Article: A preview of 2015 MLB Team Defenses.

[TheLeviathan]

I can't believe defense is extremely important and still doubt that this writer and the stats are way overstating their case?

 

A run a game difference is pretty high, especially considering it takes no account of any of the other factors that differentiate teams.  Like talent, ballpark, divisional strength, etc.  If these numbers are to be as trusted as the author seems to believe, they eat a pretty unbelievable chunk out of any differential that exists.

 

Once again, the problem isn't the stats or the argument he's making - it's the use and presentation of them.  This continues to be an alarming trend in baseball metrics.  You don't get to fill an entire paragraph of exclamation point laden "holy craps!" about what these numbers say about defense and then hide behind "people are taking these numbers too literally".  That just doesn't work.

[gunnarthor]

 

  On 3/19/2015 at 3:29 PM, jimmer said:

Yeah, and unfortunately that ALWAYS happens.  

 

Even if we don't want to believe there was exactly a 120 run difference between the Indians and Royals defense, do we believe it was only like 25-35 runs?  I personally have no issues with the 120 run difference and certainly believe that if it isn't QUITE that much, the number it actually is comes much closer to 120 than the 25-35.  

 

And yeah, the actual point of starting this thread was to point out what some of us already know, that our defense is really bad and didn't get any better which does have a HUGE effect on preventing runs. Seems some want to believe it practically has no real effect if they believe the difference between one of the best defense and one of the worst defense ONLY saved the team with the great defense about 30 runs over the team with the horrible defense. If that's true, why even care at all about defense? It explains TRs approach to building this team :-)

I won't answer for others but the inaccuracy of the numbers bothers me a lot. I don't think anyone is suggesting that the Twins defense isn't bad - at least I haven't read that - but if you want to say that something is a problem, you might want to make sure your valuation is correct or at least close to it.  (And the fact that these are projections based on questionable measurements multiplied by inaccurate values kind of mucks the issue up more).  

 

[Mike Sixel]

Except the authors are clear, they are not precise numbers, any more than most economic numbers, or polling numbers, or numbers used in six sigma process work, or the numbers Google uses to send back search results........that's not how number actually work at the medium/highest levels of use.

[Otto von Ballpark]

Also, eye test: KC had 3 Gold Glove winners, plus a shortstop who won the previous year.  And that doesn't even include starting CF Cain and 4th OF / defensive replacement Dyson, who by all accounts have been fantastic defenders.  Plus Moustakas and Infante, who are no slouches.

[Mike Sixel]

It's not inaccurate, it is less precise than you want......that's the nature of statistics. They come with error bars and confidence levels.

 

You all ride airplanes and cars and use google......and they use the same kinds of processes/numbers in all their work.

[Otto von Ballpark]

 

  On 3/19/2015 at 3:46 PM, TheLeviathan said:

I can't believe defense is extremely important and still doubt that this writer and the stats are way overstating their case?

 

A run a game difference is pretty high, especially considering it takes no account of any of the other factors that differentiate teams.  Like talent, ballpark, divisional strength, etc.

First of all, it's not a run a game, it's .075-.8 runs per game.  That's a difference of 25-30% -- and you're the one complaining about someone "overstating their case"?

 

Second of all, RAR figures do have adjustments in them.  Otherwise, Seattle and San Diego would probably lead the league in defensive RAR every year.

 

Third, look at the RAR breakdown numbers I posted above.  What would be a more appropriate breakdown, in your eyes?  It may not be exactly the same as Fangraphs or B-Ref, but I can't imagine it being more than 25-30% different, which is apparently within an acceptable range.

[Otto von Ballpark]

Cleveland's OPS+ was 102, KC's 91.  That's not a huge difference when comparing two players, but that's two whole TEAMS.  Two groups of nine full-time spots.

 

KC's ERA+ was 114, and Cleveland's was 104.  Another big difference.  But that's pitching + defense.  Can we try to isolate the two?  What would be your best guess for doing that?

 

Considering that Cleveland had the Cy Young Award winner in Kluber, Carrasco/House pitching very well, and a solid bullpen, I don't know if their pitching was really that much worse.  Every one of KC's starters outperformed his FIP.  Every one of Cleveland's except for House under-performed.

 

Add in the eye test and the names that each team was putting in the field, and it doesn't surprise me that there would be a pretty big defensive gulf between these two clubs.  No, I don't know that it was exactly 120 runs worth, but something approaching that should not be a shocker.

[gunnarthor]

 

  On 3/19/2015 at 3:50 PM, mike wants wins said:

It's not inaccurate, it is less precise than you want......that's the nature of statistics. They come with error bars and confidence levels.

 

You all ride airplanes and cars and use google......and they use the same kinds of processes/numbers in all their work.

If my airplane going to Ft. Myers landed in Orlando, I wouldn't be ok with it being .1% accurate - although my kids would probably enjoy going to Disneyworld.

 

And I do think it is inaccurate as opposed to "less precise".  Defensive numbers, especially one year ones, can be wildly inaccurate, even for teams.  (which doesn't mean I don't think the Twins are a good defensive team right now.  No one is arguing that.  I don't think anyone is arguing they won't be one of the ten worst again either).

[Otto von Ballpark]

 

  On 3/18/2015 at 6:37 PM, gunnarthor said:

Anyway, if their measurements are right, the valuation of that means about a six win difference between the best team and the worst team, (if I'm doing the math right).

Yup.  (Actually about 7 -- it appears a WAR "win" was worth about 9.1 runs in 2014.)

 

And if a regressed projection has a 7 WAR gulf between the best and worst defenses, it shouldn't be that surprising to see a 12-13 WAR actual difference between the best and worst.  Just like hitting and pitching projections aren't going to reflect the over performers (Danny Santana, Phil Hughes) and the under-performers (Hicks, Nolasco).

[Otto von Ballpark]

 

  On 3/19/2015 at 4:17 PM, gunnarthor said:

And I do think it is inaccurate as opposed to "less precise".  Defensive numbers, especially one year ones, can be wildly inaccurate, even for teams.  

Cite on that last point?  Last time I looked, I didn't think year to year team defensive performance was any more volatile than year to year batting or pitching performances.  Are batting and pitching numbers "inaccurate" too?

[gunnarthor]

 

  On 3/19/2015 at 4:25 PM, spycake said:

Cite on that last point?  Last time I looked, I didn't think year to year team defensive performance was any more volatile than year to year batting or pitching performances.  Are batting and pitching numbers "inaccurate" too?

Here's the first thing I found - http://www.sports-reference.com/blog/2014/09/the-real-problem-with-baseballs-defensive-stats/

 

"It's also true that fielding stats don't correlate quite as strongly year to year as the batting stats do (see chart after break)."  

 

You also see pretty big variations by every team, year to year, on fangraphs lists as well.

 

If you want to say the grouping of a teams defense is right over a few years - i.e. Twins from 02-10 were a top 5 defensive team but from 11-14 a bottom 5, I'd probably agree.  But I'd probably disagree if you wanted to say they were in the bottom half of the AL in 04.

[jay]

Started a poll here - http://twinsdaily.com/topic/17287-impact-of-defense/

 

Curious to hear your thoughts.

[Otto von Ballpark]

 

  On 3/19/2015 at 4:51 PM, gunnarthor said:

Here's the first thing I found - http://www.sports-reference.com/blog/2014/09/the-real-problem-with-baseballs-defensive-stats/

 

"It's also true that fielding stats don't correlate quite as strongly year to year as the batting stats do (see chart after break)."  

 

You also see pretty big variations by every team, year to year, on fangraphs lists as well.

 

If you want to say the grouping of a teams defense is right over a few years - i.e. Twins from 02-10 were a top 5 defensive team but from 11-14 a bottom 5, I'd probably agree.  But I'd probably disagree if you wanted to say they were in the bottom half of the AL in 04.

Thanks for the link, but "quite as strongly" does not equal "wildly inaccurate".  Also, that was looking at year to year player stats.  You claimed that year to year TEAM stats (over 9 times larger than the sample of a single player) were "wildly inaccurate."

 

Also, why couldn't the 2004 Twins have ranked 8th in a 14 team league?  Cuddyer logged significant innings at 2B that year, an injured Shannon Stewart was often in LF, and even Matt LeCroy played the field.  Hunter had a DL stint that year too.

 

So what if they should have ranked 6th instead, like they did in 2003?  Again, that example does not suggest anything "wildly inaccurate" at the team level.  Less accurate than batting and pitching stats at the team level, sure, no problem.  But that's not what you claimed (and it's not what some posters seem to rail against at every mention of defensive stats).

[Otto von Ballpark]

 

  On 3/19/2015 at 4:51 PM, gunnarthor said:

Here's the first thing I found - http://www.sports-reference.com/blog/2014/09/the-real-problem-with-baseballs-defensive-stats/

 

"It's also true that fielding stats don't correlate quite as strongly year to year as the batting stats do (see chart after break)."  

Furthermore, that link doesn't claim that year to year correlation is the "big issue" with defensive stats.  It says that parsing out positioning vs. skill is the issue for individual players.  But that's not really an issue at all on a team level (team positioning IS a skill for the team).

[TheLeviathan]

  On 3/19/2015 at 3:49 PM, mike wants wins said:

Except the authors are clear, they are not precise numbers, any more than most economic numbers, or polling numbers, or numbers used in six sigma process work, or the numbers Google uses to send back search results........that's not how number actually work at the medium/highest levels of use.

This author was anything but clear, the dude had an exclamation fest to boost the justification for his breakdown. Where the heck was he clear about that? Even if the site has a sobering take elsewhere, this guy sure didn't.

[Mike Sixel]

 

  On 3/19/2015 at 4:17 PM, gunnarthor said:

If my airplane going to Ft. Myers landed in Orlando, I wouldn't be ok with it being .1% accurate - although my kids would probably enjoy going to Disneyworld.

 

And I do think it is inaccurate as opposed to "less precise".  Defensive numbers, especially one year ones, can be wildly inaccurate, even for teams.  (which doesn't mean I don't think the Twins are a good defensive team right now.  No one is arguing that.  I don't think anyone is arguing they won't be one of the ten worst again either).

 

Airplanes to land at the wrong runway, or even wrong airport sometimes, but that's not I was talking about, I was talking about maintenance and construction and cleaning and all the other stuff.....but ya, sometimes they do go to the wrong place.

[TheLeviathan]

  On 3/19/2015 at 3:55 PM, spycake said:

Third, look at the RAR breakdown numbers I posted above.  What would be a more appropriate breakdown, in your eyes?  It may not be exactly the same as Fangraphs or B-Ref, but I can't imagine it being more than 25-30% different, which is apparently within an acceptable range.

Your breakdown is more palatable for sure, fangraphs claimed 120 runs and that defies basic common sense. (And doesn't account for the things i mentioned) A little less than half a run a game seems far more likely and a more "precise" measurement. (Since we are getting silly about terms)

[jimmer]

 

  On 3/19/2015 at 6:58 PM, TheLeviathan said:

Your breakdown is more palatable for sure, fangraphs claimed 120 runs and that defies basic common sense. (And doesn't account for the things i mentioned) A little less than half a run a game seems far more likely and a more "precise" measurement. (Since we are getting silly about terms)

It doesn't defy basic common sense at all.  Statements like that is what separates people who value defense a lot as opposed to people who don't.  That's it.  Those kind of comments also show a complete inability to admit that, 'Hey, maybe I could be wrong about something' .

 

 

 

[Brock Beauchamp]

 

  On 3/19/2015 at 8:20 PM, jimmer said:

It doesn't defy basic common sense at all.  Statements like that is what separates people who value defense a lot as opposed to people who don't.  That's it.  Those kind of comments also show a complete inability to admit that, 'Hey, maybe I could be wrong about something' .

No, it doesn't admit that at all. It's questioning the validity of a metric that places a higher priority on defense than it does pitching when placed into real-world run scenarios, which doesn't pass any kind of smell test.

 

That takes issue with the validity of a number, not whether defense has a significant impact on the game (as both Levi and I have stated multiple times now).

 

In a world where the run allowed gap in the AL was ~200 runs last season, saying defense can account for 120 runs defies logic.

[TheLeviathan]

  On 3/19/2015 at 8:20 PM, jimmer said:

It doesn't defy basic common sense at all.  Statements like that is what separates people who value defense a lot as opposed to people who don't.  That's it.  Those kind of comments also show a complete inability to admit that, 'Hey, maybe I could be wrong about something' .

 

No, my statement has nothing to do with the value of defense and everything with the validity of the stat.  I think it's sad that just because something is posted by fangraphs and considered a metric it becomes so indisputable that reasonable questions about it can't be asked without it being boiled down to: "Well....I guess you don't like defense."

[Mike Sixel]

 

  On 3/19/2015 at 8:22 PM, Brock Beauchamp said:

No, it doesn't admit that at all. It's questioning the validity of a metric that places a higher priority on defense than it does pitching when placed into real-world run scenarios, which doesn't pass any kind of smell test.

 

That takes issue with the validity of a number, not whether defense has a significant impact on the game (as both Levi and I have stated multiple times now).

 

In a world where the run allowed gap in the AL was ~200 runs last season, saying defense can account for 120 runs defies logic.

 

How? If the pitching accounts for 50 runs, and the actual delta is 20.....doesn't that mean that defense accounts for more delta than the "total delta"?

 

I am 100% certain your last sentence is 100% incorrect, in terms of the "defying of logic". 

[jimmer]

 

  On 3/19/2015 at 8:22 PM, Brock Beauchamp said:

No, it doesn't admit that at all. It's questioning the validity of a metric that places a higher priority on defense than it does pitching when placed into real-world run scenarios, which doesn't pass any kind of smell test.

 

That takes issue with the validity of a number, not whether defense has a significant impact on the game (as both Levi and I have stated multiple times now).

If one feels that the difference between one of the best defenses and one of the worst differences is around 50-60 points (or lower as some seem to think) and that 120 is just so darn ludicrous as to go way past 'basic common sense' then what kind of actual value is one actually putting on defense? What would be the point of putting any priority on defense if the difference in runs saved is so little over the course of a whole season?

 

And here's the thing.  Unless the math can be simply explained to anyone at all levels of comprehension, it's always going to be slammed as B.S. I would, for once, like to see someone come out and explain why the math is wrong by doing it right for all to see.  It's so easy to say it's wrong (outside the margins of error) without providing any proof that it actually is.  The inputs are wrong? ok, why?  It doesn't take into account such and such? o.k., show me that it doesn't. . As if the people who do these things for a living don't think of these things when creating their info.

 

It seems to be so common place to slam the people who do all these metrics for a living.

[Brock Beauchamp]

Who is slamming them? We're questioning their accuracy because in a world where teams usually score within 200-250 runs of one another within a season, suggesting that defense (who has less input on plays than either pitching or offense) can make up half that number seems unreasonable.

 

The guys who develop metrics are generally smart folks... but even smart folks get it wrong a bunch of the time, particularly when we're dealing with human error in the scoring of these metrics.

 

Someday, we'll get to the point where a complex algorithm can track a 123 mph line drive to coordinates 254, 138 on the baseball field with a lift of 32 feet and air time of 2.3 seconds, tracking the fielder exactly 34 feet from point of impact when ball met wood. That's when real progress will be made with fielding metrics, not some guy/gal sitting in a box, saying "that was a 90%er... I think". Couple that potential scoring inconsistency with the infrequency of outlier defensive chances and you have a recipe for questionable data.

 

IMO, the problem is with the data collection, not its parsing.

[jay]

 

  On 3/19/2015 at 8:35 PM, mike wants wins said:

How? If the pitching accounts for 50 runs, and the actual delta is 20.....doesn't that mean that defense accounts for more delta than the "total delta"?

 

I am 100% certain your last sentence is 100% incorrect, in terms of the "defying of logic". 

 

Yeah, it's confusing the two and they aren't directly comparable.  I've been trying to think through the correct logic, but can't quite put my finger on it.  In any fashion, directly comparing a 200 real-world run gap (it's actually more like 250) to a 120 run gap in defenses doesn't actually work that easily.

[TheLeviathan]

Why so defensive?  Why can't a metric just not be as accurate as it's portrayed? Why is questioning the result of that study turned into some kind of blasphemy that you get tarred and feathered for?  

 

  Quote

If one feels that the difference between one of the best defenses and one of the worst differences is around 50-60 points (or lower as some seem to think) and that 120 is just so darn ludicrous as to go way past 'basic common sense' then what kind of actual value is one actually putting on defense?

 

 

The idea that defenses saving a bit less than half a run a game versus .75 renders it meaningless is preposterous.  It really is.  About a half run a game is a very serious impact.

[jimmer]

 

  On 3/19/2015 at 8:45 PM, Brock Beauchamp said:

Who is slamming them? 

Who?  Anyone who is saying that 120 is so much that it defies basic common knowledge, that's who. Because these people came up with that number.  When someone puts out info and someone comes backs and says that info defies all basic common sense, that's a slam on the people who came up with it.

Unless saying someone defies basic common sense isn't a slam.

[TheLeviathan]

 

  On 3/19/2015 at 8:51 PM, jimmer said:

Who?  Anyone who is saying that 120 is so much that it defies basic common knowledge, that's who. Because these people came up with that number.  When someone puts out info and someone comes backs and says that info defies all basic common sense, that's a slam on the people who came up with it.

Unless saying someone defies basic common sense isn't a slam.

 

We're questioning the conclusion of the analysis.  You know...like is done all the friggin time to all sorts of analyses to make sure they are valid.

 

You've been incapable of understanding that questioning since Chief's first challenge to it.  It is a valid and necessary part of this sort of work.

[Brock Beauchamp]

They're not directly comparable but when the Twins are both one of the worst pitching and worst defending teams in the same season, the numbers should start to align more closely. You still can't account for the upper team being a top team in both fielding and pitching but the numbers shouldn't be completely out of line.

[Brock Beauchamp]

 

  On 3/19/2015 at 8:51 PM, jimmer said:

Who?  Anyone who is saying that 120 is so much that it defies basic common knowledge, that's who. Because these people came up with that number.  When someone puts out info and someone comes backs and says that info defies all basic common sense, that's a slam on the people who came up with it.

Unless saying someone defies basic common sense isn't a slam.

I'm questioning the metric's outliers, not "slamming them".

 

And reel it in a bit. Unless you created this metric and have guardianship over it, there's no reason to get so upset that people don't buy into it 100%.

[jay]

 

  On 3/19/2015 at 8:53 PM, Brock Beauchamp said:

You still can't account for the upper team being a top team in both fielding and pitching but the numbers shouldn't be completely out of line.

 

Right, that's where the logic doesn't align.

 

The best and worst team FIPs in 2014 (which for those less familiar is how many runs a pitcher should allow before accounting for luck and defense by calculating the outcomes he can control) shows a gap of 227 runs between the best AL team and the worst.

 

A gap of 140 runs between the best and worst defense is not saying that defense is the cause of 140 of those 227 runs. They are completely separate.

 

If we are comparing the two against each other, it'd be on the net whole of them both... 227 + 140. In that scenario, it's saying defense is 38% of the runs allowed equation while pitching is 62%.

 

edit: this darn double-space thing is still eating me up.  I can't help it.