What is Dice?

[yankees4life]

I haven't really ever known what it is so can someone give me a good explanation.

[petrel]

DICE = defense-independent component ERA

DICE = 3.00 + (13*HR + 3*(BB + HBP) -2*K)/IP

The idea is that ordinary ERA is dependent on what kind of defense you have behind you. If you have quick infielders and fast outfielders, a lot of hits get turned into outs. If you have a crappy infield and a lethargic outfield, fly balls and infield grounders end up turning into hits.

In general, the concept is: "one a pitcher throws a ball, he has no control over what the fielders will do with it". DICE (and other stats like DIPS) strip out the fielding luck in ERA. Note that all the stats used in calculating DICE --- home runs (given up), baseball on balls, hit by pitcher, strikeout, and innings pitched -- are largely under the control of the pitcher and not the fielders.

You get a projection of what a "normalized, defense-independent" ERA would be. The strength of this stat is that you can look at a pitcher's stats at the end of the year and get some idea of what his true value is. Did he have a better than normal ERA because of good fielding? Did he have a worse one because of bad fielding? DICE helps you pick out the good ERAs from the phony ones.

--Pet

--Pet

[Alloutwar]

Thanks Petrel, I was always wondering that too.

Yankees4Life - I would congratulate you on the good question (brave even?), but you're a Yankees fan, so...instead, I will simply not say anything negative about you.

:)

[slowrx7]

In general, the concept is: "one a pitcher throws a ball, he has no control over what the fielders will do with it".
-----------------
Uh, no. I mean that's what the concept is, but it is no where near correct. So dice assumes a pitch that is grooved is just as effective as one on the corner with nice movement. This is most retarded thing to base a concept on I have seen in a long time. I was ok with Petrels answer (as he did answer the question correctly) until he added his little 2 cents with this paragraph.

........................................................
You get a projection of what a "normalized, defense-independent" ERA would be. The strength of this stat is that you can look at a pitcher's stats at the end of the year and get some idea of what his true value is. Did he have a better than normal ERA because of good fielding? Did he have a worse one because of bad fielding? DICE helps you pick out the good ERAs from the phony ones.
....................................................


Ok, bad fielding. This may come as a surprise to some of slobbermetric lovers here, BUT AFTER THE PITCH THE PITCHER IS A FIELDER. They even hand out gold gloves to the good ones. True Value? I think a pitcher who gives up less runs is more valuable than one who gives up more. The formula is so broken, that I laugh at people who use it. (No offense Clay, but I do). Where are the GDIP's? (Oh wait, that must be all luck) The balks and wild pitches (Oh I guess the pitcher had no control of where the ball would end up after he threw it) And don't get me started on how overrated a K is. Yes great effective pitchers can rack up a lot of K's. But that is a side affect of greatness, not the reason they are great. Fielding luck? As if it's all luck? Give me a break. LOL. Ever stop to think a weakly pulled grounder is a **** of a lot easier to field than a line drive hit with authority? Oh wait it was 'luck' that the fielder handled the grounder but couldn't react the the liner. Right.

Ok, rant over. Continue with your ignorant bliss.

[HoustonGM]

Uh, no. I mean that's what the concept is, but it is no where near correct. So dice assumes a pitch that is grooved is just as effective as one on the corner with nice movement. This is most retarded thing to base a concept on I have seen in a long time. I was ok with Petrels answer (as he did answer the question correctly) until he added his little 2 cents with this paragraph.
DICE works with things that pitchers have NEAR 100% CONTROL over. It removes the fielder's influences. That's it.

DICE doesn't assume anyting about pitches. It doesn't take into account anything but the outcome of the at bat.

Focusing on the things that a pitcher can completely control by himself is not a retarded concept. The retarded concept is you assuming that people think DICE is the only meaningful stat. Defense-independent stats are ONE method of evaluating a pitcher. To evaluate a pitcher, you need to use multiple methods if you want the best possible evaluation.




Ok, bad fielding. This may come as a surprise to some of slobbermetric lovers here, BUT AFTER THE PITCH THE PITCHER IS A FIELDER. They even hand out gold gloves to the good ones.
Uh, so?


True Value? I think a pitcher who gives up less runs is more valuable than one who gives up more.
Okay, a pitcher who has given up less runs than one who has given up more runs HAS been more valuable. But DICE isn't used as a tool to evaluate how valauble a player HAS been. It's a tool used to evaluate how valuable a player WILL be. A pitcher with an ERA of 3 and a DICE of 5 SO FAR is likely to see his ERA rise, and a pitcher with an ERA of 5 and a DICE of 3 (SO FAR) is likely to see his ERA decline. So, when having to decide between two pitchers for the remainder of the season (or the following year), it's best to look BEYOND ERA. ERA is highly influenced by outside events not controlled by the pitcher, which leads to ERA's being high or low due to luck.

An example. Jason Marquis had a spiffy 2.35 ERA after April. Using your logic, Marquis will be very valuable going forward. However, you have to look at beyond ERA - Marquis walked 13 and struck out just 14. Its impossible to sustain such a high ERA with such a shitty walk:strikeout ratio. according to you, you should just assume that he somehow is able to prevent runs jsut by virtue of being on the mound, and it's because of how he's pithced that his ERA is low, so, obivously, that means he's a valuable pitcher and he'll keep preventing runs. Using the logic of "slobbermetrics", you see that Marquis was likely getting lucky and stranding an inoridnate amount of baserunners, and likely getting a lot of help from his fielders. If he continues to pitch the same way, he will not sustain that 2.35 ERA. And...the next month, it rose to 3.38 throughout May...then June, 5.09. July, 5.79. If you look beyond the ERA, you could've predicted such a decline.


The formula is so broken, that I laugh at people who use it. (No offense Clay, but I do).
ERA is broken as a singular stat. I laugh at people who use it and reufse to look beyond it.

YOU MUST USE MULTIPLE STATS. NOBODY SHOULD USE JUST DICE. NOBODY SHOULD USE JUST ERA.


Where are the GDIP's? (Oh wait, that must be all luck)
No, that's heavily influenced by the ability to induce groundballs, which is why, in addition to using DICE, you should look at groundball rates, and how many double plays the pitcher has induced.


The balks and wild pitches (Oh I guess the pitcher had no control of where the ball would end up after he threw it)
They're not in the DICE formula, which is why you should also look at how many iwld pitches a guy throws. Balks are basically irrelevent.


And don't get me started on how overrated a K is. Yes great effective pitchers can rack up a lot of K's. But that is a side affect of greatness, not the reason they are great.
That's retarded logic. Using that logic, a great pitcher who walks very few batters isn't great because he walks few batters. He walks few batters because he's great. NO! One of the reasons he's so good is because he has good control and walks few batters.


Fielding luck? As if it's all luck? Give me a break. LOL. Ever stop to think a weakly pulled grounder is a **** of a lot easier to field than a line drive hit with authority? Oh wait it was 'luck' that the fielder handled the grounder but couldn't react the the liner. Right.
Nobody says anything is ALL luck.

How about this one dude. A pitcher allows a fly ball into the corner in left field. His left fielder happens to be Jack Cust, and the pitcher isn't also the manager, so it's not the pitcher's fault Cust is in left field. Cust doesn't reach the ball (and it's a ball a guy like Carl Crawford would have reached), and the runner ends up getting an inside the park home run. ERA rises. WASN'T THE PITCHERS FAULT. IT WAS JACK CUSTS FAULT.

This isn't "luck." It's "not being able to control everything." We had this discussion in the other thread. Its' nice to see that you read my posts. If you read anything I said, you'd have learned by now that DICE doesn't say that everything besides K, BB, and HR's is luck. It simply says that everything else is largely out of the pitcher's control.


Ok, rant over. Continue with your ignorant bliss.
:rolleyes: You're the one that refuses to even read a study or article that challenges your view point. Talk about ignorance.

[slowrx7]

UH, for one I did read the article you are talking about. So lets squash that stuff right now.

Just because you post something, doesn't make it right - and it certainly doesn't mean it would change my mind. Obviously, the original poster didn't read them, and that's who I was responding to (and petrel), not you. Conceited *grumbles*

Half of your responses were 'isn't in the formula' - no way dude! I was talking about how the formula is broken and how people that rely upon are silly. If this isn't you, then don't worry about it! You basically proved my whole argument that DICE does not meet its goals of being 'a better ERA'.

As far as your fielding argument goes - hes in left field. (Nevermind how a runner can get a homerun, hitters get those, but I digress) Assume the batter is right handed - you know theres a weak fielder in left, so PITCH HIM OUTSIDE. He tries to pull it, he more than likely winds up grounding out to short (if the fielder is 'lucky' enough)

K's ARE overrated.

The assumptions of what a pitcher 'controls' are in the formula are all false.

HR's - You blew that up with your inside the park analogy (Thanks though, I wouldnt have been able to come up with that without you!)

HBP - Some of these are in retaliation, and others can serve another purpose - they are not always 'mistakes'

Walks - Say Mr. Cust drops a foul ball 3-2 pitch on 12 consecutive pitches, and finally the pitcher gives up ball 4 (pitchers fault!)

K's - overrated at the least - always in the control of the pitcher - most of the time. Sometimes the batter will swing through a pitch intended to induce a GDIP

Clay himself says DICE is a major part of the sim engine. That reliance on a broken stat scares me.

Walks (part deux) - sometimes are the unintentional intentional type to face an easier batter if putting the runner on is no or little risk (or sometimes an advantage)

Hits arent even in there! I realize why, because of the whole basis of 'defense independent'. But if homers are in there, why not triples? Is not a triple a major offensive weapon?

The stat needs revising. As it stands, it is a worthless measurement. Is era flawless? Of course not. I never said it was. But DICE is about 13 steps down in accuracy. Houston, I will work with you in coming up with a better 'perfomance evalutor' if you like, but seriously, DICE in its present form is in no way accurate. I *DO* look at several things when I evaluate pitchers. But dice? No way. Never.

[HoustonGM]

Half of your responses were 'isn't in the formula' - no way dude! I was talking about how the formula is broken and how people that rely upon are silly. If this isn't you, then don't worry about it! You basically proved my whole argument that DICE does not meet its goals of being 'a better ERA'.
Nobody should rely solely upon ONE stat. I do use DICE, but I combine it with WHIP, BAA, OOPS, K/BB, K/9, BB/9, HR/9, GB%, etc. I use many stats to evaluate a pitcher. That is the best way to evaluate pitchers.


(Nevermind how a runner can get a homerun, hitters get those, but I digress)
When a hitter hits a ball, he becomes a runner, whatever, that's such a technicality that I don't know why you even bothered to much it.


the batter is right handed - you know theres a weak fielder in left, so PITCH HIM OUTSIDE. He tries to pull it, he more than likely winds up grounding out to short (if the fielder is 'lucky' enough)
And yet, it's still possible for the outcome I outlined to happen. The pitcher does not have full control over where the ball is hit.


HR's - You blew that up with your inside the park analogy (Thanks though, I wouldnt have been able to come up with that without you!)
Inside the parkers are rare enough that they don't distort anything.


HBP - Some of these are in retaliation, and others can serve another purpose - they are not always 'mistakes'

Walks - Say Mr. Cust drops a foul ball 3-2 pitch on 12 consecutive pitches, and finally the pitcher gives up ball 4 (pitchers fault!)

K's - overrated at the least - always in the control of the pitcher - most of the time. Sometimes the batter will swing through a pitch intended to induce a GDIP
I never stated that the pitcher has complete control over any of them. Many factors are involved in every different stat. However, homers, walks, and strikeouts are able to be controlled by the pitcher A LOT more than hits, or what have you.

Don't believe me? It's been proven that K rates, BB rates, and HR rates correlate from year to year at a much higher level than other pitching stats. This suggests a repeatable skill (or lack thereof).


Clay himself says DICE is a major part of the sim engine. That reliance on a broken stat scares me.
When did he say that? The "sim" engine can't rely on any stats. All it does is produce stats.


Walks (part deux) - sometimes are the unintentional intentional type to face an easier batter if putting the runner on is no or little risk (or sometimes an advantage)
I know.


Hits arent even in there! I realize why, because of the whole basis of 'defense independent'. But if homers are in there, why not triples? Is not a triple a major offensive weapon?
A home run is more directly the pitcher's fault than a triple. The triple is a ball in play that the fielder's deal with, not the pitcher.

But again, stop acting like people rely solely on DICE! There are many stats, all of which should be taken into consideration. Opponent Slugging takes triples into account. Use it if you're concerned about triples.


The stat needs revising. As it stands, it is a worthless measurement.
No, it's not!!!! I'll prove this after you reply to my questions below


Is era flawless? Of course not. I never said it was. But DICE is about 13 steps down in accuracy.
In what kind of accuracy?


Houston, I will work with you in coming up with a better 'perfomance evalutor' if you like
There's tons of other performance evaluation stats out there that aren't in Mogul, a good amount of which are better than DICE. FIP, xFIP, DIPS ERA, Pitching Runs, SNLVR, VORP, etc.


but seriously, DICE in its present form is in no way accurate. I *DO* look at several things when I evaluate pitchers. But dice? No way. Never.
Answer these questions individually, please.

Do you look at a pitcher's tendency to allow home runs?
Do you look at a pitcher's control?
Do you look at a pitcher's strikeout rate?

[slowrx7]

1) Yes, but I look to see how many are of the solo variety

2) To an extent, but I prefer BA against. Statistically speaking, I know a walk is worse than pitching to a guy with a .900 OPS, but in some circumstances, I prefer conventional wisdom (first base open, get to a weaker hitter, his run wouldnt affect the outcome - ie runner on 3rd is the winning run, load the bases up say if the score is tied in the 9th and the leadoff hitter triples)

3. Almost never, unless he is relief ace, (in bill james definition of the word - I believe Bill's model for a bullpen is more efficient than the LaSorda model)

[slowrx7]

July 19, 2000

If you play Baseball Mogul, you have already encountered Defense Independent Component ERA ("DICE"), even though you don't realize it. This is because the artificial intelligence in Baseball Mogul uses DICE to evaluate pitching talent.

We also use it at Sports Mogul to create our annual player projections.


Pretty much sums it up. So the sim engine will use projected stats, of which DICE is used for the pitchers.

[ohms_law]

Which creates very realistic outcomes...

[slowrx7]

Uh huh.

[TexanBob]

Ok, bad fielding. This may come as a surprise to some of slobbermetric lovers here, BUT AFTER THE PITCH THE PITCHER IS A FIELDER. They even hand out gold gloves to the good ones. True Value? I think a pitcher who gives up less runs is more valuable than one who gives up more. The formula is so broken, that I laugh at people who use it. (No offense Clay, but I do). Where are the GDIP's? (Oh wait, that must be all luck) The balks and wild pitches (Oh I guess the pitcher had no control of where the ball would end up after he threw it) And don't get me started on how overrated a K is. Yes great effective pitchers can rack up a lot of K's. But that is a side affect of greatness, not the reason they are great. Fielding luck? As if it's all luck? Give me a break. LOL. Ever stop to think a weakly pulled grounder is a **** of a lot easier to field than a line drive hit with authority? Oh wait it was 'luck' that the fielder handled the grounder but couldn't react the the liner. Right.

Ok, rant over. Continue with your ignorant bliss.

Thank you. (and thank you, Petrel, for the explanation).

Someone once said of Greg Maddux "He has the ability to make batters hit grounders right to his guys." So I guess that skill is totally useless in DICE. And yet he's won over 300 games. Must have been pure luck.

[etothep]

i dont want to get into the middle of this one, but if one is to bash a stat due to its inaccuracy, i would need to see some evidence of how it is an inaccurate stat to use

by evidence, i am looking for numerical data, as i hardly see much use for theoretical situations in the validation of a statistics's accuracy...obviously, if dice is 13 steps down from era in terms of accuracy, i would like to see your data in relation to era if you would be so kind

[Alloutwar]

Methods of Evaluating Pitchers - ranked best to worst in Accuracy:
#37 - ERA
#38 - Body Mass Index
#39 - High school Chemistry grades
#40 - Pure Looks
#41 - Lines of Monty Python they can quote verbatim
#42 - Ability to grow facial hair
#43 - Regional (southerners are just better - especially if from Texas)
#44 - Not allowing runs 'n stuff
#45 - Nickname-a-bility, i.e. "Oil Can", "Rocket", "Spaceman", "El Duque"
#46 - Lack of other pertinent skills
#47 - Fear of high-fiving teammates (this would be a negative factor)
#48 - Height (what is Randy Johnson, 7' 11"??)
#50 - If the Yankees are willing to spend millions on you (again, negative factor)
#51 - DICE
#52 - Wins

[slowrx7]

I don't really understand what you're getting at.

[slowrx7]

lol alloutwar

[etothep]

I don't really understand what you're getting at.

you stated how dice is 12 steps below era in terms of accuracy in regards to rating pitchers and a pitcher's ability, so im simply asking for your statistical proof in such a claim, so that i can become more informed in which methods of rating pitchers are most helpful

[slowrx7]

Baseball isn't a statistical game. We use statistics to put into a form easier understood. You can't statistically prove one is better than the other, but we can find fallacies in the logic for such formulaes.

[ohms_law]

lol

[jcbarr]

Baseball isn't a statistical game. We use statistics to put into a form easier understood. You can't statistically prove one is better than the other, but we can find fallacies in the logic for such formulaes.

Do you even realized what you just said there?

Are you arguing just for the pleasure of doing so?

Basically you just said that your whole basis of argument is because you have no basis of argument...

[Alloutwar]

Ok - so what have we learned today?

We all have many opinions; we are all knowledgeable about Baseball stats, in varying degrees, and some of us feel passionate for/against certain statistical measurements being taken as paramount in the game. ERA isn't perfect because defenses can suck, DICE isn't perfect because hits can be due to a pitcher missing location, etc. Wins has always sucked, I think we can all agree.

So to get the best possible feel, you examine a BUNCH of stats - WHIP, OBA, and the other sabermetric stuff. Done? Hug? HUG! I LOVE you guys!

And even then, your guys always "want to live somewhere colder." I'm paying them MILLIONS, darnitall!! Why does a few degrees matter!! WEAR LONG SLEEVES!

Factors in making players happy, according to BM:
1) temperature in the part of the country your team happens to be in
2) where his best friend plays
3) order in pitching rotation
4) "change of pace" factor. Apparently, major league is boring. Try sitting in my CUBE all day, you spoiled brat!!
and then, distant runner-ups...
5) Money
6) Contender status

does this seem odd to anyone else? Now THIS I will fight about!!

[HoustonGM]

1) Yes, but I look to see how many are of the solo variety

2) To an extent, but I prefer BA against. Statistically speaking, I know a walk is worse than pitching to a guy with a .900 OPS, but in some circumstances, I prefer conventional wisdom (first base open, get to a weaker hitter, his run wouldnt affect the outcome - ie runner on 3rd is the winning run, load the bases up say if the score is tied in the 9th and the leadoff hitter triples)
Evidence shows that walk rates are a much more repeatable skill than hit rates. And the "intentional" walks (even "unintentional" intentional walks) barely cloud the stats of a pitcher. I think you look at control without even realizing it. You see a pitcher who's walking 6 batters per 9 innings, you know that's a red flag.


3. Almost never, unless he is relief ace, (in bill james definition of the word - I believe Bill's model for a bullpen is more efficient than the LaSorda model)
Well, just so you know, starting pitchers really need to have a decent strikeout rate to succeed unless they also do something else extremely well (like Chien-Ming Wang), so if you're ignoring strike out rates, and some guy has a fluke in which he strikes out 3 per 9 innings but manages a low ERA, you're GOING to get burned.

Anyway, nearly everybody looks at a pitchers home run rates, control, and strikeout rates. All DICE does is combine them into one stat. It's far from "worthless."


Pretty much sums it up. So the sim engine will use projected stats, of which DICE is used for the pitchers.
No. The sim engine uses the predicted stats, and the predicted stats include everything - including hits. You predict a guy to give up 1 hit in 200 innings, but strikeout just 50 guys, he's gonig to be ridiculously good, even though DICE woudl say he wasnt.

The AI uses DICE, not the sim engine.


Someone once said of Greg Maddux "He has the ability to make batters hit grounders right to his guys." So I guess that skill is totally useless in DICE. And yet he's won over 300 games. Must have been pure luck.
Just because "someone once said" something, doesn't make it true. You're foolish if you really think a pitcher has a significant amount of control over where the batter hits the ball. Yes, he can locate it to try and get the batter to pull it or hit to the oppsoite field or whatever, but he CANNOT MAKE the batter hit it directly to a fielder. The pitcher has no control over where the fielder is positioned and how the fielder reacts to a ball.

And no, even if that "skill" existed, it woulnd't be totally useless in DICE. Why? Because he gets the batter out, which improves DICE. Also, because the ability to induce grounders lends itself to the ability to prevent home runs well.

But, even if the skill isn't taken into account in DICE, that doesn't make DICE a worthless stat. It makes it a stat that doesn't take into account something. By your logic, slugging percentage is worthless because it doesn't take into account a player's walks. All DICE is meant is to be a stat that takes into account three things that are largely controlled by the pitcher himself - walks, strikeouts, home runs. That's what it calculates, just like slugging percentage is a calculation of singles, doubles, triples, and home runs. DICE isn't meant to be, nor does anybody claim it to be, the be all and end of all of stats that is the ultimate stat for a pitcher.


Also, Greg Maddux has a really good DICE of 3.22. Why? Because except for 2004 and 2005, he allowed very very few homers, and from 1993 on, walked very few batters, and because a strikeout rate of 6.9 per 9 innings is decent. And, his career ERA is 3.07. You'll notice that when given significant sample sizes, DICE matches up pretty nicely with ERA...

Watch. I'm going to take a look at every active player (this is without 2007, just for your information) with 75 career starts, and then look at their ERA and their DICE. This gives a sample of 134 pitchers.

The results were actually pretty god **** astonishing, and surprised even me, because I didn't think the correlation was this good.

Of this sampling of pitchers, their average ERA was 4.30. Thier average DICE was 4.26. :eek:

Two players, Roger Clemens and Jose Contreras, have a DICE that matches exactly to their ERA.

The largest difference between ERA and DICE is Ryan Drese, with a 5.31 ERA and a 4.36 DICE. He's the only player who has a difference of .9 runs or more. The only other player to have a difference of .8 runs or more is Jeremy Bonderman, with a 4.72 ERA and a 3.92 DICE. The only other player with a difference of .7 runs or more is Horacio Ramirez with a 4.13 ERA and a 4.91 DICE. Only two more players have a difference of .6 runs or more - Jaret Wright (5.07 ERA, 4.47 DICE) and Ryan Franklin (4.35 ERA, 5.00 DICE). And, only two more players have a difference of .5 runs or more - Barry Zito (3.55 ERA, 4.12 DICE).

7 out of 134 players (5 percent) had a DICE that differed from their ERA by half a run or more, meaning 127 of 134 (95 percent) had a DICE that differed from their ERA by less than half a run. 81 of 134 players (60 percent) had a DICE that differed from their ERA by a quarter run or less.

I've attached the spreadsheet with the list of pitchers, their ERAs, DICEs, and the difference between the two.

Now, confronted with this overwhelming evidence that DICE correlates extremely EXTREMELY well with ERA, are you still going to deny its usefulness?

Obviously there are individual seasons where DICE can differ from ERA greatly, because the sample size is smaller. But, given a large sample size, DICE tends to match up perfectly with ERA. Given that, and the fact that a pitcher's DICE correlates from year to year better than a pitcher's ERA, I think it's pretty obvious that DICE is a better predictor of future ERA than past ERA is.


i dont want to get into the middle of this one, but if one is to bash a stat due to its inaccuracy, i would need to see some evidence of how it is an inaccurate stat to use

by evidence, i am looking for numerical data, as i hardly see much use for theoretical situations in the validation of a statistics's accuracy...obviously, if dice is 13 steps down from era in terms of accuracy, i would like to see your data in relation to era if you would be so kind
Well, luckily for you, I just proved that DICE is a near perfect predictor of ERA. How's that for accuracy?


Baseball isn't a statistical game.
All I have to say to this is :eek:


We use statistics to put into a form easier understood.
No. Here are why we use statistics:

1) to keep a record of what happens on the baseball field
2) to evaluate players
3) to predict future performance


You can't statistically prove one is better than the other
I did just statistically prove that DICE is a great indicator of ERA.

[HoustonGM]

Those results really did surprise me. I did not expect to see such a high correlation. Given this which is news to me, I'm actually going to start incorporating DICE into my pitcher evaluations even more.

[slowrx7]

Evaluations are subjective is I was trying to get across, 100 years of stats, and we still argue about who is a HOFer and who isn't. We use statistics to explain what happened in game, a season, or a career. You made ANOTHER flawed statement in saying you MATHEMETICALLY proved DICE is a decent evaluator, when common sense says elsewise.

[slowrx7]

I find it severly amusing when people who never played the game or pitched competively think they understand pitching and that the pitcher can't influence how well or where a ball is likely to be hit.

[Gaedel]

Evaluations are subjective is I was trying to get across, 100 years of stats, and we still argue about who is a HOFer and who isn't. We use statistics to explain what happened in game, a season, or a career. You made ANOTHER flawed statement in saying you MATHEMETICALLY proved DICE is a decent evaluator, when common sense says elsewise.

Is any stat a decent evaluator under the definition which you use? Even counting stats like Hits aren't necessarily good measures of value. Pete Rose has more hits than any other player in history. Is this because Rose is the greatest player ever or is it because Rose had so many plate appearances?? If you add up the lifetime stats of Billy Goodman, Barney McCosky, and Dale Mitchell, you get almost a perfect match right down the line with Pete Rose's stats. Pete Rose is thus one of those players with a longer career.

DICE is a "regression" formula which takes three measures of pitcher value (HR/IP, BB/IP, and K/IP) and combines them into a single measure which approaches ERA in the same manner in which Pythagorean W-L approaches real life W-L. The data show that if DICE and ERA diverge significantly over a short term, the ERA will tend to regress towards the DICE.

Regression formulae weight several data elements in a methodology to produce a figure of merit which is an artificial, yet usefull, overall measure.

[HoustonGM]

We use statistics to explain what happened in game, a season, or a career. You made ANOTHER flawed statement in saying you MATHEMETICALLY proved DICE is a decent evaluator, when common sense says elsewise.
I did mathmaticlaly prove that DICE corralates very well with ERA. Care to dispute that?

"Conventional wisdom" or "common sense" may say otherwise, but the actual cold hard facts PROVE IT. And facts are indisputable, whether or not they disagree with common sense. Common sense is just "common" not necessarily CORRECT all the time.


I find it severly amusing when people who never played the game or pitched competively think they understand pitching and that the pitcher can't influence how well or where a ball is likely to be hit.
Well, first of all, you don't know whteher or not me or anybody else here has or has not played the game or pitched competitively. I played Little League, so I've played the game. There are also many people who have played at high levels and still believe in defense-independent pitching statistics.

NOBODY IS SAYING THAT THE PITCHER CANNOT INFLUENCE HOW WELL OR WHERE THE BALL IS LIKELY TO BE HIT.

Jesus christ.

Yes, the pitcher can attempt to influence how hard a ball is hit and the direction in which he goes. He does not have full control over this, though, and you cannot say he does. Also, once the ball is in play, whether or not he influenced where it went and how hard it was hit, IT IS THEN ENTIRELY OUT OF HIS CONTROL (unless it is hit to him).

And jesus ****ing christ. THIS IS JUST ONE STATISTIC. It's doesn't say that the pticher can't influence anything else. All it does is tell you in one number how well he keesp the ball in the yard, prevents wlaks, and strikes people out. THAT IS ITS PURPOSE. It isn't meant to tell you how well a pitcher possibly prevents hits. There's other stats that you use for that.

Just like slugging percentage doens't tell you how well a hitter gets on base. You use more than one stat to evaluate the hitter.


DICE is a "regression" formula which takes three measures of pitcher value (HR/IP, BB/IP, and K/IP) and combines them into a single measure which approaches ERA in the same manner in which Pythagorean W-L approaches real life W-L. The data show that if DICE and ERA diverge significantly over a short term, the ERA will tend to regress towards the DICE.

Regression formulae weight several data elements in a methodology to produce a figure of merit which is an artificial, yet usefull, overall measure.
I don't see how that is so hard to understand....


Now that you've been confronted with evidence completely proving that DICE correlates very highly with ERA, and thus is a good predictor of ERA, you have to rely on accusing the other side of "never playing the game" which is just an old, tired cliche that means absolutely nothing.

DICE correlates year-to-year for a single pitcher better than ERA.
DICE correlates very well with ERA.
Therefore, DICE is a better predictor of ERA than past ERA.

[etothep]

I find it severly amusing when people who never played the game or pitched competively think they understand pitching and that the pitcher can't influence how well or where a ball is likely to be hit.

isnt that the same argument bonds used the other day when chastising bob costas for questioning the "cleanliness" of bonds' hr total?

[HoustonGM]

It's a ridiculous argument that "anti-stat" people try to use to demean the intelligence of people who use statistics, as if playing the game somehow means you automatically are more knowledgable that one who hasn't played they game. They also don't care if the "stat guy" actually has played the game, as you know, anybody who has played the game knows that stats are useless.

[slowrx7]

No, I argued the whole basis of thinking behind DICE is flawed, which it is.

[HoustonGM]

No, I argued the whole basis of thinking behind DICE is flawed, which it is.
No, it isn't.

Dude.

Is the whole basis behind slugging percentage flawed?

It's the same thing.

HR, BB, and K are three stats for pitchers. DICE is a stat that measures those three stats.

THAT'S IT. THERE'S NO MORE BASIS BEHIND IT.

Also, if it's so flawed, than why can it predict ERA so well?

[slowrx7]

Ok, Houston GM placed on ignore

[HoustonGM]

That's absolutely lovely. I'm proving you wrong, so you place me on ignore. If you're so confidant in your position that DICE is a useless, worthless stat, defend it.

But, right, you can't defend it, because I just proved using indisputable facts, that is isn't meant to be anything more than an estimation of a pitcher's ERA using just HR/BB/K, and I also proved that it actually correlates very well with ERA, and is a better predictor of ERA than ERA itself.

And since you can't argue back against those points, rather than concede that DICE actually has a use, you place me on ignore.

That's just absolutely ridiculous. It's not like I'm insulting you, harassing you, or anything. If I was, those would be good reasons to place me on ignore, but ignore me simply because I'm proving to you that DICE is a usable stat? That's ridiculous

[petrel]

Interesting. Houston GM, I just ran an R-squared on your list of ERA vs. DICE and came up with a value of 0.74614. If I understand R-squared correctly, this implies that 74.61 percent of the variance in ERA can be attributed to DICE.

Component ERA, from which DICE is derived, has an R-squared of about 0.843, according to some website somewhere. So Component ERA is a better predictor than DICE.

A question: on your ERA vs. DICE list, are the ERAs/DICEs lifetime ERA or season ERA?

--Pet

[HoustonGM]

A question: on your ERA vs. DICE list, are the ERAs/DICEs lifetime ERA or season ERA?
Lifetime. Season ERA is much more prone to random variation, and if we just looked at a single season of X amount of pitchers, the results would be more varied, as there isn't enough to time for ERA to really even out with the DICE.

[slowrx7]

Heh, you thought I really placed you on ignore!? I was joking.

Nah, I realize it shows Hr BB and K, but I can just look at those per 9 ip myself. Fine in itself, but really no purpose to it. No reason that it really needs to exist.

It's fine for people to look at DICE. But does it accomplish what is was intended to do? No. It is based on flawed assumptions, and as such, I choose to disregard it in the context of what it's creator intended. It's just way too simplistic and disregards several factors while overemphasizing others.

Now perhaps if it took Range Factor of the defenses behind him for every at-bat - weighted heavier at the positions that receive more balls, Home runs other than solo types, ground balls with runners on base, tossed out walks thats were obviously pitch arounds, lowering the value of a k, and other things that actually make sense, I would probably endorse it.

I in no way endorse wins and era as the end all to evaluating performance. I am sure there are better evaluators out there, and if not, I will work with you to come up with one.

Does DICE have a use? Sure, a very very limited use. It is what it is. A value based on a few isolated stats. But I refuse to give it the credit people give it due to it being based on flawed and not very well thought out concepts.

I think you have to understand where I come from. People are giving DICE much more value than it deserves, and I am simply pointing out that perhaps the value of this stat should be rethought for the reasons I have listed (and I listed more than a few reasons)

[ohms_law]

...I think that the only person here who's giving DICE more credit that it deserves may be you.
:eek:

[slowrx7]

Ohm's - save your charm for someone it will work on.

[shepster]

Ok - so what have we learned today?

We all have many opinions; we are all knowledgeable about Baseball stats, in varying degrees, and some of us feel passionate for/against certain statistical measurements being taken as paramount in the game. ERA isn't perfect because defenses can suck, DICE isn't perfect because hits can be due to a pitcher missing location, etc. Wins has always sucked, I think we can all agree.

So to get the best possible feel, you examine a BUNCH of stats - WHIP, OBA, and the other sabermetric stuff. Done? Hug? HUG! I LOVE you guys!

And even then, your guys always "want to live somewhere colder." I'm paying them MILLIONS, darnitall!! Why does a few degrees matter!! WEAR LONG SLEEVES!

Factors in making players happy, according to BM:
1) temperature in the part of the country your team happens to be in
2) where his best friend plays
3) order in pitching rotation
4) "change of pace" factor. Apparently, major league is boring. Try sitting in my CUBE all day, you spoiled brat!!
and then, distant runner-ups...
5) Money
6) Contender status

does this seem odd to anyone else? Now THIS I will fight about!!

This one I have a few post to fun abought :) factor.
This team is not Violent enough? YOUSE your BAT!
This team is too Charming? Whats with that Doesn't like Girls?
Zito likes to Bean Polar Bears.
O Yea Pitcher Wins game Good Pitcher "Throws" game Bad

[ohms_law]

Ohm's - save your charm for someone it will work on.
Awww... I'm heartbroken!
:)


This one I have a few post to fun abought :) factor.
This team is not Violent enough? YOUSE your BAT!
This team is too Charming? Whats with that Doesn't like Girls?
Zito likes to Bean Polar Bears.
O Yea Pitcher Wins game Good Pitcher "Throws" game Bad
hehe
Nice!

[MeetDaMets]

Ignore DICE
and just
roll the bones.

[etothep]

Interesting. Houston GM, I just ran an R-squared on your list of ERA vs. DICE and came up with a value of 0.74614. If I understand R-squared correctly, this implies that 74.61 percent of the variance in ERA can be attributed to DICE.

Component ERA, from which DICE is derived, has an R-squared of about 0.843, according to some website somewhere. So Component ERA is a better predictor than DICE.

A question: on your ERA vs. DICE list, are the ERAs/DICEs lifetime ERA or season ERA?

--Pet

im not saying right or wrong on your assumptions pet, but i need some better supporting evidence than "according to some website somewhere" due to the fact that "according to some website somewhere" smoking is actually healthier than exercising

[HoustonGM]

Heh, you thought I really placed you on ignore!? I was joking.
Good to know.


Nah, I realize it shows Hr BB and K, but I can just look at those per 9 ip myself. Fine in itself, but really no purpose to it. No reason that it really needs to exist.
Than there's no reason that on-base percentage really needs to exist? After all, you can just look at hits and walks and hit-by-pitch. And how about batting average? You can just at hits and at bats.


It's fine for people to look at DICE. But does it accomplish what is was intended to do? No. It is based on flawed assumptions, and as such, I choose to disregard it in the context of what it's creator intended. It's just way too simplistic and disregards several factors while overemphasizing others.
What was it intended to do? It was intended to be a stat that focuses on what a pitcher's ERA "should be" based on the things he has near total control over. Going by the fact that it correlates well with ERA, I think it did accomplish what it was intended to do.


Now perhaps if it took Range Factor of the defenses behind him for every at-bat - weighted heavier at the positions that receive more balls
DICE is a rather simplistic "defense-independent" stat. There are other stats out there, though not in Mogul, that actually do account for the specific defense behind a pitcher.


Home runs other than solo types
It takes into account all home runs.


ground balls with runners on base
The effect those have on a large scale of numbers isn't all that great. And, why does that need to be taken into account with DICE? What if that runner got on base via a hit, which doesn't show up in DICE anyway? What if that groundball is a single? What if that groundball doesn't result in a double play? I'm not sure why ground ballers with runners on base needs to be taken into account for DICE. DICE isn't mean as a all-in-one pitching stat. That's not what it is, so why does it matter if it doesn't take everything into account? And again, DICE DOES ACCOUNT FOR A PITCHER'S GROUNDBALL TENDENCIES? How? Groundball pitchers give up less home runs.


tossed out walks thats were obviously pitch arounds
I'm all for removing intentional walks from a pitcher's actual walk total, but there's absolutely no way to tell which walks were "obviously pitch arounds."


lowering the value of a k
which would deflate the whole idea behind what defense-independent pitching statistics are. Pitchers need to be able to strike batters out. There are very few pitchers that can get by with a below-average strikeout rate. The only guy I can think of off the top of my head is Chien-Ming Wang, who still gets a good DICE because he gives up few home runs and doesn't walk many. Strikeouts aren't overvalued in DICE. It's just that pitchers aren't going to succeed with low strikeout rates if they don't also excel well elsewhere, as Wang does. If they do other things good, they'll still have a strong DICE.

It's a fact that pitchers who strike out few batters won't last long. Look at Jeremy Sowers. He did great last year, but many people saw that he had a measly strikeout rate, and he wasn't very exceptional anywhere else. These guys said that the Indians should expect Sowers to struggle this year. "FEAR NOT!" said the people who refuse to look at better statistics. "Sowers had a good record and a low ERA! He'll be fine!" A few months later, Sowers has been demoted to the minors. Those that read deeper into his statistics correctly predicted that his low strikeout rates will start to haunt him.


I in no way endorse wins and era as the end all to evaluating performance. I am sure there are better evaluators out there, and if not, I will work with you to come up with one.
Wins are the worst evaluator there is, and should never be taken into consideration. ERA should always be taken into consideration, but needs to be adjusted for context as well. And, like I said, there are many stats out there. I listed a bunch.


Does DICE have a use? Sure, a very very limited use. It is what it is. A value based on a few isolated stats. But I refuse to give it the credit people give it due to it being based on flawed and not very well thought out concepts.
But it correlates very highly with ERA...suggesting that it is a very good predictor of ERA...why is that bad?! And how is that not useful?

And yes, I agree with ohms, it seems as though you're the only one giving DICE more credit than it deserves. You make it out as if people use it as the be-all-and-end-all of pitching stats, and nobody has even come close to suggesting that. It's just one stat that should be used in evaluating pitchers, and considering it correlates very well with earned run average, it's a strong predictor of ERA, which is good.

[BubbaDrew]

I don't see anyone ranting about how great DICE is.

slow doesn't like dice, so he doesnt have to use it to judge pitchers.

houstonGM understand dice and will use it in correlation with other stats to judge how good a picher is.

i want to meet all the imaginary people ranting and raving about DICE, really.

[HoustonGM]

I don't know who here plays fantasy baseball, but one of the main things a lot of fantasy baseball writers try to do is decipher which players are getting "lucky" or "unlucky" and then buying low or selling high on such players.

Dan Haren was the top fantasy pitcher (and basically the top pitcher in the league) for nearly the entire first half. However, starting in early June, there were articles suggesting that people should try and trade Haren, despite his pretty ERA. Why? Because his underlying numbers suggested he was going to regress. Yes, he was giving up few hits...but it wasn't due to him being some exceptional pitcher, but rather a miniscule BABIP that I think hit a low of like .200, which is entirely unsustainable and due nearly entirely to great luck. If we looked simply at his W-L record and his ERA and hits allowed, we'd see this amazing pitcher who would never start dropping. But those that looked at his numbers saw a decline in his future...and what do you know? His last 5 starts, he has a 4.60 ERA and a 1.50 WHIP.

[slowrx7]

You like DICE? Fine, use it. Anything to get you to shut up.

[Rongar]

Methods of Evaluating Pitchers - ranked best to worst in Accuracy:
#37 - ERA
#38 - Body Mass Index
#39 - High school Chemistry grades
#40 - Pure Looks
#41 - Lines of Monty Python they can quote verbatim
#42 - Ability to grow facial hair
#43 - Regional (southerners are just better - especially if from Texas)
#44 - Not allowing runs 'n stuff
#45 - Nickname-a-bility, i.e. "Oil Can", "Rocket", "Spaceman", "El Duque"
#46 - Lack of other pertinent skills
#47 - Fear of high-fiving teammates (this would be a negative factor)
#48 - Height (what is Randy Johnson, 7' 11"??)
#50 - If the Yankees are willing to spend millions on you (again, negative factor)
#51 - DICE
#52 - Wins

LOL:D...I like your witty signature, too...

[OldFatGuy]

Personally, I think number 44 is highly overated :).

[HoustonGM]

You like DICE? Fine, use it. Anything to get you to shut up.
I'm just arguing against the flawed notion that it's a useless stat. I don't care if you don't use it or not, but to call it useless is wrong, and I've proven that.

[slowrx7]

Well that didn't work as well as I had hoped. And you've 'proven' nothing. But seriously, if you feel it has a use, then use it. I choose to assign it a weight of 0. Got a friggin' problem with that?

[HoustonGM]

I have aproblem with you claiming its a useless stat, and everytime someone mentinos using it, having to chime in and tell them how it's worthless.

And I Have porven that it correlates high with ERA. Care toe xplain how a high correlation with ERA is abad thing?

[jcbarr]

Well that didn't work as well as I had hoped. And you've 'proven' nothing. But seriously, if you feel it has a use, then use it. I choose to assign it a weight of 0. Got a friggin' problem with that?

You as well have not proven anything for your side of the argument either. DICE isn't for everyone, but it is far from worthless. It is a good prediction of ERA as Houston has shown, that is all that he has ever said it was...

[Gaedel]

I have aproblem with you claiming its a useless stat, and everytime someone mentinos using it, having to chime in and tell them how it's worthless.

And I Have porven that it correlates high with ERA. Care toe xplain how a high correlation with ERA is abad thing?


Well, it would correlate highly with ERA since the pitcher is trying to prevent runs and home runs and walks are significant elements of hitter's "runs created". If you produced a "batter's DICE" based on HR, W, and K per Plate Appearance, it would probably correlate very highly with batter's RC/27.

HR per AB probably has a very high correlation with Isolated Power.

This in no way denigrates DICE as a useful tool in evaluation, but points out that a large part of the correlation is causeed by the definition of (and intercorrelation between) the two measures.

[HoustonGM]

Well, it would correlate highly with ERA since the pitcher is trying to prevent runs and home runs and walks are significant elements of hitter's "runs created". If you produced a "batter's DICE" based on HR, W, and K per Plate Appearance, it would probably correlate very highly with batter's RC/27.

HR per AB probably has a very high correlation with Isolated Power.

This in no way denigrates DICE as a useful tool in evaluation, but points out that a large part of the correlation is causeed by the definition of (and intercorrelation between) the two measures.
Well, you're exactly right and that's exactly the point. DICE takes the three elements that are under near-complete control of the pitcher, three elements which all play a high part in run prevention, and puts that into one number. Since HR, BB, and K rates are all very repeatable skills, and they are all major factors in run prevention, using them (or a stat that combines them, like DICE) becomes a better predictor of ERA than prior ERA because ERA is much more prone to random variation and outside influence.