166 replies to this topic

[Mackus]

The runs saved is better than a run scored is just a micro-example of that.  A guy who is worth 15 runs offensively and -10 runs defensively is slightly less valuable than a guy who's worth -10 runs offensively and +15 runs defensively.  Team makeup matters a lot there, too. 

 

It's really pretty much equal, though.  The difference over 162 games between a .534 and .529 winning percentage is less than a win.

 

Working this example a little further, I think it'll show how little it matters whether a player "creates" his runs offensively or defensively.

 

The average run environment last year in the AL was about 8.65 runs per game.  That's roughly 700 runs allowed and scored for the average team.  Obviously that would be a .500 pythagorean record.  Let's say it's a slightly above average team, so they outscore their opponents by 20 runs, 10 created and 10 saved, so they are 710 scored and 690 allowed.  That's a .5143 pythagorean record, or 83.31 wins.

 

So say that same team had a LF who was exactly league average offensively and defensively.  Let's then replace that guy with someone who's +10 offensively but -10 defensively.  So instead of scoring 710 and allowing 690, they score 720 and allow 700.  That's a .5141 winning percentage, or 83.28 wins.

 

Let's also consider replacing him with a guy who's -10 offensively but +10 runs defensively.  So instead of 710 and 690, they are now at 700 and 680.  That's a .5145 winning percentage, or 83.35 wins.

 

So you're slightly worse off if you replace a neutral player with someone who's offensively minded but gives it all back defensively.  And slightly better off if you replace that guy with someone who has a good glove but gives it all back offensively.  It's a very minute difference though.  I think it's very fair to say that a run saved is equally as valuable as a run created.

[CA-ORIOLE]

Right......but the community suggests that the best correlation to ERA is SIERRA (includes balls in play) which is slightly ahead of xFIP (doesn't include balls in play).

 

 

So neither of those measures includes a defensive contribution.....but both claim* to highly correlate to ERA.

 

* I'm not questioning the claim, that a statistical measurement, which is fine.

-------------------

 

So if we take the ERA predictors, multiply by the innings load that gives us runs allowed in a season and then we can add up dWAR, subtract from the runs allowed and we'll use that as a predictor for the runs the team gives up?

 

Anyone ever tried that?  as a prediction tool?  In retrospect?

rWAR basically does this when adjusting pitchers WAR. They use BIS hit data (previously TZ data) to make the adjustments to the pitchers adjusted ERA/WAR. Not that uncommon to see quarter point adjustments to ERA based on defense. Not sure how useful that would be as a predictive tool. Also, as far as I know FIP is still the best predictive tool (correlation with ERA). Even better than xFIP and SIERRA (which seems odd to me). 

[Dr. FLK]

So you're slightly worse off if you replace a neutral player with someone who's offensively minded but gives it all back defensively.  And slightly better off if you replace that guy with someone who has a good glove but gives it all back offensively.  It's a very minute difference though.  I think it's very fair to say that a run saved is equally as valuable as a run created.

 

This is what it all comes down.  Look at the winning % for 3 separate teams who allow 650, 700, and 750 runs respectively.  In MLB, teams hover around .500 for the most part.  And, in that region, curves if win% vs. runs scored are pretty much linear.  That means, that changing your run differential by scoring more runs provides about the same benefit if you saved the same number of runs defensively.For a constant run differential, the team allowing the fewest runs wins the most.  But, the difference really isn't all that meaningful.  For a team with a limited budget, improving your run differential defensively is probably cheaper that adding a big bat to the lineup.  We aren't talking huge changes in runs allowed and/or scored here.  RunsScored.png   416.76K   10 downloads

[dude]

Working this example a little further, I think it'll show how little it matters whether a player "creates" his runs offensively or defensively.

 

The average run environment last yea ring the AL was about 8.65 runs per game.  That's roughly 700 runs allowed and scored for the average team.  Obviously that would be a .500 pythagorean record.  Let's say it's a slightly above average team, so they outscore their opponents by 20 runs, 10 created and 10 saved, so they are 710 scored and 690 allowed.  That's a .5143 pythagorean record, or 83.31 wins.

 

So say that same team had a LF who was exactly league average offensively and defensively.  Let's then replace that guy with someone who's +10 offensively but -10 defensively.  So instead of scoring 710 and allowing 690, they score 720 and allow 700.  That's a .5141 winning percentage, or 83.28 wins.

 

Let's also consider replacing him with a guy who's -10 offensively but +10 runs defensively.  So instead of 710 and 690, they are now at 700 and 680.  That's a .5145 winning percentage, or 83.35 wins.

 

So you're slightly worse off if you replace a neutral player with someone who's offensively minded but gives it all back defensively.  And slightly better off if you replace that guy with someone who has a good glove but gives it all back offensively.  It's a very minute difference though.  I think it's very fair to say that a run saved is equally as valuable as a run created.

 

....but here's the problem.  It's not that the theory against Pythag Record is incorrect, I'll generally support Pythag Record over a season (remember the mocking of it in 2012?)....

 

...the real question is whether defense translates to runs in the same way that offense translates into runs.

 

You sort of said that earlier, which I agree with (people need confidence in the metrics for defense), but to translate good defense with specifically reducing the Runs Allowed is hugely problematic today.  Just because you make great defensive plays doesn't translate directly to the Runs Saved nad the Pythag Record as you are ascribing in theory.

 

The challenge is the OPPORTUNITY to change the RS (offense) versus RA (defense) 

You just might not ever get the chance to make that run-saving play. (Defensively)

I guarantee you will get ABs. (Offensively)

 

Again, the discussion isn't whether there's value in defense.....I think it exists on more levels than will be related in the way some want to create numbers.....the question is the ability to impact the game the same way.

 

Look at a couple examples....Jack Z tried to leverage the 'market efficiency' of defense in Seattle several years ago (2010).  He wanted to build a speed defense club that played better to Safeco and the avert the challenges of bringing premium (expensive) hitters hitter there.  How did King Felix do (13-12....2.27 era)?  the team (scored a horrible 513 runs....couldn't prevent runs.....698)?   So now he's throwing all the efficiencies out and going after the best hitter on the market.

 

AZ got tired of not scoring enough runs and defense be damned (Trumbo in LF) they wanted some power out there and worked hard to get it.

 

Again, it's not really a question of what is trying to be done, it's a question of whether there's enough credibility in the way it's being done (or even the ability to do it) to merit the unequivical answers you are providing through those assessments

[Matt_P]

....but here's the problem.  It's not that the theory against Pythag Record is incorrect, I'll generally support Pythag Record over a season (remember the mocking of it in 2012?)....

 

...the real question is whether defense translates to runs in the same way that offense translates into runs.

 

You sort of said that earlier, which I agree with (people need confidence in the metrics for defense), but to translate good defense with specifically reducing the Runs Allowed is hugely problematic today.  Just because you make great defensive plays doesn't translate directly to the Runs Saved nad the Pythag Record as you are ascribing in theory.

 

The challenge is the OPPORTUNITY to change the RS (offense) versus RA (defense) 

You just might not ever get the chance to make that run-saving play. (Defensively)

I guarantee you will get ABs. (Offensively)

 

Again, the discussion isn't whether there's value in defense.....I think it exists on more levels than will be related in the way some want to create numbers.....the question is the ability to impact the game the same way.

 

Look at a couple examples....Jack Z tried to leverage the 'market efficiency' of defense in Seattle several years ago (2010).  He wanted to build a speed defense club that played better to Safeco and the avert the challenges of bringing premium (expensive) hitters hitter there.  How did King Felix do (13-12....2.27 era)?  the team (scored a horrible 513 runs....couldn't prevent runs.....698)?   So now he's throwing all the efficiencies out and going after the best hitter on the market.

 

AZ got tired of not scoring enough runs and defense be damned (Trumbo in LF) they wanted some power out there and worked hard to get it.

 

Again, it's not really a question of what is trying to be done, it's a question of whether there's enough credibility in the way it's being done (or even the ability to do it) to merit the unequivical answers you are providing through those assessments

 

You're comparing apples and oranges.

 

I can guarantee you that a hitter can get an at bat. I can also guarantee that a fielder will get a ball hit to him.

 

Whether or not an at bat or a defensive play has a chance to be relevant is reasonably random even if the impact of a good at bat is larger than the impact of a good defensive play on average.

[Dr. FLK]

The challenge is the OPPORTUNITY to change the RS (offense) versus RA (defense) 

You just might not ever get the chance to make that run-saving play. (Defensively)

I guarantee you will get ABs. (Offensively)

 

Again, the discussion isn't whether there's value in defense.....I think it exists on more levels than will be related in the way some want to create numbers.....the question is the ability to impact the game the same way.

I'm not sure I understand this.  Every ball in play provides someone an opportunity to save a run, does it not?  If you make a play, you have increased your team's chance of winning.  If you don't, the other team stands to benefit.  Not every ball in play gives every player a chance - I get it.  But, over time, every players gets plenty of opportunities.  Not every play "saves a run" in the box score.  But every play impacts the leverage of subsequent plays.

[dude]

You're comparing apples and oranges.

 

I can guarantee you that a hitter can get an at bat. I can also guarantee that a fielder will get a ball hit to him.

 

Whether or not an at bat or a defensive play has a chance to be relevant is reasonably random even if the impact of a good at bat is larger than the impact of a good defensive play on average.

 

No, it's all about opportunity to leverage your skills.

 

You are not guaranteed to get a defensive opportunity that leverages the skill that creates value (Defensive Runs Saved) in defensive metrics.

You are guaranteed to get offensive opportunity that leverages your skill set to create value (Runs Scored) in offensive metrics.

 

Also....don't get me wrong....the offensive metrics are flawed too (for how people usually try and use them).....but there's more relationship in there for opportunity.  It's entire foundation is expected run theory (which is generically ok, not specifically ok) and we know historically over a broad and fairly clear data set that when you (individually) get on base the opportunity to score runs (team) increases.

[hallas]

I agree with the theory (the methodology is flawed, but that's semantics) in the simplest terms: a run saved on defense is equal to a run created on offense. Where I take issue is where a LF in 95 games can have a near equivalent defensive value as Manny Machado (and you touched on this). The reason you trade some bat for glove at SS, 2B, C, and maybe CF is because of the opportunities and the number of difficult plays needed to be made at these positions. LF isn't a place where you trade bat for glove.

 

There are several things at play here. 1: The reason you don't (normally) trade bat for glove at LF is because it's usually a place where it's difficult to save a ton of runs.  You absolutely trade bat for glove if you have a LF that will save you 40 runs a year, and that same player is less effective than your current option at CF.  This is terribly unlikely, and requires a player with a bizarre skillset that only plays at a corner OF position, or an amazingly good CF.  2: A 30 UZR LF is worth less than a 30 UZR 3B because a 3B gets assigned a positional credit, while a LF gets a positional debit.  3: Outfielders are generally involved in very high-leverage plays; balls that fall in the gaps are nearly always XBH, and over-the-wall catches are obviously home runs taken away, which are assigned about 1.4 runs from a DRS standpoint.  The position is also most subject to park effects and other randomness.  A RF in Camden Yards has very few chances for wall catches because the wall at his home park is 18 feet high.  So it's definitely possible for a LF to save a ton of runs because of this, but because they receive a smaller number of higher-leverage chances, you have to take any fielding numbers with a grain of salt due to the increased volatility.

[Russ]

I think you are extrapolating my point. A LF in opacy isn't going to have the opportunity to save a ton of runs so it doesn't make sense to emphasize defense in LF. You and Mackus said it better than I did.

[Dr. FLK]

I think you are extrapolating my point. A LF in opacy isn't going to have the opportunity to save a ton of runs so it doesn't make sense to emphasize defense in LF. You and Mackus said it better than I did.

 

But what was our alternative?  It's not like the O's were choosing between adding a big time hitter and big time fielder, and went with the later.  The choices were to use existing mediocre hitters, sign a slightly better mediocre hitter, or get Lough.  Lough isn't a tremendous hitter, but he's better than the options currently on the roster.  And, he adds a dimension that we didn't have as a great defender, that adds to his overall value.

 

They didn't emphasize defense out of philosophy.  it was a move out of necessity.

[Russ]

I think you are extrapolating my point. A LF in opacy isn't going to have the opportunity to save a ton of runs so it doesn't make sense to emphasize defense in LF. You and Mackus said it better than I did.

But what was our alternative? It's not like the O's were choosing between adding a big time hitter and big time fielder, and went with the later. The choices were to use existing mediocre hitters, sign a slightly better mediocre hitter, or get Lough. Lough isn't a tremendous hitter, but he's better than the options currently on the roster. And, he adds a dimension that we didn't have as a great defender, that adds to his overall value. They didn't emphasize defense out of philosophy. it was a move out of necessity.

I was talking in general terms. Like Mackus alluded to, you get the most value out of the money or prospects you're willing to spend whether that value is offensive or defensive. I think Lough is a good get.

[Matt_P]

No, it's all about opportunity to leverage your skills.

 

You are not guaranteed to get a defensive opportunity that leverages the skill that creates value (Defensive Runs Saved) in defensive metrics.

You are guaranteed to get offensive opportunity that leverages your skill set to create value (Runs Scored) in offensive metrics.

 

Also....don't get me wrong....the offensive metrics are flawed too (for how people usually try and use them).....but there's more relationship in there for opportunity.  It's entire foundation is expected run theory (which is generically ok, not specifically ok) and we know historically over a broad and fairly clear data set that when you (individually) get on base the opportunity to score runs (team) increases.

 

Yes. A home run will always score runs while a great defensive play will not necessarily matter.

 

But we can surmise that when an individual gets on base that the opportunity for the team to score a run increases than the opposite is also probably true.

[JeffLong]

Using SEA or AZ as examples for why a defense-first team won't win a ton of games is very flawed logic.

 

How do we know that was their emphasis? Because they had Gutierrez and signed Chone Figgins? The reason Jack Z signed Cano is because his job is on the line and he needed to make big moves to keep it.

 

Towers, he's a whole other story. The guy called Didi Gregorious the next Jeter. Enough said there.

[dude]

Using SEA or AZ as examples for why a defense-first team won't win a ton of games is very flawed logic.

 

AZ is an example of a FO willing to ignore defense in LF to get some offense that they (I guess) feel they need.

 

I think the 2010 Mariners team is certainly an example of the challenges of believing a defense-only team can take you to the playoffs.

 

Again, I'm pro-defense, so I'm not making any grand statements about what it does or doesn't do.  I do think it's hugely problematic to lack offense at the collective expense of defense.  Shared before, but oWAR's foundation is expected run theory....and when you can't produce the average (neutral) array of conditions then individually, players will be worth less than and the opposite should be true too.....but neither are considered when producing the number (because it's intentionally context neutral).

 

...and again....I'm a pro-metrics guy more than anyone would understand here....but my issue is when we take reasonable (but still flawed) swags at assessments and make specific assertions based on the numbers. it just doesn't fly.  Every step of the fangraphs approach includes all kinds of caveats and recognition of the challenges......everyone says they understand them and 'we're only using them appropriately' .....and then we - and there's a lot here and everywhere - like to make very specific statements about what a player will or will not do based on those numbers.

 

I'm comfortable with the defensive metrics that suggest Lough is a solid defensive player and that has value.

Start telling me his defense will reduce the Orioles Runs Allowed by 30 or we don't need offense because DRS says he is worth +XX runs.....uh.....no.....

[CA-ORIOLE]

AZ is an example of a FO willing to ignore defense in LF to get some offense that they (I guess) feel they need.

 

I think the 2010 Mariners team is certainly an example of the challenges of believing a defense-only team can take you to the playoffs.

 

Again, I'm pro-defense, so I'm not making any grand statements about what it does or doesn't do.  I do think it's hugely problematic to lack offense at the collective expense of defense.  Shared before, but oWAR's foundation is expected run theory....and when you can't produce the average (neutral) array of conditions then individually, players will be worth less than and the opposite should be true too.....but neither are considered when producing the number (because it's intentionally context neutral).

 

...and again....I'm a pro-metrics guy more than anyone would understand here....but my issue is when we take reasonable (but still flawed) swags at assessments and make specific assertions based on the numbers. it just doesn't fly.  Every step of the fangraphs approach includes all kinds of caveats and recognition of the challenges......everyone says they understand them and 'we're only using them appropriately' .....and then we - and there's a lot here and everywhere - like to make very specific statements about what a player will or will not do based on those numbers.

 

I'm comfortable with the defensive metrics that suggest Lough is a solid defensive player and that has value.

Start telling me his defense will reduce the Orioles Runs Allowed by 30 or we don't need offense because DRS says he is worth +XX runs.....uh.....no.....

This is a pretty solid post. When I see a years worth of defensive metrics that a guy saved xx amount of runs, I certainly don't think about it the same way as I do offense. I'll generally look at a 3-4 year running average and consider that that number is closer to the truth to what actually happened. Where that number will go in the future is another question with a lot more variables.

 

Even then there are certainly issues with accounting and uncertainties in defensive numbers that you don't have with offense. Adding individual defensive rate stats is one of the biggest perils. Even carefully accumulating individual DRS/UZR component values won't match team efficiency values all that accurately in many cases. As a strategy, taking numerous guys with high defensive efficiency ratings probably has to lead to diminishing returns and loss of efficiency at some level...particularly if done at a simplistic counting level. That said, i do think guys out there are smart enough to look at all that and recognize it. Maybe SEA did try to go in too far with defense and got burned. I don't know what their logic was, but I think everybody recognizes no one can afford to be a "defense only" team.  

 

All that said I do think the defensive stats provide fairly reasonable data for planning and analysis. Like you've indicated it just has to be done smartly. 

[Matt_P]

We're going to use the best numbers we have. It should be understood that they have their limits.

 

If we accept the argument that a run scored is equal to a run allowed then it makes sense that defense isn't as valuable as offense. Runs allowed is determined by pitching (pitchers) + defense (fielders)  while runs scored is determined by offense (hitters). It only follows logically that hitters contribution is worth more than fielders contribution.

 

If so, then a team that ignores offense in order to maximize defense will be in trouble. Teams that have excellent defense and excellent pitching but terrible offense should end up around .500.

[PatrickDougherty]

BSL: David Lough wants to improve at the plate; Here's how to do it
http://baltimorespor...-improve-plate/

 

I'm actually pretty excited to see Lough this year. He's apparently a workout freak and Google autocompletes "David Lough" as "David Lough abs," so worst case scenario is that sales of the pet calendar skyrocket.

[GaryArmida]

I really like the idea of giving Lough the chance here. If anything, he's a plus defender who makes contact. There's use in those skills like you said. Good points on waking up and suddenly knowing your strike zone. 

 

Good job getting some info from inside the O's and their expectations. 

[dude]

David Lough is an easy guy to like.

 

I hope he does well and really the Orioles really need him to do something special and develop into the leadoff hitter they need next year.

 

He's probably the RFer next year.

 

That said, the article Jon did on Camden Depot is worth re-reading in terms of expectations.

 

I hope we know something KC didn't.

[PatrickDougherty]

David Lough is an easy guy to like.

 

I hope he does well and really the Orioles really need him to do something special and develop into the leadoff hitter they need next year.

 

He's probably the RFer next year.

 

That said, the article Jon did on Camden Depot is worth re-reading in terms of expectations.

 

I hope we know something KC didn't.

Yeah, he was really nice. It wouldn't surprise me if he took over RF if he works out at the plate, which is where the Royals were playing him. Thanks for pointing me to the article on Camden Depot. I'm slightly more optimistic at this point than Jon, because Lough had really good number in the minors and plays pretty good defense and I'm willing to put my stock in that, but my optimism generally wanes in the first few weeks of the season. Royals fans were bummed to see him leave, but that's probably in part because he's active on twitter and seemed like a good guy.

 

Like Jon pointed out, we've been stockpiling a bunch of LFers. Seems to me like they're taking a "throw it at the wall" approach, known in coachspeak as "we'll go with whoever has the hot hand."