Back to Boards

dWar vs. D+/-


dWar vs. D+/-

11/18 sws33 Random Chatter 0 0
I'm curious how D ratings are calculated on PC. I know dWar is a good indicator when researching, but they don't seem to be the exact same rating.

Re: dWar vs. D /-

D Rating is the number of hits saved or allowed due to an individual players DWar rating - negative DWar numbers is the number of hits he is allowing through lack of range or positioning

Re: dWar vs. D /-

Along these lines does anyone have thoughts on whether to prioritize a high fielding percentage vs. a high DWAR? I've seen guys with low fielding percentages but high DWAR and wondered if they are worth it?

Re: dWar vs. D /-

Just curious about that nyself

Re: Re: dWar vs. D /-

11/18 sws33 Random Chatter 0 0
I should have been more clear. I know WHAT dWar and D+/- is. My question is how is the PC Defense rating calculated? I'm sure it's somehow based on real life dWar, but I know they aren't quite the same number. For example, someone's real life dWar might be 2.4 but his PC D rating is 3.1

Re: Re: dWar vs. D /-

I did some quick math with one of my favorite Ruth Era players, Dick Bartell. He's a shortstop with a .959 Field% and 2.94 DWar. I'm rounding those to 3 and .960

He's averaged about 72 hits saved in the seasons with my team. And averaged about 34 errors
Averaged about 850 chances.
So by my quick math:
.10 percentage in fielding% is about 8.5 errors
So if he had a .970 fielding percentage, he'd make 25.5 errors in those same amount of chances.
1 DWar is about 24 hits saved over a season.
So therefore, and I know this is very crude:
.33 DWar is the same as .10 Fielding % assuming a hit robbed is equally positive compared to the negative of making an error.
Or rounded up:
1 DWar (for a shortstop) is equal to .30 Fielding %
That's my approximation at least!

Re: dWar vs. D /-

some of the errors could be of the 2 base type and some of the hits robbed could turn into double plays so the math is tricky

Re: dWar vs. D /-

I agree that 1 DWAR is ~24 for SS, and possibly the same (or at least very close to that) for 2b and cf... but I’d estimate about 1/2 that for corner OF and corner IF.

That’s just observational, I don’t claim to know that perfectly, but there is certainly a difference up the middle and in the corners.

Keep that in mind when considering low fld% higher DWAR guys at positions where they will NOT save you 24 hits per win.

Re: dWar vs. D /-

DWar is the secret sauce to winning in PC. Look at the team stats in your leagues, most of the time the teams at the top of your league are at the top of D +/-. The teams that are struggling are at the bottom with a negative number.

Re: Re: dWar vs. D /-

A players PC DWar rating is derived from is actual DWar rating for an entire season, it should not be different unless you are in a custom League that has maximums or due to progression or something else.

But if go to Baseball reference and look up a player and his specific stats for a season, his DWar should match his PC DWar.

Re: dWar vs. D /-

11/19 sws33 Random Chatter 0 0
Thanks @SinCity. That's what I was thinking. But then I see an example like this one -- 2013 Carlos Gomez has a dWar of 3.6 according to Baseball-Ref. In PC he's a BEAUTIFUL +4.57. I was wondering if I missed something on this calculation.

Re: Re: dWar vs. D /-

I can't answer that one- almost every guy I have seen is spot on with BB Reference- maybe it's a mistake by PC but that DWar rating is off 1.1

Only the Webmaster can answer that one...

I have seen a few other odd ones though - I know that IP can say effect WHip in PC a bit like if you have 143.1 IP and your Whip is 1.199 your Whip in PC it could be 1.200

Re: dWar vs. D /-

Gomez has come up before. That's an error on baseball reference. 4.6 is correct. If you check other sources you'll see 4.6

Re: Re: dWar vs. D /-

oddest dwar number...

Adam Dunn, -5.2 Dwar in 2009... At one point this makes me question the integrity of this stats.

So would be some JT Snow 1995 -2 dwar number... he won a gold glove that year.... meaning flat foot / iron glove Mo Vaughn should of win the gold glove instead of JT Snow.

and some Roberto Alomar below average dwar number year in and year out,

these guys (Alomar and Snow) were no doubt the greatest defenders at their respective position in that ERA...

Re: dWar vs. D /-

You have to consider that DWar is NOT an exact Science, it's someone's impressions of how many games a team would have won without that player playing at his position - it's so subjective as to be laughable I see guys who NEVER Played the field, DH'd the entire season have a negative DWar?? How their defense didn't matter

I don't care about Launch Angle and exit velocity either - I remember the Old Yankee and Tiger Stadiums with upper deck's that hung out over the field, guys developed swings to hit the ball high enough so that they landed in the upper deck - If you play in Fenway Park and hit screaming 20' high line Drives to left your going to get a lot of wall ball doubles and singles. Taylor your swing to your park, everything else is garbage

Re: Re: dWar vs. D /-

The comment about JT Snow was an interesting one. And after a little research, I found that DWar is adjusted for the player's position. So someone like Keith Hernandez (career DWar 0.6) was also considered a great defensive first baseman. But that DWar number is in comparison to all fielders and adjusted for playing a less important position.

Which makes me think that DWar is a little broken in PC. A shortstop with a 2 DWar should provide the same defensive value over a season as a first baseman with 2 DWar - because the calculation of that number has already been adjusted for position. But in PC, the 2 DWar SS or CF is way more valuable than the 2 DWar 1b.

Re: dWar vs. D /-

you guys confuse me so much

Re: Re: dWar vs. D /-

No I wouldn't say it's broken Krtyler. A +2 dWAR 1B is uber rare air to begin with, only one or two guys ever hit 2 dWAR at 1B (I know Pujols is one) but even if a player was a 2.0 1B, he simply won't be made privy to as many opportunities (SS, CF, 2B, 3B) as his counterparts. The distribution where the ball is hit around already sets up a nice selection bias. A 2 dWAR 1B SHOULDN'T have the same impact as a 2 dWAR shortstop. The 1B isn't getting the "big play" opportunities that the SS in our example is.

Re: dWar vs. D /-

Jay is spot on. SS gets probably 3.5 times as many balls hit to him as a 1B yet will rob as many hit percentage wise as a 1B with the same DWar rating

Re: Re: dWar vs. D /-

No I think you're both wrong on this. The reason that there isn't a 2 DWar 1b is because the rating has already normalized it against every fielder on the field. So it's nearly impossible to provide the same value from 1b as a normal shortstop because they don't see as many plays, and as many high impact plays. That's built into the rating.
So a 1b that robs hits at the same rate as a shortstop will have a much lower DWar. In order to have the same DWar as a shortstop, they'd have to rob hits or make plays at a much higher rate.

if they did do that - they should do that in PC. And that's my point.

A 2 DWar 1b would have the same impact over a year as a 2 DWar SS - which means they'd make as many hit saving plays in a smaller amount of opportunities. But that's not how it's simmed out in PC - at least not in the sandoxes I've tried.

Re: dWar vs. D /-

Agreed KRtyler, position adjustments are already accounted for. Think it's a great point that hasn't been brought up before.

Not saying it's worth tuning though. It's probably more of a custom league problem with out of proportion dwars at certain positions.

Re: Re: dWar vs. D /-

Hard for me to know if it's working as it should, because I don't play in a lot of leagues with high DWAR players at positions like 1B, but technically, the way the logic works, a 2.0 dwar first baseman has just as much of a chance of saving a hit on every opportunity as a 2.0 DWAR shortstop, if that was what you were asking for. The 2.0 dwar SS is going to save a lot more hits because of the high rate of opportunities.

Re: dWar vs. D /-

Yeah that's exactly what I was asking for Guy - thanks for verifying that. It's how I would expect it but after finding out that there are positional adjustments baked into the DWar number, it changes how I think it should be simmed.

Re: Re: dWar vs. D /-

I think I am reading your post correctly now - you were saying a 2.0 DWAR first baseman should save *as many* hits as the SS even over fewer opportunities.

This is going to take some thinking on my part, but I am not sure I am squaring up with that... A 2.0 dwar first baseman is still 2 games better than the average fielder. A 2.0 shortstop is also 2 games better than the average fielder. So let's say for the moment then that they are both equal defensive athletes... put that player at SS, he saves more hits by virtue of his chances. Put him at 1B he will save fewer. Put the first baseman at SS and he'll save the same number of hits as the SS would have at that position.

The position adjustment is not saying that a 2.0 first baseman is more amazing than a 2.0 shortstop... I think most pundits argue the 2.0 shortstop, even with the positional adjustment, is still the better player.

But I am not claiming to be a DWAR expert!! :)

Re: dWar vs. D /-

I think KRTYLER is correct. Here is some evidence in support.

DWAR is a combo of fielding runs above average AND a position adjustment. Baseball reference gives data on the fielding runs before the position adjustment (rfield).

In 2007, Albert Pujols had 2.3 dWAR at 1B
In 1966, Rico Petrocelli had 2.2 dWAR at SS

But Pujols had 31 Rfield, while Petrocelli only had 12

Pujols was worth nearly 3 times as many runs in raw fielding, but because Rico is a SS and SS fielding takes place over more chances, their final Dwars are the same.

It would seem that Pujols is a better defender and that position adjustment makes him equally defensively valuable with Rico. In PC, either all dWAR should be equal impact regardless of position OR we should use Rfield instead of dWAR and then use the PC chances by position to adjust value.

Re: Re: dWar vs. D /-

I am a newbie on the site but have been following sabermetric stats for a long time. And I'm with krtyler and joshbonds for sure.

DWar is a counting stat (like runs or RBI, though way more complicated to calculate), not a rate stat (like AVG or OBP). A 2 DWar 1B's defense was worth 2 wins at 1B; a 2 DWar SS's defense was worth 2 wins at SS. Both saved their teams 2 wins through defense. Since 1B do get fewer chances than SS, presumably the 2 DWar 1B was worth more *per chance* than the 2 DWar SS. That's why the numbers joshbonds wrote up come out the way they do.

You really can't think of DWar as some measure of raw defensive value - it's all relative to the position. Position really matters. Given the different skills required, I'd guess even most 2 DWar 1Bs would have very negative DWar at SS - if they could play SS well they would not be playing 1B. There are far more humans that can play 1B well than there are humans that can play SS well.

In my *very* limited experience, I feel that the value of defense is nicely calibrated relative to offense in this sim. (I know that some feel otherwise.) But, given how the DWar stat is calculated, its PC implementation makes it currently perhaps *too* important for positions that get lots of chances, and maybe not important enough for the corners. I care very little whether this is changed - again, in the big picture, I'm quite happy with defense. But, I thought I'd weigh in on my understanding of the stat.

Re: dWar vs. D /-

Let me just agree with Jade regarding how defense is valued in PC.

I think this is one of the best sims at valuing D. While I believe using Rfield (fielding runs) instead of dWAR would be an improvement in extreme cases at the corners like Pujols or Brooks Robinson. It’s a small improvement in the grand scheme of things.

I’m happy with D how it is right now (I also enjoy the value of speed)

Re: Re: dWar vs. D /-

This is a really good thread. I recall mentioning when DWAR was rolled out that we were not using it for what it really was. We're sort of hijacking it as a rate-stat when it isn't.

I can agree it feels corners should have a little more value in PC while middle of the field maybe a little less.

My point was only that PC is a per-chance game. The 2.0 DWAR first baseman is still more value than a replacement 1B in PC *and* that same player could be placed at SS and be just as valuable as a 2.0 SS. (That part is probably unrealistic, but the FLD% penalty evens things out.)

I have a hard time wrapping my head around "wins above replacement" when comparing two players of different positions. In the example above, are we to assume Albert Pujols made more great plays than Rico Petrocelli, or is it just that Pujols made more plays than a typical first baseman? Or was it just the timing of those plays happened to be more valuable?

Re: dWar vs. D /-

My understanding (and this may be wrong) is that the fielding runs numbers suggest Pujols saves his team more runs compared to the average player (not just 1B) in 2007 than Petrocelli did in 1966.

That said, because lots of players can play 1B, while very few can play SS, they make an adjustment for positional scarcity that gives extra credit to up the middle positions and penalizes corners.

When WAR is calculated, offensive and defensive runs are added together and then the positional adjustment is made.

When oWAR and dWAR are separate the positional adjustment is made to both separately.

So dWAR is runs saved above average adjusted for positional scarcity

Re: Re: dWar vs. D /-

No first base man ever robbed more hits than a 2B or SS ever did, maybe he had a great glove and made the catch on a ball fielded by those guys and that should not be discounted, but in PC its strictly hits robbed, and that stat goes to the fielder

Re: dWar vs. D /-

Well, a hit robbed could certainly include a scoop on a bad throw, although you're right the credit right now goes to the fielder... so that does seem to align with the argument that corners in PC should make more plays.

I still probably wouldn't deviate too greatly from how it works, because right now there's a pretty realistic strategic element to solidifying your up-the-middle defense.

Re: dWar vs. D /-

I like the balance of D to hitting and up the middle to corner right now in general. Given that, it may not be worth it to try to implement what I’m going to say next. But I know nothing of programming. So, in case this wouldn’t be hard, let me lay out an idea:

Currently, I estimate that 2B/SS/CF get twice as many chances as 1B/3B/LF/RF

I’ve done some tests that allow me to estimate expected hits robbed based on dWAR and position. As Kingdude mentioned earlier that’s about 24 per 1 dWAR for up the middle guys, and 12 per 1 dWAR for corner guys.

Examples

1989 Ozzie Smith is 4.8 dWAR. I expect about 115 D*
2007 Pujols is 2.3 dWAR. I expect about 28 D*
1985 Keith Hernandez is 0.6 dWAR = 7 D*
1956 Jackie Robinson is 2 dWAR = 48 D*

I propose switching to Rfield, but keeping the twice as many chances for up the middle guys.

1989 Ozzie is 32 Rfield. To keep a similar hits robbed I estimate about 3.5 hits robbed per fielding run for up the middle and 1.75 hits robbed per fielding run for corners (I know you’d need to assign chances of robbing a hit per chance).

If we use these for Rfield with our examples

1989 Ozzie 32 Rfield = 112 D*
2007 Pujols 31 Rfield = 54 D*
1985 Hernandez 14 Rfield = 25 D*
1956 Jackie 19 Rfield = 67 D*

With fielding runs, an up the middle defensive wizard is still the way to go. More than twice as valuable as the best 1B season ever.

But the best 1B season ever is now close to as valuable defensively as a good 2B season like 1956 Jackie Robinson.

And a good 1B like Keith Hernandez may not be close to a good up the middle guy, but his glove is useful vs now where 7 hits robbed doesn’t matter much.

If you switched to Rfield, I don’t think it would overvalue corners in general- up the middle D would still be a championship formula. BUT, it would better reflect the value of guys like Keith Hernandez, Don Mattingly, and Brooks Robinson- the rare corners who were so good with their glove that it mattered.

Re: Re: dWar vs. D /-

Intriguing idea definitely worth consideration in the future.

While it wouldn’t be hard to switch the logic over to a different stat, the lengthy effort is bringing a new stat into PC. The dwar project was a multi month endeavor. It’s not hard, it’s just time consuming because every aspect of the site (importing, copying, compiling, archiving, etc etc etc) has to be changed to accommodate a new stat. Not to mention that stat has to get imported for every player and every season in baseball history!

Re: dWar vs. D /-

I'm having deja vu all over again. Happy to see some other guys making the point I was making last time I saw this on the boards.

I want to preface this by saying that I think the sim is great as is, and I have no strong preference for changing it, or not. It's a argument of stat nerds (I count myself as one) who don't see dW lining up with how the real-life MLB dWAR works. And so when I say "fix" or "should" below, I just mean it in the sense of "to make dW line up with dWAR" and I'm not sure how much of a priority that is for anyone other than stat nerds. :) The site is great as is - just want to make that clear!

To join in with Josh, Jade, and krtyler... and summarize:

dWAR is "wins above replacement." A 2 dWAR 1B contributed as many wins as a 2dWAR SS.

In the PC world, that would mean (loosely) that a 2 dWAR 1B saved as many hits as a 2 dWAR SS. And THAT means that a 2 dWAR 1B would save a higher percentage of his opportunities than a 2 dWAR SS.

As for programming... I think the easiest "fix" would also be really accurate.

I would imagine in the programming that the chance of a saved hit is just a function of dW converted to a percentage. For players at 1B, 3B, C, and LF/RF, that percentage "should" be doubled.

Maybe that would lead to too much defense across the site, so perhaps they can/should all be weighted differently. (CF/SS/2B adjusted to 75% of current levels, while 1B/3B/C/LF/RF adjusted to 1.5x current levels.)

For what it's worth, I ran a couple sandbox leagues and came away with these averages.... hits saved per dW:

C - 11.4
1B - 12.7
2B - 24.4
3B - 12.4
SS - 26.8
LF/RF - 11.7
CF - 25.7

Lots of variability in league structures, number of teams, etc... But that was a base 2019 stats league run a couple of times. Theoretically, those numbers for "hits saved per dW" should all be the same... And they all fall in one of two categories: 12 or 25. Doubling the 12s works... or to not hurt offense... making them all 18 or something like that might be best.

All that said... I'm not lobbying for a change... or against a change. But that would make dW line up closer with dWAR.

Re: Re: dWar vs. D /-

Love this idea Ritz
"Maybe that would lead to too much defense across the site, so perhaps they can/should all be weighted differently. (CF/SS/2B adjusted to 75% of current levels, while 1B/3B/C/LF/RF adjusted to 1.5x current levels.)"

Re: dWar vs. D /-

Lots of good points and research but you are not considering the fact that a pitchers H/9 already take into effect all the DWar they got in compiling their stats.

Pitchers already are reaping the value of whatever defense they got in compiling their actual stats, I for one think they deserve less benefit from DWar then they already get.

Re: Re: dWar vs. D /-

Ill have to patch in corresponding links, but we've really only become more precise (using video, etc) since 2002, prior to that, we use Total Zone as a 'guesstimator" to how many balls were hit in a player's direction. We use play-by-play from Retrosheet, LH/RH tendencies of hitters, it's hardly exact throughout history.

Re: dWar vs. D /-

Yeah agree generally with what’s being said that you can’t tweak it to its true intent or you’re going to have way too much defense and you’re also going to lose the value of focusing on the most important defensive positions. The point about great defense already being baked into the stats is legit.. pitchers have those plays baked in so they are already double dipping to an extent. It works out okay on PC because most leagues have such loaded lineups the extra defense doesn’t skew things.

Re: Re: dWar vs. D /-

To the last point, yes, pitchers are double-dipping on the hits saved, but that is the way all of these stats work.

The best players play a lot more on PC. All of those guys who spend the full season on waivers... they all played in MLB.

MLB pitchers got to face many much worse hitters. MLB hitters got to face many much worse pitchers/fielders.

I think it all works out well in the auto-leagues, and then the custom guys can tweak how they want.

I don't think we'd want significantly more D* plays overall. If any change was made, I'd assume it would be a positional weight to take dWAR back into it's positional weighting.

Re: dWar vs. D /-

Another thing to keep in mind if adjustments are made in PC are players with secondary positions. If defensive value was increased in the infield for 1B, 3B, and C it would then need to be reduced if those players had a secondary position of IF, SS, or 2B when playing out if position. Otherwise you might end up with, for example, a 3B playing 2B or SS and making way too many plays (with a few extra errors) that would be very unrealistic. It gets even more complicated in the OF. If you take a +3 DWar CF and move them to LF or RF they would likely have a much different Dwar in reality over a season.

I agree with everyone that DWar is being used differently in PC than in MLB but once you understand it I think it is a pretty good compromise for not greatly overcomplicating things.

Just thought I would weigh in with my thoughts.

Re: Re: dWar vs. D /-

Just my one note of caution:

When you make a change in the relationships to address one issue, you tend to create imbalances elsewhere. MLB is a perfect example: Pretty every time MLB makes a rule change to 'speed up the game', the game slows down measurably.

If it ain't broke, don't fix it.

Re: dWar vs. D /-

Just to jump back in real quick- I think the many comments worried about throwing things off are right to think about that. It is useful to ask what the purpose is of a thing before you change it. To me, a defensive stat on PC’s purpose is 1) increase strategy by diversifying the pathways to building a winning team, and 2) better reflect the value of glove guys like Ozzie Smith

DWAR is doing a good job of 1 and 2, but it currently fails at reflecting the value of glove guys at the corners- especially 1B. That’s because dwar adjusts for position penalizing 1B especially and then PC does too in total chances.

The worry is that in trying to rectify this it results in 1) overvaluing D and thus reducing the different ways you can win, and/or 2) overvaluing corners and thus undervaluing the Ozzie Smiths and Johnny Evers of the world.

I still contend that keeping the basic system we have now but using fielding runs (Rfield on baseball reference) instead of dWAR is a fix. Rfield is what dWAR is based on, so it’s not radically different and it is available for every player that dWAR is available for. BUT, it is not adjusted for position, so PC’s adjustment isn’t a double dip. MOST guys will get about the same D*, but a few avg 2B/SS get less as they don’t get a double bonus and a few great defensive 1B/3B get more as they aren’t double penalized

Re: dWar vs. D /-

And it has slowed down! That new RP rule, games still took longer. Like 3 hr and 10 min I believe. Players are slowing down and there are too many pitches. More pitches in the strike zone would mean more balls in play. yea, yea I know. Wrong thread.