Is FIP any better than ERA or WHIP?

Why is this not in the statistics and scorekeeping forum?

Also warning to anyone, do not download anyone's files here unless you scan first!

You don't need the FIP constant to use FIP, the constant's just there to scale FIP to ERA so that a "good" FIP is roughly equivalent to a "good" ERA for the league in question.

What you would need for a non-MLB league was a method of measuring the relative values of BB/HBP, K, and HR at the level you're interested in. I suspect that the values of all 3 in the HS run environment differ at least somewhat from the values assigned for the MLB level (HR especially, given how relatively rare they are). In theory, this shouldn't really be hard to do given enough data.

All that said, there's no particular reason you'd have to go to the trouble. If you're willing to assume that the underpinnings of FIP hold true at the HS (or college, or whatever) level, you can just look at K/PA, BB/PA, and HR/PA to get a pretty good idea of how good a pitcher is relative to his competition.

Of course, it's far from a given that the underpinnings of FIP do actually hold at the HS level (again, with regards to HR especially), and coming to a conclusion on that is going to be much harder given the data it would be necessary to have to do it right.

FIP's main value is as a predictive stat, BTW, since it correlates better with future ERA than actual current ERA does. But I know you're not actually interested in predictive stats...

Does the attached data show players who pitched in pre-BBCOR days?  Looks like it might. If so, you may need to start over.

You've always been able to calculate FIP without the constant, which is all you need for comparison purposes within a data set.  The constant only exists to adjust the number to be in line with ERA, the comparisons remain the same without it.

FIP is not a measure of performance in the sense that ERA or WHIP is, so you can't really say that it's better than either of those. It's a better predictor of future performance than either of those, but that's not the same thing. And, since you don't care about predictions anyway, that difference shouldn't be germane.

FIP essentially is just a combination of BB/PA, K/PA, and HR/PA, and it is easier to combine them for the purposes FIP is used for, but the weights given to each in the FIP formula for MLB aren't going to be the same as you'd need for HS or college ball (or, probably, for at least some other levels of pro ball). Rather than spend time on calculating the FIP constant from your non-MLB data, you'd be better off doing the necessary legwork to calculate the appropriate coefficients for BB/K/HR for the level you're looking at.

Or, if you're lacking the time/ability/data to do that, you'd be better off comparing the individual pieces K/PA, BB/PA, and HR/PA for your data set, since those are easy to look at individually and don't require any normalization if you're not trying to combine them into a single useful statistic. But, hey, keep re-inventing the wheel if you like.

Originally Posted by jacjacatk:

You've always been able to calculate FIP without the constant, which is all you need for comparison purposes within a data set.  The constant only exists to adjust the number to be in line with ERA, the comparisons remain the same without it.

I know that, but obviously you haven’t produced many metrics for players, parents, and coaches. Believe me, there’s always someone who looks up the metric, and when they see the algorithm isn’t the same, I’m the one always stuck with trying to explain why. Most people who aren’t into metrics in depth, don’t want to hear about how FIP would be the same without the constant. I’ve produced it under a different name for myself, but I like others to be able to see what I see.

FIP is not a measure of performance in the sense that ERA or WHIP is, so you can't really say that it's better than either of those. It's a better predictor of future performance than either of those, but that's not the same thing. And, since you don't care about predictions anyway, that difference shouldn't be germane.

I didn’t say it was a better measure of performance, JH did. So if you want to argue its merits, you take it up with him.

FIP essentially is just a combination of BB/PA, K/PA, and HR/PA, and it is easier to combine them for the purposes FIP is used for, but the weights given to each in the FIP formula for MLB aren't going to be the same as you'd need for HS or college ball (or, probably, for at least some other levels of pro ball). Rather than spend time on calculating the FIP constant from your non-MLB data, you'd be better off doing the necessary legwork to calculate the appropriate coefficients for BB/K/HR for the level you're looking at.

I’m not looking to solve the world’s problems, but I do like to see if what I work with can be used to show the same things that work on ML data. It doesn’t need to be exact because no one’s gonna use it to draft anyone or pay them, and even if it were, its next to impossible to get the data.

Or, if you're lacking the time/ability/data to do that, you'd be better off comparing the individual pieces K/PA, BB/PA, and HR/PA for your data set, since those are easy to look at individually and don't require any normalization if you're not trying to combine them into a single useful statistic. But, hey, keep re-inventing the wheel if you like.

I don’t feel like I’m re-inventing anything, but I am trying to do more than say the HS numbers mean nothing. I’ve asked you before, if you can do better, please do. I’ll be the 1^st one to use what you come up with.

Originally Posted by JCG:

Does the attached data show players who pitched in pre-BBCOR days?  Looks like it might. If so, you may need to start over.

Why? Are you saying the numbers would be significantly different?

Aren't your team's and league's  HR numbers significantly different before and after BBCOR?  They sure are around here.

HR rates would be substantially different, which would lead to the coefficients being different, not that you're looking at dealing with that anyway. HS/college walk and K rates are probably enough different to warrant another look as well.

http://www.fangraphs.com/commu.../why-i-dont-use-fip/

Yeah, because parent's eyes aren't going to glaze over when you introduce FIP anyway. To the extent that they care, giving them FIP with the MLB constant would be fine anyway.

FIP with the MLB coefficients doesn't really tell you anything with any particular accuracy about non-MLB ball anyway. If you're not going to do the work to adjust those, I don't know why you'd bother with the work to calculate the constant. Just use FIP and call it a day.

I'm not compelled to re-invent stats with poor/inaccurate data (for the stated purpose) just to post stuff on message boards. Given the necessary to data to calculate the proper coefficients (which I'm not sure exists for the levels I'd care about), it ought to be fairly easy. But it's easier still to look at K/BB/HR rates, which I can essentially do in my head for the ones I'm interested in.

…But it's easier still to look at K/BB/HR rates, which I can essentially do in my head for the ones I'm interested in.

And therein lies the tale. I’ve been doing this for almost 15 years now. I don’t do it for the coaches, the players, causal fans, or even message board wonks. I do it for the parents because its pretty much their last chance to be connected to their kid on this kind of personal level, and I can tell you a lot of parents like looking at the numbers because they are interested.

I know those numbers don’t mean much when it comes to moving on in the baseball world, but most parent don’t. They’re gonna talk anyway, so I give them something to talk about, rather than guess. I really don’t much care what the message board people say about what I do, but I do care what the parents say, and I can assure you that they appreciate what I do, and that’s my reward.

There is an additional benefit that I really enjoy as well, but I’ll keep that to myself.

The parents who understand advanced metrics (even when properly calculated) well enough to put them in the correct context don't need/care about them at the HS level. Giving people who still think squishing the bug is a good plan FIP is going to cause more problems than it could ever possibly solve. Hell, you can't even convince a lot of people who think they understand the game that the underpinnings of FIP are important/accurate.

Originally Posted by jacjacatk:

The parents who understand advanced metrics (even when properly calculated) well enough to put them in the correct context don't need/care about them at the HS level. Giving people who still think squishing the bug is a good plan FIP is going to cause more problems than it could ever possibly solve. Hell, you can't even convince a lot of people who think they understand the game that the underpinnings of FIP are important/accurate.

Its not about the numbers, its about parents and their kids and a way to give them that last connection. If you don’t want to take the time to do it, don’t. It won’t bother me.

You’re so woefully misunderstanding me and what I do, it’s a pity. I’m not trying to “solve” any problems. All I’m trying to do is allow people to have more in their baseball experience than batting average, ERA, and wins and losses in connection with playing at the next level. Baseball is about much more than that.

Do you really think that the people who are looking for "that last connection" are looking for FIP? Maybe baseball parents in your neck of the woods are different than in mine, but around here that last connection has literally nothing to do with numbers on a piece of paper. My son and I are both inveterate baseball stats-geeks, but when it comes to his playing and the connections we've both built because of it, the numbers mean almost nothing.

Beyond that, the parents interested in stats are almost universally interested only at the surface level. If Harold Reynolds isn't a fan of the stat in question, they've never heard of it. The Brian Kenny fans don't need anyone to calculate FIP for them, anyway (or are aware enough to realize it's not especially instructive with regards to HS stats).

But hey, go ahead and build hsbaseballreference.com for your parents, if that's what floats your boat. But if you're going to pontificate about it online while misapplying the math, don't be surprised when people attempt to point it out.

Originally Posted by JCG:

Aren't your team's and league's HR numbers significantly different before and after BBCOR?  They sure are around here.

As you can see, our hitters have actually done better while our pitchers have too. But the thing is, HRs are so infrequent in HS around here, I don’t see BBCOR being that much of a factor.

Hitters

62 Total

24 2007 – 2010

38 2011 – 2014

Pitchers

44 total

29 2007 – 2010

15 2011 – 2014

To tell the truth, I haven’t checked our league’s totals, but I can’t imagine they’re much different than ours. We just don’t see that many HRs. There are some teams in the area that do hit more than most, but not very many.

Your hitters have hit 58% more HR since BBCOR than before it? While your pitchers have allowed 48% fewer?

In any event, the fact that they haven't changed substantially in raw numbers isn't really the issue.  The issues are how much have they changed on a rate basis, and how does the current rate basis compare to the MLB one since you're relying on the MLB FIP coefficients. In the AL HR occur in 2.4% of PAs. I'm guessing based on the length of the typical V schedule here, but the HR numbers you're listing should result in something south of half the AL rate, which is going to make a difference in the math you should be trying to do.

Originally Posted by jacjacatk:

Your hitters have hit 58% more HR since BBCOR than before it? While your pitchers have allowed 48% fewer?

Yes. So what? Because team personnel never last more than a year at a time, why would anyone be surprised at wildly varying numbers? Arguably the “best” team in NorCal hit 24 in 2009 with BESR giving up 5, 37 HRs in 2010 with BESR giving up 4, 24 in 2011 giving up 6 with BBCOR, 10 in 2012 giving up 4, 10 in 2013 giving up 1, and 1 giving up 3 in 2014.  

So was it the bats, the hitters or the pitchers that affected the HR rate? One player on offense or defense can really change the numbers.

In any event, the fact that they haven't changed substantially in raw numbers isn't really the issue.  The issues are how much have they changed on a rate basis, and how does the current rate basis compare to the MLB one since you're relying on the MLB FIP coefficients. In the AL HR occur in 2.4% of PAs. I'm guessing based on the length of the typical V schedule here, but the HR numbers you're listing should result in something south of half the AL rate, which is going to make a difference in the math you should be trying to do.

Ours is 0.8%. 62 HRs in 7,685 PAs. And?

Why don’t you concentrate on ML numbers and stay away from everything else since you don’t care about them nor do you want to do anything to improve on what’s available or being done?

FWIW... I have discovered, all too slowly, that pointing out mathematical, statistical, or baseball realities to Stats re his 'presentations' is much like stopping by the neighborhood Kool-Aid stand to explain best business practices to the kids... A) they don't comprehend and B) since they're basically playing make believe to begin with, they don't care or much need to know.

Guessing any HS parents drinking Stats' Kool-Aid are doing so much like those neighborhood parents at the Kool-Aid stand... With a kind smile on their faces. After a supportive token purchase, better to simply wave and a give a big thumbs up whenever you happen to drive by thereafter...

You're the one using the MLB coefficients, not me, you sort it out.  But if you're at 1/3 the MLB rate that's substantial, and acting like it doesn't matter makes what you're doing with FIP fairly pointless.

As to the BBCOR effect on HR, they declined dramatically around here, though to be fair there are a few bandbox fields at some of the HS, so that could make quite the difference. If you're honestly shocked at the expectation that the bats would be a big factor in HR rates and that HR rates being higher for your team post-BBCOR would be unusual, you really need to re-examine your expectations.

Talk about history repeating itself........

Originally Posted by jacjacatk:

You're the one using the MLB coefficients, not me, you sort it out.  But if you're at 1/3 the MLB rate that's substantial, and acting like it doesn't matter makes what you're doing with FIP fairly pointless.

As to the BBCOR effect on HR, they declined dramatically around here, though to be fair there are a few bandbox fields at some of the HS, so that could make quite the difference. If you're honestly shocked at the expectation that the bats would be a big factor in HR rates and that HR rates being higher for your team post-BBCOR would be unusual, you really need to re-examine your expectations.

You’re a hoot, that’s what you are. You tell me not to even bother using the coefficients and use straight rates, then you chastise me for using the same coefficients they use in the ML.

I have nothing to sort out because I don’t use the numbers the way most people do. I’m not trying to do anything other than give people the best way possible to compare the players on our team. I realize there are others who think it’s a fruitless activity, but those aren’t the people who have kids playing in our program.

I’m not at all shocked that BBCOR bats affect the HR rates. But what difference does it make? I’ve computed the league ERA from 2014. If I was to compute FIPs as a regular thing, they would only be for the season after I’d computed ERA for.

Look, you can pi$$ and moan all you want about what I do, but I couldn’t care less because you mean absolutely nothing to me and what I do for the people I service. As I’ve said, if you can do better, why don’t you do it and I’ll learn from the master. If you can’t, why not just let it go? Just exactly who am I hurting that you’ve taken such an interest in making me a project?

I guess I’ll just do what I’ve done before with you. I honestly attempt to communicate with you, but you aren’t interested in communication. All you want to do is prove to everyone how brilliant you are and how stupid I am, so I’ll just let this be the last time I bother with you.

I told you not to bother using figuring the FIP constant for your league because it doesn't matter, especially if you're going to use coefficients from another league. I told you to use the straight event rates for your league because you're using coefficients from another league that need to be calculated based on the rates in your league in order to be accurate.

HR rates being different from MLB for the data you're using are one example of those issues. You're using the MLB coefficient for a league that it's not appropriate for. The pre/post-BBCOR thing was just an aside.

If you're going to post your methods to a discussion board as though you've made some fundamental new discovery or insight, be prepared to defend them and/or learn from constructive criticism.

Or whine about how you're right and everyone else is wrong, while providing inaccurate and/or useless information to the "people you service". Doesn't really matter to me.

This is the issue.  The constants (13 for HRs, 3 for BB + HBP and 2 for K) are all calculated for MLB players.  These would be different constants for a different level of play.  FIP is not really accurate for a level using a different level's constants.

Originally Posted by bballman:

This is the issue.  The constants (13 for HRs, 3 for BB + HBP and 2 for K) are all calculated for MLB players.  These would be different constants for a different level of play.  FIP is not really accurate for a level using a different level's constants.

You guys talk as though creating a table of linear weights is simple, given that its next to impossible to get much data to work with. Would I like to have linear weights for HS? Sure would. Perhaps you have a way to do that. I’m all ears.

Stats- If access to HS data to create linear weights isn't available, why extrapolate constants from another league? Just ignore the constants and use the raw data. The results will read differently than those you'll see at the big league level, but will be more accurate for comparing the players that created the data. I think that's what jacjacatk is trying to point out - using incorrect constants makes the data less accurate among the sample of participants you're analyzing.

Which is the exact same argument you typically use for rejecting the comparison to or use of MLB stats in other contexts where it is actually useful.

http://www.fangraphs.com/commu.../why-i-dont-use-fip/

Quoting from the conclusion to that article:

This in fact may be the ultimate problem with FIP. On its own it doesn’t give us any information. Even with the most extreme differentials we always have to look to other statistics to draw any conclusions. So why don’t we make things easier and just look at those other statistics to begin with instead of trying to draw conclusions from a flawed stat with incorrect parameters?

       

No one said it's easy stats. It's a pain in the a$$. That's part of why no one wants to do it. If you want to use these metrics though, you have to do it correctly. If you are using incorrect numbers, then the results are useless. You can make things look anyway you want if you use whatever numbers you want.

You keep saying for others to do the work if they don't like what you are doing. The people here don't want to go thru the calculations and use these metrics for HS players. You do. All people are saying is if you are going to do the calculations and use them for whatever you want to use them for, use the correct numbers. You asked for feedback, that's what you are getting. Nothing personal. It's all about the numbers.

Originally Posted by J H:

Stats- If access to HS data to create linear weights isn't available, why extrapolate constants from another league? Just ignore the constants and use the raw data. The results will read differently than those you'll see at the big league level, but will be more accurate for comparing the players that created the data.

That’s exactly what I’ve done for all these years! It wasn’t until just recently I acquired the capability to come up with the “League ERA”, so I stuck it in there and took a look at what shook out. I was surprised to see that even using the MLB constants for HR, K, and BB, the results showed of the top 4 pitchers in FIP, 3 of the 4 kids we’ve had make it to the pros were there. That sure seems to be a pretty good predictor, even if predicting using HS numbers isn’t something I’d ever recommend. But it sure is quite a coincidence.

I think that's what jacjacatk is trying to point out - using incorrect constants makes the data less accurate among the sample of participants you're analyzing.

I’m not mentally deficient. I understand that doing what I did isn’t the “correct” way or even a good way to come up with the result. But it is a way, and no matter how bad, it isn’t going to be off that far, and sure appears to be at least as good a way to compare players on the same team as anything else.

I’m not trying to get scouts or college coaches to stop what they’re doing and glom onto the number I come up with! In fact, unless one of those people contact me personally, they won’t have access to that number. I do this stuff strictly for the entertainment of the coaches, players, and parents in our program, and to be able to evaluate whether something some talking head claims is fact or fiction.

Those folks like what I do and I try to give them an honest effort in doing it. I spend literally thousands of hours every season devoted to helping the program where I can and providing this stuff. What do people like jacjacatk and SG do? They could help me improve what I do, but instead all I get is derision, then get told what I do makes no difference. If it makes no difference, why bother judging me in the 1^st place?

In the end, nearly everything I’ve read on FIP says HRs which are a major part of the formula need at least 2 seasons worth of work for a ML starter in order to be of real value as a reflection of a pitchers talent to prevent HRs. Assuming that to be true, there’s no HS pitcher who’s ever going to get that much work, so the metric becomes pretty useless just on that, which is another big reason I’ve never generated it. When you mentioned it in a different thread, I went back and looked at it again, and with the ability to come up with at least a semblance of the main factor, I played with it a little bit, and got a kick out of the result. I had no idea I was going to kick off such a firestorm for playing with numbers.

We get it, when other people compare to MLB numbers, they're wrong.  When you do it you're right. Whether or not you do it an a statistically valid way.

Originally Posted by bballman:

No one said it's easy stats. It's a pain in the a$$. That's part of why no one wants to do it. If you want to use these metrics though, you have to do it correctly. If you are using incorrect numbers, then the results are useless. You can make things look anyway you want if you use whatever numbers you want.

I’ve never used it or many other more advanced metrics because I couldn’t use good factors. The main reason is, there’s no way to generate a table of linear weights for run expectancy by base/out state, men on base state, or any other state. As far as I know, I’m the only one who could do it, but to do it for only 1 team isn’t something even I would do.

When I run the numbers for FIP, I wasn’t trying to make them look any special way, and it’s a bit insulting to have someone insinuate that I did, or would even think of doing that. I don’t have a lot going for me, but I do have some integrity.

You keep saying for others to do the work if they don't like what you are doing. The people here don't want to go thru the calculations and use these metrics for HS players. You do. All people are saying is if you are going to do the calculations and use them for whatever you want to use them for, use the correct numbers. You asked for feedback, that's what you are getting. Nothing personal. It's all about the numbers.

You’re right. In general the people here don’t want to see any kind of improvement in using metrics on HS players, and it wouldn’t make a lot of difference if Tom Tango himself were doing it. There are some who don’t want to see anything new or innovative. It may only be a very small percentage of people who are like that, but it doesn’t take a lot of people to throw a wrench into a gearbox.

FWIW, by the time I’d gotten the final numbers I posted here, I’d already decided FIP wasn’t going to be something I produced, even if I did have the ability to generate all the correct constants. I’d forgotten about the need for a sample larger than a HS player could possibly produce, but after reading a lot about it again, I was reminded. So, no one has to worry about me producing a bunch of illegitimate information that will ruin HS baseball forever. To me its just a matter of deleting a few small files and its gone forever.

So all the arguments you've been making have been to support the way you did something when you had already decided it wouldn't work anyway, in ways largely similar to those the people you were arguing against were trying to explain?

Brilliant (which you can define in the same unique way you do derision and criticism).

Well I have to admit... for me it's become a little bit personal with Stats. This is because I have come to believe that he is in reality just one more troll on the old baseball website. I don't believe that his motivations have much to do with developing insightful statistical metrics... I believe his posts are mainly intended to enable Stats to play Stats.

Trying to debate and/or discuss points logically with him is a true fool's errand. Most trolls are obvious enough... they make inflammatory statements for a period of time until they realize that this really isn't that type of forum, so they move along. Stats, on the other hand, throws out odd bits of random baseball-like premises... usually couched specifically for maximum controversy, thus insuring more chatter... and always looking for someone... Anyone... to take the bait. And then it's the familiar steady descent into him pontificating about whatever twisted bit of data he was wanting to gab about in the first place... weaving in, generally, his invaluable service to the parents and coaches... yada yada yada. To me, it's the boy who cried wolf (way) too many times... Even when he occasionally brings something up that might have some actual merit, I know from experience that it will quickly degenerate into just one more long-winded romp through the ditch.

Having followed along through that ditch a few times personally, I know where it leads: always nowhere. I've made it a point on a couple of occasions to pin Stats down on his specific numbers and calculations. It's not that I have all the answers or consider myself some statistical wiz... far from it... But I do know my way around the subject pretty thoroughly and so have made the honest effort to talk it through with Stats on a couple of occasions. What I found was a guy who simply cannot or for whatever reasons will not digest demonstrably correct statistical, and even mathematical, realities. As you drill past all the homespun theory and anecdotes to try to walk step by step through his actual data... Stats starts doing back flips to change his arguments. Eventually, when bluster and posturing (or boredom) don't get him off the hook, he shrinks down to size and resorts to putting on the pity act. I've read along as others have done the same thing with similar results. I've debated numbers on occasion with others here including jacjac, who I tend to disagree with almost entirely! BUT... at least when debating specifics with him and others, the goal posts never move. It's simply a debate. You know... he makes a point, you refute it, he refutes that, and back and forth you go... Each point relating to the previous one. With Stats, it's not about discussion, analysis and trading viewpoints... I believe it's all just an excuse for him to generate a forum to blather to in a mock authoritative voice. End of story...

I have to hand it to you though, Stats... trolls come and go here, but you're clearly determined to maintain that role for the long haul.

Edited: Case in point, the post above. Another circular argument dissected... Stats making it up as he goes per usual... and just as he will do next time. If you want to spend time debating statistics with Stats, just know going in that it's the debate he wants... not the statistics.