Anything You Don't Agree With?

[OneTuzSea]

Thought of this while looking at the thread about WARP and VORP; are there any things sabr-related you don't agree with? Whether it's certain princicples, or even a stat you just think is bogus?

This isn't a thread to just bash sabrmetrics if you don't like and don't care to learn about it. Just trying to get some discussion.

[VRP723]

Not that I don't use it, but there's some cases where I don't believe UZR to be 100% effective. Not to be a homer, but Matt Kemp didn't cost the Dodgers 25 runs in the field last year. Is he a good defender? Of course not. But I probably can't be convinced that he was 25 runs below average.

[bosox3431]

Not that I don't use it, but there's some cases where I don't believe UZR to be 100% effective. Not to be a homer, but Matt Kemp didn't cost the Dodgers 25 runs in the field last year. Is he a good defender? Of course not. But I probably can't be convinced that he was 25 runs below average.

Yea I don't trust UZR all that much, and done find it do be 100% effective, especially for Fenway. Crawford every full season but one has had a UZR over 10, some well over, now all of sudden his defense has just fallen off? I don't buy it.

[OneTuzSea]

UZR the biggest one I have too, along with it's effect on WAR.

It'd be awesome if they eventually make FieldFX public

[papipapsmanny]

xFIP

It doesn't even take into account whether the HRs given up were solo HRs or Grand slams. All it takes into account is flyball/hr ratio and meshes it with FIP (which is kind of flawed to begin with)

[Nymfan87]

I agree with papi, I really could care less what a player's xFIP is except for small sample sizes where a pitcher has exceptionally high or low HR rates and then I basically use that instead of FIP. And I also don't love FIP because some pitchers are simply much better at stranding runners than others. Johan Santana will always strand more runners than Ricky Nolasco but FIP doesn't account for that.

And UZR in small sample sizes, even over an entire season, is basically useless. Andrew McCutchen's UZR/150 was -12.9 last season and this year it's +12.7, a 25.6 run swing. That is absurdly huge.

[OneTuzSea]

xFIP

It doesn't even take into account whether the HRs given up were solo HRs or Grand slams. All it takes into account is flyball/hr ratio and meshes it with FIP (which is kind of flawed to begin with)

xFIP actually takes HB/FB% out of the equation. All it is, is just FIP with HR/FB normalized. It was pretty useful to see Michael Pineda's true value early in the season for example. He didn't give up any home runs through his first few starts, but his xFIP took out the luck that's involved between a flyball dying in the outfield or leaving the park. His FIP now is pretty close to what his xFIP was before, now that he's giving up home runs at a more normal rate. It's one of the best predictive stats for pitchers.

And whether it's a grand slam or solo HR doesn't really matter in what FIP is trying to prove. You can't take FIP as an all-inclusive "this-is-exactly-what-happened" stat. It's just a nice summary of everything a pitcher can control.

But I do kind of agree with you. I wish they could factor hits in to FIP somehow and make some middle ground between it and ERA, but still have it be fielding independent. Maybe some how get a baseline of what plays SHOULD be made in any given play? And if they're not made then assume in the equation that it is? If that makes any sense.

I agree with papi, I really could care less what a player's xFIP is except for small sample sizes where a pitcher has exceptionally high or low HR rates and then I basically use that instead of FIP. And I also don't love FIP because some pitchers are simply much better at stranding runners than others. Johan Santana will always strand more runners than Ricky Nolasco but FIP doesn't account for that.

That's because (as far as I know) most people consider a high strand rate purely luck because there's no real to quantify or explain why someone can strand runners better than someone else. Not saying that's right, but that's why it's also assumed it'll come down to league average.

This line of thinking in general is something I usually don't like. The fact that, just because it's hard to understand why it happens, it must be luck to have it better than anyone else. Flyball pitchers are the biggest example, ala Matt Cain.

[bosox3431]

I agree with papi, I really could care less what a player's xFIP is except for small sample sizes where a pitcher has exceptionally high or low HR rates and then I basically use that instead of FIP. And I also don't love FIP because some pitchers are simply much better at stranding runners than others. Johan Santana will always strand more runners than Ricky Nolasco but FIP doesn't account for that.

And UZR in small sample sizes, even over an entire season, is basically useless. Andrew McCutchen's UZR/150 was -12.9 last season and this year it's +12.7, a 25.6 run swing. That is absurdly huge.

The biggest reason for that is because it's based on league average. So the league average for his position changed this year.

xFIP actually takes HB/FB% out of the equation. All it is, is just FIP with HR/FB normalized. It was pretty useful to see Michael Pineda's true value early in the season for example. He didn't give up any home runs through his first few starts, but his xFIP took out the luck that's involved between a flyball dying in the outfield or leaving the park. His FIP now is pretty close to what his xFIP was before, now that he's giving up home runs at a more normal rate. It's one of the best predictive stats for pitchers.

And whether it's a grand slam or solo HR doesn't really matter in what FIP is trying to prove. You can't take FIP as an all-inclusive "this-is-exactly-what-happened" stat. It's just a nice summary of everything a pitcher can control.

But I do kind of agree with you. I wish they could factor hits in to FIP somehow and make some middle ground between it and ERA, but still have it be fielding independent. Maybe some how get a baseline of what plays SHOULD be made in any given play? And if they're not made then assume in the equation that it is? If that makes any sense.



That's because (as far as I know) most people consider a high strand rate purely luck because there's no real to quantify or explain why someone can strand runners better than someone else. Not saying that's right, but that's why it's also assumed it'll come down to league average.

This line of thinking in general is something I usually don't like. The fact that, just because it's hard to understand why it happens, it must be luck to have it better than anyone else. Flyball pitchers are the biggest example, ala Matt Cain.

This is just a guess, and based off three pitchers and what I noticed. I looked at Cain, Santana, Webb and Halladay. Maybe there's a correlation of high LOB% with being a stikeout pitcher and also being a GB or FB pitcher. Cain isn't really a strikeout pitcher, but does have a low Gb%. Sanatana in the early part of his career was a great strikeout pitcher and had a high LOB%, but now he's not striking out nearly as much, but his GB% is also a lot lower over the past two seasons. Then Roy Halladay who over the length of his career has began striking out more people, but giving up less GB along the way, and his LOB% has increased along the way. For most years those guys LOB% are above average. Now look at Brandon Webb, never a high SO guy, but his GB% were always very high, 61% or better. But his LOB% is always a lot closer to league averages, although most years is slighly above average.

It makes sense so me, considering a hitter is more likely to get a hit on the ground then in the air and of course a guy who can strike guys out is also more likely to strand runners.

I don't ven know if what I said makes sense, but something I noticed.

[Jilly Bohnson]

I still think defensive metrics are crap. It's already starting, but I have a feeling in 5 years we're going to look at +/- and UZR like we look at FRAR today. Particularly for infielders I think they're bad.

I also think people are too much a slave to FIP/xFIP at times. They're very useful, and generally they're right. But the idea that the entire difference between a players' ERA and his FIP/xFIP is luck or defense I think is a gross oversimplification.

Basically, my problem with SABR stats is more people(generally people that don't even know how they're calculated too) who treat them as gospel and don't recognize the faults and limitations of them.

[Yendil]

FIP and xFIP, they don't take into account trajectory, situational hitting, batting order, etc. Also they assume ALL pitchers throw the same amount of HR's based on FB%. It tends to favor GB pitchers over FB pitcher.

[Wake's Fastball]

As has been said, and is generally pretty well agreed-upon in the SABR community, defensive metrics are still pretty primitive.

[Mell413]

I'm starting to become less confident in defensive metrics. I've never been one to treat them as gospel, but I do think most of them meet the eye test. I would never say Kemp is -25 run fielder, but I would think it's fair to call him a below average one. I also have a hard time believing Fukudome is worth -8.5 runs at this point. I won't say I'm a scout and I'm not trying to be a homer, but I consider him a good OF.

My big gripe with saber stats is really the people that use them more than the stats themselves. I hate hearing everytime someone under performs their xFIP it is automatically due to luck and not skill. Obviously no stat is perfect, but it seems to do a disservice to a guy like Zambrano or Johan and overrate guys like Nolasco. For the most part the stats are fine, but there are exceptions and I don't think ppl notice that

[OneTuzSea]

My big gripe with saber stats is really the people that use them more than the stats themselves. I hate hearing everytime someone under performs their xFIP it is automatically due to luck and not skill. Obviously no stat is perfect, but it seems to do a disservice to a guy like Zambrano or Johan and overrate guys like Nolasco. For the most part the stats are fine, but there are exceptions and I don't think ppl notice that

That seems like more of a problem with the people than the stats, but I agree. Especially when someone uses WAR and only WAR throughout a season. Especially since it's so heavily influenced by UZR (for FanGraphs anyway).


Another thing I've found I kind of have a problem with is when someone says a catcher has zero effect on a pitcher. I'm not saying cERA is worth anything, but there definitely has to be some sort of comfort zone with a pitcher that knows his catcher and they're on the same page.

[nyyfan4life]

UZR... Does A-Rod really have a +6.0 UZR so far this year? That's his best mark since 04.

[quiksilver2491]

Although it's not my favorite stat and I can agree that it will be obsolete in the near future...I'm going to do the best job I can at defending UZR.

Not that I don't use it, but there's some cases where I don't believe UZR to be 100% effective. Not to be a homer, but Matt Kemp didn't cost the Dodgers 25 runs in the field last year. Is he a good defender? Of course not. But I probably can't be convinced that he was 25 runs below average.

Well, no stat is 100% effective and shouldn't be judged on that criteria. As with all stats, sample size has to be taken into perspective. It's impossible to watch a player and determine how many runs he was worth defensively, way too many variables that go into effect on each ball in play.

Yea I don't trust UZR all that much, and done find it do be 100% effective, especially for Fenway. Crawford every full season but one has had a UZR over 10, some well over, now all of sudden his defense has just fallen off? I don't buy it.

Sample size is a HUGE factor with any stat, more so with defensive stats then hitting or pitching stats. Judging UZR 2 and 1/2 months into the year is never going to give you an accurate portrayal of a player's defense.

However I don't expect Crawford to ever have the defensive stats he put up in TB while playing half his games at Fenway park. The green monster in LF takes away a ton of range that other parks usually have so it's a lot harder for him to be a plus defender when he doesn't have to cover as much range.

It'd be awesome if they eventually make FieldFX public

No doubt, unfortunately I don't think they are going to let that happen

And UZR in small sample sizes, even over an entire season, is basically useless. Andrew McCutchen's UZR/150 was -12.9 last season and this year it's +12.7, a 25.6 run swing. That is absurdly huge.

Not really, the data can be changed a whole lot in a small period of time. Players can also drastically change their true talent level from year to year but overall you want to look at how well McCutchen did combined, not just year to year.

I'm starting to become less confident in defensive metrics. I've never been one to treat them as gospel, but I do think most of them meet the eye test. I would never say Kemp is -25 run fielder, but I would think it's fair to call him a below average one. I also have a hard time believing Fukudome is worth -8.5 runs at this point. I won't say I'm a scout and I'm not trying to be a homer, but I consider him a good OF.

Even for scouts, the hardest aspect of a player to assess accurately is defense. A scout can watch a player one game and see a couple of plays that are absolutely spectacular and he is going to walk away thinking the player is amazing defensively. Meanwhile the player actually isn't good defensively, but just has to make spectacular plays to make up for his lack of range. Defense is hard to judge by the naked eye.

UZR... Does A-Rod really have a +6.0 UZR so far this year? That's his best mark since 04.

In season UZR is garbage, we aren't even at the halfway point of the year yet. While it's possible for A-Rod to maintain that level, based on his past performance at 3B I think it's fair to say that Rodriguez's UZR will drop








The UZR Primer, although long, does a much better job then I can of explaining many of UZRs biggest complaints.

http://www.fangraphs.com/blogs/index...uzr-primer/#12

As already stated in the thread, one can only hope that camera recorded data is made available to the public which would in all likely hood, leave UZR in the dust. In the meantime it's the best tool that us average fans have and does an adequate job of giving defensive marks to players we have never watched.