Neil Weinberg FanGraphs Q&A – 9/10/14

2:28
Neil Weinberg: Hey all. It’s Wednesday afternoon, so let’s chat.

Remember, this chat focuses on questions about advanced stats, FanGraphs data/features, etc, but you’re more than welcome (encouraged?) to ask regular baseball questions as well. Fantasy and prospect questions aren’t banned, but I’m just not very good at answering them.

I’m @NeilWeinberg44 on Twitter if you want to reach me other times.

Also, my dog usually isn’t at home on Wednesdays, but he is today. That may mean a slight “please take me outside” delay, or it might not. I’ll make up for any pauses on the back end.

3:00
Neil Weinberg: Alright. Let’s discuss the events of the day. Current dog status: sleeping

3:00
Comment From Mike
why did so few teams recognize Abreu’s abilities as a hitter?

3:01
Neil Weinberg: I don’t think anyone did. Not even the White Sox. Players who come up against competition different from the normal MiLB are harder to judge because you have no idea how his numbers and performance is skewed by the opponents he faces.

If I got to play against 8 year olds, I’d look great. I don’t think we’re good at determining the talent of all Cuba opponents.

3:01
Comment From Guest
Does a foul out that occurs in the infield count as an IFFB? For example, a ball count one step in foul territory next to first base.

3:01
Neil Weinberg: Theoretically, yes. Although it’s my understanding that it is at the discretion of the stringer to call something a “popup” or not

3:01
Comment From Miller
How are the different permutations of RE24 calculated? I understand how an inning would start with .461 runs expected, but don’t know how it changes from there

3:03
Neil Weinberg: It’s based on the average number of runs the league scores based on reaching those situations. I’m not sure if we use just that season or last 365 days or what, but it’s average runs scored after reaching that state

3:03
Comment From CStat Guy
When using a custom report with certain fields, why do the fields revert to “Dashboard” when I change to some of the “Splits” (i.e. 2nd Half)? Thanks.

3:03
Neil Weinberg: I assume it’s just a glitch of some nature. Report it in the contact field!

3:03
Comment From Vin
We love you Neil!

3:04
Neil Weinberg: That’s nice of you to say. I secretly hope you are Vin Scully.

3:04
Comment From Guest
More of an observation: The Steamer RoS numbers + the season so far do not add up with the Steamer total numbers.

3:05
Neil Weinberg: Yeah, I think this is because one of the data streams is coming from an outside source and one is use calculating based on inputs we get from them. Noted!

3:05
Comment From kid
Hamilton or deGrom. Can Billy win ROY with a sub .300 OBP and a -3.1 OWAr?

3:06
Neil Weinberg: I am starting to think deGrom, but I will admit I haven’t done a deep dive into either player in a month or two. ROY is typically based on name value. Hamilton is an interesting case because he’s all defense and speed, which voters usually don’t reward.

3:06
Comment From dom
Why isn’t WPA discussed more in MVP conversations?

3:07
Neil Weinberg: I don’t think WPA is a very useful tool in this regard. Rewarding a guy because he happened to get his hits when the game was on the line doesn’t really tell you anything about value. RE24 is a slightly different story, although I’d still prefer context neutral

3:07
Comment From hscer
if “Rookie of the Year” were named “Most Valuable Rookie” would team contention magically become a criterion for a lot of voters

3:08
Neil Weinberg: I think it would. You can guess my view on this.

3:08
Comment From CStat Guy
If you are interested in looking at hitting/pitching consistency month to month, which stat(s) would you use?

3:10
Neil Weinberg: For pitchers, I’d look at components. So K, BB, HR, GB, BABIP. We can yell lots about the whole “who gets credit for a batted ball” thing, but we definitely don’t want to judge pitchers based on sequencing if you care about consistency.

Hitters, I’d look at basic plate discipline stuff. Maybe something like ISO and BABIP too.

I’m assuming you’re trying to look for underlying performance in answering this question rather than just how well they hit.

3:10
Comment From Zach
Will we see Rusney Castillo in the major leagues for all of next season, with Jackie Bradley Jr back in AAA? Or will the Red Sox have Castillo get his feet wet in the minors for a while?

3:10
Neil Weinberg: I bet Castillo is pretty much a regular right away. JBJ probably gets dealt?

3:11
Comment From Scott
In the post the other day about uncertainty in WAR, Dave illustrated batters getting 57% of available WAR and pitchers 43% by pointing to fielding as having an impact on run prevention, explaining the 7%, and the 50% being run production. How was it derived originally that run production is responsible for 50% of WAR and prevention the other 50%?

3:12
Neil Weinberg: This is sort of tautological. Every run scored is positive for one team and a negative for another. There is one run and two teams.

(I don’t mean for this to be a snarky reply at all, FYI.)

3:12
Comment From Pennsy
Is there a stat that account for well-roundedness of a player? Something that speaks to a lack of general deficiencies versus a single elite skill making up for other shortcomings? (Ortiz’s power, Hamilton’s speed, so on versus a player like McCutchen, Rendon). Or is there not any use if knowing a players skills are well-rounded or not?

3:13
Neil Weinberg: This is a very useful thing to know. I don’t think there is a stat that tells you this thing. You don’t really care if a player is well-rounded or not in terms of who is better, but sometimes we care about the ‘how’ question.

Best bet would be to look at rankings or standard deviations relative to each part of a player’s game

3:13
Comment From haishan
Something I’ve been wondering about: Why isn’t replacement level different for different positions? Isn’t it almost definitionally easier to find a Quad-A player who can DH than one who can play shortstop? Or is this baked into the positional adjustments?

3:16
Neil Weinberg: Positional adjustment.

A replacement level DH(avg runner) would hit much better than a replacement level SS (avg run, avg field).

3:16
Comment From IsIt2015Yet?
Couldn’t something equivalent to hard hit balls data + PitchFX + Field FX help us compare Cuban baseball to MLB? And Korean and Japanese and Mexican for that matter? Some sort of league-aggregated metrics that’d tell you how to generally discount performance by league?

3:17
Neil Weinberg: We don’t have that data for Cuban baseball. Really you want a long track record to develop good translations. We don’t have enough Cuban players to be great at this. Someday, we should. Just not yet.

You can use scouting to get a sense, but scouting comes with uncertainty, just like stats. More info is always better, but you’re never going to be right and teams would rather miss on a great player than overpay a bad one

3:17
Comment From Guest
What are the advantages of wRC+ over wOBA and vice-versa? Is one stat “better” than the other?

3:19
Neil Weinberg: wRC+ is park adjusted, which is better. wRC+ also compares to league average, which is good if you want to talk about players across seasons and run environments.

Typically, wRC+ is superior. Although, it’s easier to use wOBA in a pinch sometimes and you can compare differences among run environments if you like.

3:20
Comment From Lucas
Couldn’t we use the same type of metrics in UZR to evaluate pitchers for their batted ball profiles? Let me try to explain what I mean. In UZR if a ball is hit hard to x zone on the ground it has y amount of runs attached to it. If the fielder makes that play then he is awarded z runs. Why don’t we just reward or punish the pitcher for those run values (average run value based on the batted ball)? You could combine that with FIP to get a pitchers WAR and it would be still be defensive independent.

3:22
Neil Weinberg: There is something to this basic idea. If you allow a ball to be hit to X at Y speed, what is the average expected run value of that play?

3:23
Neil Weinberg: Hopefully StatCast data can help us understand batted balls better.

3:23
Comment From Doug
I was trying to do some research on BABIP decliners from 1st to 2nd half. I found the pages on FG and exported them to excel, but it was tedious matching player ID’s/names then just browsing the list for noticeable drops. Easier way to do this?

3:25
Neil Weinberg: I haven’t found a super efficient way to automatically line up two lists in Excel if they don’t contain the same exact players (anyone?). But I always create another column and do BABIP1-BABIP2 and then sort. I know how to do the merge thing in a different stat program for my other job, so I’ve never looked too hard

3:25
Comment From Lucas
What would a Tigers extension on Porcello look like? I would think similar to Homer Bailey. 5 or 6 years 85 to 110 million?

3:27
Neil Weinberg: My guess is a 5 year deal is probably $90 million, 6 is probably $110 million or so. If he gets to FA, it’s obviously more and I don’t think he signs one unless they dial it up quickly

3:27
Comment From Cman
Curious, I know the difference between Baseball Reference’s WAR for pitchers is based on ERA and Fangraphs on FIP. What is the difference between the two sites in regards to position player WAR?

3:28
Neil Weinberg: Slightly different measurements for the components. They use DRS for defense, we use UZR. They use a different baserunning metric, slightly different batting, position, etc. I was looking today and most of the major gaps are defense.

3:28
Comment From Scott
re: tautological 50%s. If I were to argue (as many do) that pitchers have a greater influence over the outcome of a plate appearance than a batter, wouldn’t it then make sense to pick a different starting point than 50-50? Of course, that argument may be wrong, but I want to see the 50-50 split supported.

3:30
Neil Weinberg: If a pitcher allows a home run, he allowed one run and the batter contributed one run. The pitcher didn’t allow more runs than the batter scored.

I understand your point about a PA, but I’m not sure there is evidence that it is true. And a 50/50 split should be a default belief.

3:30
Comment From Lucas
Wouldn’t it make sense to have positional adjustment be different for every year? Adjust it so that all positions come out with the same WAR per 650 PA.

3:32
Neil Weinberg: There is a case to be made that positional adjustments should vary year to year, but the problem is that you need multiple years of data to create them, so it’s never going to be a true one year estimate. I think there is some evidence that LF need to get docked more and C need more of a boost, but I’m not savvy enough to run superior calculations. If anyone out there is, I’m all ears.

3:32
Comment From Skuggs
Is there any stat that incorporates UZR into pitcher WAR? In other words, we have this stat, UZR, which tells us how often different types of batted balls were turned into outs. Can’t we (and by we I mean probably you), just use UZR to determine how much credit the pitcher/fielder gets for every play instead of arbitrarily saying “all” or “none”? I guess this would function as a more encompassing xFIP, which instead of just looking at league average home run to fly ball rates, would look at all league-wide batted ball data and give us a better idea of what *should* have happened on every play. Seems to me the data is all there, just needs to be incorporated.

3:33
Neil Weinberg: Similar to the previous question. Two points. This stat does not exist publicly, at least, right now.

There’s also some question as to how exactly to carry this out. I think the basic question you’re asking is a good one. Just not sure we’re capable of putting this together super well just yet.

3:34
Comment From Guest
Can anyone explain the SMASH statistic? I have heard it referenced but can find any information on it

3:34
Neil Weinberg: I have no idea what this is. I assume it’s a stat someone made up for an article!

3:34
Comment From hscer
your view of Passan-Cameron-Forman?

3:38
Neil Weinberg: I think Passan did more harm that good and doesn’t have as good of a grasp on the situation as he thinks he does. (PS: I said this to him directly, lest you think I’m passive aggressive).

He presents a world in which saber-writers treat WAR as gospel. We don’t. We just don’t. He acts like we aren’t aware of the flaws and don’t have any interest in fixing them. That’s not true. No one understands the limitations more than the people who created them.

His real beef is that we don’t do enough to teach people the proper way to use WAR. Which could be a valid argument (hey look FG hired a guy whose only job is to literally explain the stats), but it’s not the one he makes.

I also don’t think he has a strong foundation of skepticism on defensive stats. His argument is basically, “that doesn’t seem right.” He might be right, but you have to come to the table with more than that.

All in all, he has a platform in which he could educate people on how they should be cautious about WAR and he went the combative route. I think that hurts the movement of knowledge more than anything

3:39
Comment From Gila Monster
What are your thoughts on error bars for WAR? As of right now, WAR is very precise. It calculates WAR in a methodical manner that is uniform, but seemingly at the cost of accuracy because we probably shouldn’t calculate WAR the same way for everyone. Couldn’t we just have error bars that get smaller as a skill has been shown like defense, including weak contact, a high LOB%, or a high BABIP?

3:41
Neil Weinberg: Here’s the skinny on this. Any time you create a stat or metric that includes estimation, you have uncertainty.

We know a player had X hits and Y outs. Those are not up for debate. But when we talk about run values, we’re estimating something based on other data. We aren’t going to get it perfectly right and some plays will systematically have more uncertainty.

This is built in. We know this.

Problem is that the general public are bad statisticians. We do a horrible job teaching this stuff in our schools. That doesn’t mean we shouldn’t create good stats, but it’s a big problem for getting people to use them properly

3:41
Comment From Matt
Why is the MLBPA such as strong union compared to other sports? I imagine a work stoppage in the NFL would lead to the country falling into a state of anarchy and costing the league billions even if it only last a couple weeks. But NFL players have far fewer rights.

3:42
Neil Weinberg: I don’t really know the answer. I imagine it has something to do with some accident of history and Marvin Miller. Just don’t know enough about the NFL to say

3:42
Comment From Spark
Did you ever get a chance to look into why fangraphs lists all the components of an infielder/outfielder’s DRS (rGDP, rARM, rGFP, and rPM) but only 2 components of a catcher’s DRS (rSB and rGFP)? With the information available, the DRS totals for catchers don’t make sense. For example, Russell Martin and Welington Castillo both have 6 rSB and 3 rGFP, but Martin’s DRS is 10 and Castillo’s DRS is 5. There seems to be something missing. Even if fangraphs is not allowed to list that information for some reason, do you at least know what the missing areas are, so I can guesstimate where the difference between guys like Martin and Castillo is coming from?

3:43
Neil Weinberg: Didn’t forget. Just had a very busy week and haven’t gotten the answer. Keep asking, eventually I’ll come through for you!

3:43
Comment From Well-Beered Englishman
If a reliever pitched a literal perfect season, what kind of WAR could they post? What kind of season would it take for a relief pitcher to EARN an MVP?

3:43
Neil Weinberg: Intrigued.

3:43
Neil Weinberg: So you would need about 6 WAR to be considered for MVP

3:43
Neil Weinberg: Betances has 3 WAR right now, yes?

3:44
Neil Weinberg: Could 90 innings of like a 0.60 FIP do it? I’d have to look at it longer, but maybe.

3:45
Comment From Paul
If you look at the season splits among starters after the All Star Break, Chris Tillman, Wei Yen Chen and Miguel Gonzales are among league leaders. How well do you think this performance can translate into the post season? Or would a team like the Tigers have the advantage still?

3:45
Neil Weinberg: Tigers still have a starting pitching advantage. Although bullpens play up in the playoffs, so….

3:46
Neil Weinberg: If you don’t know i’m a Tigers fan. We have a very troublesome relationship with bullpens.

3:46
Comment From Matt
Could we create something along the lines on a xWhiff% on pitches using pitch f/x data? We know velocity is good. We know movement is good. We know seperation is important for changeups. Granted it would be a very crude tool, but it is curious on how the correlation would turn out.

3:47
Neil Weinberg: You could! I don’t know how accurate it would be. But it’s very possible. Velo, movement, location, handedness, count, and previous pitch? Throw that in a model and see.

3:47
Comment From Tripp
What do you make of Kevin Kiermaier’s current 57.7 UZR/150 in RF in 526.1 innings? The highest qualified season has been Brett Gardner’s 45.8 in 2010 (LF, 906 innings).

3:47
Neil Weinberg: Hold on. Let’s look at this.

3:48
Neil Weinberg: This is amazing.

3:49
Neil Weinberg: Two responses: One, Kiermaier gets to lots of baseballs, so I’m not surprised he grades out well. Two, sample size matters a lot! Let’s talk about that…

3:52
Neil Weinberg: Defensive stats have two issues. One is measurement error. We aren’t great at determining exact location, speed, positioning, etc. Sometimes we get stuff wrong! Room for improvement, no argument.

But sample size is a confusing one for many people. Small samples are a FACT of baseball when it comes to defense. You simply don’t get many chances to field tough plays. As a result, if you happen to nail 8 in a row or screw up 5 in a row, that will have a huge impact on your rating.

This is NOT a flaw of defensive metrics. This simply means a SSS of defensive stats might not tell you much about that player’s true talent because of random variation.

3:52
Comment From Cman
In regards to Dougs question. You could run VLookup function which can pull data from one sheet to another, say, based on player name. That could take a lot of the work out? Not sure if this helps or not.

3:52
Comment From El Guapo
RE: Doug’s excel question. The “lookup” functions are a good way to do this. But make sure to convert the info you are searching into a table first

3:52
Comment From Lucas
Explain someone like Odorizzi. His BABIP and LOB% are really close to league average yet his FIP is one-third of a run lower than his ERA. Random sequencing.

3:54
Neil Weinberg: Seems to be much better with the bases empty. That’s probably nothing, but if your bad moments all happen with men on base, more runs will score than if they are distributed evenly. Might be something else, but that’s the first glance

3:54
Comment From John
I’m a Jays fan living in Toronto and a lot of my buddies just keep raving about Sanchez as a starter, how do i explain that hes not that good?

3:55
Neil Weinberg: There is basically no way to win this argument without him failing. What I recommend is creating a record of your specific concerns (Blog, email, FB message) so that you can show them you are right in a few years.

3:55
Comment From Hank
FIP- is adjusted based on park RUN factors correct? What happens in parks where the home run and run factors are very different (Fenway, Yankee stadium etc) as HR’s drive that # quite a bit as well but the run factor adjustment may work in the opposite or lesser direction.

3:58
Neil Weinberg: So more HR leads to more runs overall, yes? I’m not exactly sure on your specific question. Can you rephrase it so that I have a clearer handle on it?

3:58
Comment From Robert Hombre
Question about Park Factors, and I’m sorry if it’s based on incorrect premises: WAR uses some version of FIP- as AN input, as I understand it. What are the park factors used for FIP? Do the park factors include only TTO effects, or do they use holistic park factors?

3:58
Neil Weinberg: Entire park factors.

3:59
Comment From Well-Beered Englishman
I made up the stat HULKSMASH for an article a few months ago. It stands for Hitting Ultra Long, Knowledgeable Stat Measuring Ability to Stroke Homers. It is the same as ISO.

3:59
Neil Weinberg: Well done.

3:59
Comment From Jake
What stats are the best starting points in finding comps for relief pitchers for simulated arbitration cases?

3:59
Neil Weinberg: I think you’re mostly looking at innings, ERA, and saves. Maybe strikeouts? Good question for Matt Schwartz if you can find him on Twitter

3:59
Comment From Chief Yahoo
What’s the other stat program you use besides excel?

4:00
Neil Weinberg: I use STATA for more complicated data stuff. I know how to use R but I don’t like it, nor really ever need it.

4:01
Comment From Scott
Maybe I’m not being clear about this 50/50 thing. What I’m asking is, since the exact split has a massive impact on WAR, is there anything I can read that attempts to prove this? I don’t agree that simply because there are two of them we should assume it’s 50/50.

4:04
Neil Weinberg: I think, perhaps, we’re talking about different things. Run scoring is 100% offense, correct? And run prevention is 100% pitching and fielding, correct?

Each run is both scored and allowed. So every run goes in both camps. I don’t really see how it could be different from that. I appreciate your inquiry, I’m just not sure it makes logical sense for anything else to be the case.

And I don’t have anything off hand to link you to, sorry!

4:04
Comment From Guest
Not strictly a baseball question: But why does is the NFL getting so much crap over the Ray Rice incident, but not the legal system? Isn’t it the legal systems responsibility to rather evidence and convict criminals? See Matt Bush. I mean if Mike Trout drowned kittens for fun, I would still enjoy watching him play baseball….

4:05
Neil Weinberg: I think we have different standards for the legal system because in that case we are talking about removing a person’s rights/liberty. Whereas with sports, it’s about getting the person off television. The legal system should be criticized for how it handles certain things, but pro sports don’t have to meet the same level of proof to act so they get criticism on more cases? Maybe.

4:06
Comment From Paul
I found a couple of references to Smashing Percentage (SMASH) which is Slugging Percentage Divided by Average. Supposedly a metric for power with a cute name. I bet I could invent a metric that Hitting Unless Struck Out (HULK) + (SMASH) and Chris Davis would be at the top of the list.

4:06
Comment From CuriousGeorge
are both WAR and WrC+ adjusted for Era? (cant think of how to word this) like league averages. So comparing babe ruths war to barry bonds is neutral

4:06
Neil Weinberg: Yes.

4:06
Comment From Gila Monster
Eric Gagne had a close .86 FIP over 82IP generating 4.4 WAR, so that sounds about right. Maybe a lower FIP would be needed.

4:07
Neil Weinberg: Gagne did that in a tougher run environment. Also an easier park than my Yankees example.

I’m going to revamp our WAR explainers. I’m hoping to build a calculator that can answer this in five seconds, but haven’t gotten it right yet

4:07
Comment From Jake
What are the first stats I should use to sort through relief pitchers to find comps?

4:07
Neil Weinberg: Depends on what you want. RE24, FIP, K%, BB%, GB%, maybe swinging strike rate

4:07
Comment From Timmy and Tommy
What kind of a miracle would it take for Sano or Buxton to debut

4:08
Neil Weinberg: This year? All of the miracles

4:08
Comment From hscer
if Kershaw leads the NL in strikeouts this year despite his missed time (currently 5 behind Strasburg and 3 behind Cueto), he moves from unfair to ____

4:08
Neil Weinberg: Kershawian?

4:08
Comment From Timmy and Tommy
Where would one go to see how a minor leaguer done over the past 15 or 30 days? It’s SSS, but it also would show tweaks.

4:09
Neil Weinberg: We have minor league game logs! Go to player page, click game log!

4:09
Comment From Timmy and Tommy
Is Stroman or Alex Wood more likely to keep their numbers up next year?

4:10
Neil Weinberg: Toss up?

4:10
Comment From RE:John
When some raves about a prospect as a starter and they don’t possess an elite changeup, call them a reliever. It will piss off your friends and you’ll be right 75% of the time

4:11
Neil Weinberg: Doesn’t have to be elite. Just good. But yes. This is a fun game

4:11
Comment From Noah
Why are pop ups not included as automatic outs in FIP? It seems like such an easy solution. Fly ball pitchers (especially extreme fly ball pitchers) still underrated by FIP (and xFIP).

4:11
Comment From Noah
Why are infield flies not included in FIP?

4:12
Neil Weinberg: We didn’t invent FIP and it was invented before this data was easily available and evaluated, I believe. However, we do treat IFFB as automatic outs when we calculate WAR. So while it doesn’t show up in the raw FIP you see, it does go into the WAR value we spit out.

4:12
Comment From Seedy Alleyway Mugger
Backdoor cutters and high-fastballs are apparently the secret of unlocking Drew Smyly’s potential.

4:12
Neil Weinberg: He’s had a nice run. Curious to see this approach all year in 2015

4:12
Comment From Hank
As an example Fenway… very high run (above league average) factor; below league average HR factor. FIP- (and WAR) I believe get adjusted based off run factors, even though the pitcher is pitching in a park that suppresses HR’s.

4:14
Neil Weinberg: So you are saying the pitcher is getting too much help from the park factor on FIP because the park actually doesn’t allow many HR? I think that”s fair. I think it’s probably a very small overall effect. But making park factors more responsible to this kind of thing is a goal of many going forward

4:14
Comment From Noah
FIP misjudging guys like Phil Hughes and Chris Young just makes fWAR look really poor. Especially after a poll was conducted on FanGraphs asking whether IFFBs should be included, and almost everyone agreed that they should.

4:14
Neil Weinberg: I have no idea why you think it’s misjudging them. And IFFB are included in WAR.

4:14
Comment From Tia
What is league average wRC+? Just discovered how to export from FB (thanks to this chat) and want to compare the Reds wRC+ as a team to league average.

4:15
Neil Weinberg: 100 for position players, (technically 96 if you include pitchers)

4:15
Comment From Noah
Just make IFFB’s equal in value to strikeouts and recalculate the constant. It’s that simple. The good thing is that FanGraphs’s failure to appreciate pop ups has allowed smart teams (like the Rays and the Mets) to feast on extreme fly ball pitchers (think Drew Smyly).

4:15
Neil Weinberg: Same answer. Popups are in WAR. Have been for a while.

4:15
Comment From Ray Rice Double Jeopardy
How do the Tigers and Cardinals know the exact time to reach down and rip the division out of the grasps of the leaders going into the final month?

4:16
Neil Weinberg: I want to make a joke about experience but it’s really just a matter of being good.

4:17
Comment From Dennis
why is wRC+ not called wOBA+? I was under the impression that wRC+ is derived from wOBA.

4:17
Neil Weinberg: It’s a different scale. wRC+ is on a runs created per PA scale while wOBA in on an OBP scale. You will make the same inferences when comparing players, but the scale will be constrained.

4:18
Comment From Alex
Sanchez is criminally underrated. Even with bad command a 65% GB% is great. Sanchez may not be an ace, but this is one case where the scouts are more right than Fangraphs.

4:19
Neil Weinberg: I think it’s too early to say for sure who is right. I think on average, these guys bust more than the scouting community acknowledges. But that doesn’t mean they all bust. I think FG (non-prospect guys) typically base our opinions on groups of players rather than individual looks at a guy. That works on average, but it will obviously miss too.

4:19
Comment From Jacob
A Whitesox beat writer claimed that with a reliever with closer experience(Listed Grilli and Jim Johnson as examples) the Whitesox would have won 8-10 more games this year. Do I feel bad for laughing at the comment and offering up Grant Balfour?

4:19
Neil Weinberg: Nope. Laugh away.

4:20
Comment From Cman
a WAR calculator would go a long way in helping me waste more time at work.

4:21
Neil Weinberg: Yeah, I have a very basic one on my Tigers site for position players. I improve it and add pitching, but there are lots of little things to do to get it right.

4:21
Comment From Mark
How do UZR/DRS handle plays like that Torii/DK fiasco from the other night? As in, who gets penalized for not making that play?

4:22
Neil Weinberg: So the in that case, the play is made by a CF like 40% of the time (guessing) and by a RF 20% of the time (guessing) and drops like 40% of the time. They each get deducted their percentage times the run value difference between and out and hit. Obviously, one could say either player is totally responsible, but it’s just too hard to do that.

4:23
Comment From Noah
Do you ever plan on updating the WAR for pitchers primer to mention that there is an additional step (including IFFB’s) besides just converting FIP into WAR?

4:23
Neil Weinberg: Yes. This is coming relatively soon!

4:23
Comment From Noah
Neil you’re making me look stupid by including my old questions after you’ve now answered my original question.

4:23
Neil Weinberg: I just answered them in order! Sorry, you asked like 1000 FIP/IFFB questions.

4:23
Comment From Noah
Also does FanGraphs plan on adjusting for pitcher defense and stolen base prevention? Johnny Cueto and Mark Buehrle, for example, can continuously beat their peripherals because of this perfectly sustainable skillset.

4:24
Neil Weinberg: So we do include this, it just goes to their position player WAR side of things. I guess that’s confusing, but it exists.

4:24
Comment From Aaron
Why does Yusmeiro Petit’s expected rest-of-season contribution amount to .1 WAR according to ZiPS(R) when his total-season WAR is expected to jump .6 WAR according to ZiPS(U). This seems to apply to other pitchers as well.

4:25
Neil Weinberg: I think this is just a matter of the data sources crossing. It’s been noted, not sure where we are on a fix.

4:25
Comment From John
How many “+ pitches” does a #3 or better starting pitcher need to be adequate? Can you succeed over the course of a season with just 2?

4:25
Neil Weinberg: Plus pitches? Yes. I think one plus pitch is enough to be a 3 if you have two or three average pitches besides that one.

4:25
Comment From Noah
Do you expect the Mets to be good next season? They have been on a nice streak of late, and with Harvey coming back, Wright hopefully injury-free, and some money to spend on a SS, I personally think they will be right in the thick of the playoff race.

4:26
Neil Weinberg: Depends on how they spend. There are some payroll questions there. If they spend their money wisely, I think they can absolutely contend.

4:26
Comment From Dave M
Do you think in the future there will simply be a variety of different WARs, each weighting inputs differently in some kind of defensible way? That way people can stop being butt hurt about WAR not giving them the exact answer they want. Additionally, will you be doing a podcast with Cistulli anytime soon?

4:27
Neil Weinberg: I think WAR is still new and people aren’t sure what to make of the idea just yet. Long term, if people buy the concept, they will be more likely to buy the idea that there are different versions. Getting there.

And no podcast planned, but if you tell Cistulli you want one I would be happy to accept his offer.

4:28
Neil Weinberg: Going to take two minute break. Fire off some more questions!

4:30
Comment From Sean
Who’s the next player to have a 40/40 season and why is speed (or at least, base stealing)+ HR power becoming more rare? Thought Trout would but he seems to be allergic to SB’s nowadays

4:31
Neil Weinberg: Not sure why exactly we see less of these guys other than that offense is just down overall. I’ll say…Springer? Gosh. Someone who’s like 12 right now probably

4:32
Comment From Aaron
With Petit, it looks like ZiPs(U) projection for the end of the season looks exactly like what ZiPs(R) projects him to do, so the data seems to add up together. It seems like it has to be the WAR calculations… weird

4:32
Neil Weinberg: Yeah. Right. I think one is taken directly from ZIPS and one is internally derived.

4:33
Comment From Scott
to clarify an above q&a, position player WAR for pitchers DOES include defense, but not “stolen base prevention” or anything else a pitcher can do between pitches, correct?

4:34
Neil Weinberg: It does include stolen base prevention…or at least, we track rSB. It might not be included in WAR. The data is on the site if you want to included it in your evaluation.

4:35
Comment From Noah
Is there a reason you guys keep the FIP formula so rigid?

4:36
Neil Weinberg: I think the reality is that the weights almost never vary all that much. I guess we could make them 12.94 and 2.94 and stuff. But the variation is tiny

4:36
Comment From DB
What do you think is the best way to compare players from different eras? I’m talking over their careers, not single seasons. Do you put more weight on 3 great years, or good production over 6? How would you adjust for the era differences, and does FG have something like this besides looking at wRC+ or something showing how they did relative to their own competition.

4:37
Neil Weinberg: I like to compare a player’s peak. Usually 6-7 years. B-Ref has WAR-7 in their HOF area. Could be useful!

4:37
Comment From Noah
If the season ended today, who would be your NL ROY (ie who do you think is most deserving)? Obviously Hamilton is favored by WAR, but that is driven largely by his defense, which DRS is not fond of at all.

4:37
Neil Weinberg: deGrom, but I haven’t thought about it all that much yet.

4:37
Comment From DB
If you were trying to make ratings for a video game or sim game (contact, power, etc) that included all-time players how would you go about it basing it off their careers. Best way to adjust for era differences, competition, etc?

4:38
Neil Weinberg: Probably their stats relative to league average during their careers

4:39
Comment From Robert Hombre
A primary pull of FIP is its simplicity. It doesn’t take much to approximately calculate FIP on the back of a napkin if you assume C’s about 3.10. This simplicity is imperfect, possibly, but that’s part of its appeal.

4:40
Neil Weinberg: Yeah, I think I recall Tango being in favor of standard constants. Might be a little wrong, but it’s more digestible that way. I’m okay either way.

4:41
Comment From Dave M
If playoff teams could not be chosen based on win-loss records, how do you think they should be chosen?

4:42
Neil Weinberg: Maybe something like run differential? Would motivate teams not to use mop up players?

4:42
Comment From DB
If you could go back and watch any player or matchup all time what would it and what stadium? Randy Johnson v Babe Ruth in original yankee stadium?

4:43
Neil Weinberg: Koufax v Bonds, Coors.

4:43
Comment From Aaron
Is there a way that WAR takes into account the fact that games are discrete events? For instance, if a pitcher is absolutely terrible only a couple game in a year (but good in the rest) vs. a pitcher who has similar end-of-season stats that is more mediocre than pitcher one on a game to game basis. Am I wrong in thinking this makes a difference when accounting for their value?

4:44
Neil Weinberg: WAR doesn’t do this. I’m not sure you really want it to. I think it’s totally fair to augment your impression of a player if they had one awful outing compared to spreading those runs out, but I think the evidence suggests it’s not a big deal going forward

4:44
Comment From Mark
What’s your take on the most appropriate split between FIP-WAR and RA9-WAR (50/50, something else), and why?

4:45
Neil Weinberg: I think it’s probably about 70/30. But that’s not based on hard data. I think you could make a case for anything between 85/15 and 40/60 without being crazy.

4:45
Comment From hscer
not a big deal going forward but isn’t WAR a look-back stat?

4:46
Neil Weinberg: Correct. I think I just read the question wrong at the end. I don’t think it matters, but that reason was not relevant to the Q

4:47
Comment From Robert Hombre
That said, while one can comment on FIP’s simplicity, xFIP does not face the same strictures. The denominator is IP – outs, in other words – which does rely somewhat on defense/BABIP. For FIP, we’re willing to trade some accuracy for simplicity. But why does xFIP use outs rather than batters faced in the denominator? Heck, it would be pretty simple, really: TBF/(lg avg BF/inning). Just another constant.

4:47
Neil Weinberg: Same reason. So it looks like ERA. You could absolutely create a FIP or xFIP that uses batters faced.

4:48
Comment From Jacob
Say that a star MLB player was suspended 25 games for a amphetamine offense and was in process of serving it, but new evidence came out in regards to same failed test that it was a PED/HGH and fans were leaked info on it and MLB appeared as if they withheld evidence but attempted to give player 50 game suspension…would MLBPA/player be able to argue that it was double jeopardy suspension?

4:48
Neil Weinberg: I assume they would try to cap it at 50 games.

4:49
Comment From Noah
I think dividing runs scored by runs allowed is a better measure of team talent than run differential. To use an extreme, if a team allowed 0 runs in a season but only scored 162, they would be the best team in the league despite maybe not having the best run differential. They would certainly be better than a team that scored 1000 runs and only allowed 800, right? I think that’s the problem with run differential; it underrates good pitching teams.

4:49
Neil Weinberg: The question asked me to reinvent the playoffs, this is a fine idea too. Neither means anything!

4:50
Comment From DB
Hockey advanced stats have something called QoC (Quality of Competition) and factor it in to some of their other stats. Imagine this could fit in somewhere in baseball, especially for pitchers. Would have to decided how you measure it, i.e stats to that day, updated throughout year, etc. Any thoughts?

4:50
Neil Weinberg: Yup, I know this is something we’ve been talking about. I don’t know how far away we are from getting there but we’re certainly aware of the fact that facing Kershaw isn’t the same as facing Correia

4:51
Comment From hscer
Correia’s mom is like, Who is this mean old Neil?

4:51
Comment From Noah
So it’s definitely true that FIP and xFIP underrate fly ball pitchers then, right? That should be mentioned in the new WAR for pitchers primer.

4:52
Neil Weinberg: Fly ball guys who also don’t allow HR, yes.

4:52
Neil Weinberg: Any last minute, quick questions?

4:53
Comment From Guest
who’s better…kershaw or felix

4:54
Neil Weinberg: At pitching? Kershaw. At everything, probably Kershaw.

4:54
Comment From DB
Tigers postseason pitching rotation assuming they can set it up the way they want?

4:54
Neil Weinberg: Scherzer, Price, Porcello, Velander

4:54
Comment From Sean
1st in offense/last in pitching average d

4:54
Comment From Sean
or last in offense, first in pitching, average d

4:55
Neil Weinberg: Would rather watch the good pitching team, think the offensive team would be better.

4:55
Comment From Dave M
If the Red Sox go into next year with basically just one legit left-handed bat (Ortiz), which is how they are rolling now, do you think that will be a major problem?

4:55
Neil Weinberg: Not really. Good hitters are good hitters. It’s a place to improve, but it’s not vital

4:56
Neil Weinberg: Alright, thanks for the questions! We’ll do it again next week. I’m @NeilWeinberg44 on Twitter if you need me before then. Take care.





Neil Weinberg is the Site Educator at FanGraphs and can be found writing enthusiastically about the Detroit Tigers at New English D. Follow and interact with him on Twitter @NeilWeinberg44.

newest oldest most voted
Andy
Guest
Andy

Pennsy, who asked about well-roundedness of players: There was an article here a few weeks ago that addressed that, using five or six different criteria. I can’t remember them all, but I think they included SB, iso, K%, and some others. I can’t find that article now, but it’s around somewhere.

Neil, in response to Lucas, you said, “There is something to this basic idea. If you allow a ball to be hit to X at Y speed, what is the average expected run value of that play?”

I want to take this even further. I’ve been wondering if we couldn’t dispense with the entire wOBA formula, which is based on the traditional view that a player’s value depends on his hitting singles, doubles, etc., and just, in effect, cut to the chase. A batter’s real skill depends on how hard and how squarely he hits the ball, with the outcome of that hitting depending to a large degree on chance. We all know that a player who hits a hard line drive that is caught up against the fence has performed better than one who hits a blooper that falls for extra bases, but he doesn’t get credit for this. We assume that over time these things average out. But why wait for them to average out if we can just measure the probable run value of a ball as it leaves the bat? Its speed and two trajectories (angle wrt the ground and direction on the diamond) should uniquely determine an average expected run value. The speed and first trajectory reflect the batter’s skill, leaving only direction on the diamond as possibly an outcome of to a large extent (but not entirely) random chance.

We would still have to work in walks to get a full measure of offensive value, but as it stands now, they are considered to reflect skill, not luck, so that shouldn’t be much of a problem.

In theory, if we did this right, we could determine wOBA and the rest from these speed/trajectory data that, over a large enough sample size, should agree with the now conventional manner of using various types of hits. I would love to see a pilot study that tested this idea.