects of winning. In football it is all about the Quarterback. In Basketball and Hockey the addition of that one special player can dramatically change a teams fortunes in a single off season. This is just not so in Baseball. Individual stats matter less in this sport than in the other major sports.There are players that are better than their numbers indicate and players that are not as good as the stats. I am not saying they are worthless they are not, but numbers will never overcome the randomness of baseball. I think this is why the best team on paper rarely wins the WS in this age of parity.
GO CARDS!
|
|
Reply |
Quote |
|
|
Posted: 12/8/2012 4:45 PM
Re: New Fangled Statistic Talk
I wouldn't go that far - in baseball the addition of the *right* one player can very often put a good team over the top.
|
|
Reply |
Quote |
