Effectively Wild: A FanGraphs Baseball Podcast - Effectively Wild Episode 188: The Yankees and PECOTA/Earned Run Ratio/Least Likely MVP Candidates/When to Trust 2013 Stats

Episode Date: April 24, 2013

Ben and Sam answer listener emails about whether PECOTA underestimates the Yankees, how to improve ERA, Coco Crisp the MVP candidate, and in-season stats vs. projections.

Good morning and welcome to episode 188 of Effectively Wild, the daily podcast from Baseball Prospectus. I am Ben Lindberg, joined as always by Sam Miller, and we are here to answer your email questions today. We got some good email questions that we are here to answer your email questions today. We got some good email questions that we are going to answer. You just said that. Yeah, I did. Okay.

Do you want to start in any particular place? Nope. Start where you want to start. Okay. Okay. Then I'll start, I guess, with the Yankees question. This is from Steve. Steve says, I'm a Red Sox fan, expat, living in Brooklyn.

Right now the Yankees seem to be holding their own, and then some despite an array of injuries that one would expect to be crippling. I can't remember a season when the New York Yankees did not outperform their Pocota expected win total. Where do the Yankees rank in terms of realized versus Pocota expected wins since the inception of Pocota expected wins since the inception of Pocota? If they have been a consistent outperformer, do you think it is luck or something else? If you conclude that Pocota does have an anti-Yankee bias, would it be appropriate to

build some kind of aura and mystique factor into the projection system, or is that idea fundamentally on Pocota. So I checked, because I compiled a list of preseason Pocotas when I was working on something else not too long ago, and it looks like if you throw out the first couple years, because for the first couple years, like 2003 and 2004, when Nate did projected standings, they were kind of crazy. I don't think they had a strength of schedule adjustment. So a team like the Yankees that was in a good division just was projected to win a ton of games because it wasn't accounting for the competition. So if you throw out those first couple of years when they were projected to win like 109 games and 106 games, since then, so from 2005 to 2012, they have been projected to win 756 games and they have actually won 765 games.

So that is nine extra wins, which is, I mean, nine extra wins in eight years. So a little over one win a year. So I don't know where that ranks in terms of how Dakota has over-projected certain teams, but I would guess that it is not as much as Steve would have guessed. is not as much as Steve would have guessed. The last time that they didn't, or that they, let's see, when was the last time that they did not outperform their Pocota expected win total? I guess it would have been, it's been a while.

2005, they were even. 2006, they outperformed by three wins. 2007, one win. I guess 2008 is the year when they missed the playoffs and they were projected to win 97 games and they won 89. So that is when. So as Sam pointed out, I think when I sent him this email from Steve, you would sort of expect a competitive team or at least the Yankees to outperform Dakota slightly because they would have a tendency to beat buyers at the deadline. So maybe that is your extra win or win in a fraction per season right there.

And I guess you theoretically could account for that. And you could, one thing that, that Pocota doesn't do, and that I guess no projection system does really is, is to make some adjustment for tendencies with rosters. And just, I guess, I don't know, something like the Orioles last season, just making a ton of moves. That seems to be something that they just do. But the projection system doesn't account for that

and doesn't, as far as I know, account for a competitive team being a buyer at the deadline. Yeah, the Orioles example you gave is a little bit probably too specific to ever actually expect any system to go for. I mean, at a certain point, you have to just expect the fan, the buyer, to kind of make his own adjustments as he sees fit and to be reasonably well-informed on these sorts of things. But the idea that a team that projects to win, say, 84 to 90-plus games is likely to be adding, and a team that is projected to win 74 or fewer is likely to be subtracting, seems consistent enough.

I mean, the problem is that you project a team to win 84, but you don't actually know what they're going to win. But, I mean, it seems reliable enough. That is a league-wide tendency that has persisted for many, many, many years. You and I, of course, don't have any say in Pocota, but if you were running Pocota, would you give sort of a one-game bonus to all teams projected to win more than 85 and a one-game demerit to all projected below 75 or something like that? Sounds reasonable. I'm trying to think of some way to empirically verify. I guess part of the question, too, is what exactly is the question that Pakoda is trying to answer. If it's trying to answer the question of how many wins the team

is going to win this year, which is sort of what it is on the surface, then yes, you might do that because that might make it more accurate. If it's trying to give you an accurate reading of how talented the team is as currently constructed, which is sort of the philosophical and fundamental question that underlies it, then you wouldn't get too imaginative about what sorts of hypothetical changes to that team might happen because those changes haven't happened. And once they do, then you adjust. Yeah, I don't know. So I don't know what the question that Dakota is trying to answer is, to be honest. I don't know that I've ever sat down and thought about it.

Yeah, I guess most people would, I don't know. I mean, at the end of the year when it is judged and people look back at how it did, they don't sort of add up how it projected the people who were on the opening day roster or on the roster when Pocota's last preseason projection was, they look at how many wins the team won or games they won. So I guess retroactively at least it's judged as if it were trying

to just predict how many games the team will win. I guess that's how most people use it. Yeah, that seems right. So yeah, I don't know. I guess maybe I'll suggest it or I'll bring it up it's a good idea i mean are you gonna are you gonna mention whose idea was uh i did i mean when you bring it up or suggest it oh sure all right great uh all right uh read the coco crisp one uh okay where's the coco crisp one uh i think it's podcast oh yeah here it is okay brad um hello ben and san alphabetical nothing to read into i'm a big fan of the show and i really

appreciate the daily podcast. I listen on my way to work and it makes the seemingly endless snow-filled commutes a little more bearable. My question is, what if Coco Crisp just won the MVP award this season? I mean, what if he just kept playing well, hitting for power, helping his team win, and then he just ended the season with great stats for a great team and he won the MVP? How crazy would that be? stats for a great team and he won the MVP. How crazy would that be? How odd would it be to watch Coco Crisp accept an MVP award? Is Coco Crisp the most unlikely of all MVP candidates? I know that in reality, a guy like Drew B. Terra is the most unlikely of all MVP candidates. I get that. However, if I had you list your choices for AL MVP at the beginning of the season, basically until

you run out of players, you would actually consider Coco Crisp would eventually be on the list, but he would be one of the last guys before you give up and stop making the list. He might even be the last guy. He's probably a better candidate than guys on his team, but still he's Coco Crisp. Yeah, I think that's probably all I have to read of this. It's funny because a couple of years ago, when I was at the register, I did a post about Vegas odds on MVP voting, on who was going to win MVP. And Coco Crisp was actually listed, and he was the guy who I mockingly included as the bottom of the MVP contender list.

I think he was like 125 to 1 or something like that to win the MVP award. And I like this question a lot. Vegas, when they do these sorts of prop bets, there's always one of the options is the field. And like they they list 50 guys and then the field is everybody who's not listed and i remember uh one time a friend of mine uh at work brought back like the the sheet that showed all the prop bets and it was for who was going to lead the majors and home runs and i was this is probably like 2005 and i was trying to imagine who from the field could do it and i believe if i'm not mistaken i believe that somebody from the field could do it. And I believe, if I'm not mistaken, I believe that somebody from the field had won it the previous year. Um, and so the question

is sort of, I guess Brad's question is sort of who from the field could do it is Coco crisp, you know, from the field. I imagine that Coco crisp was not listed this year. Um, but, uh, I, yeah, I mean, I'm not sure that, for instance, last year, I don't know that Chase Hedley would have been seen as any more likely to win. In fact, he almost certainly would have been seen going into the season as I would think less likely to win than Coco Crisp is now or is before this season. Hedley was 28. He was coming off a four-home run season.

He had never driven in more than 64 in a season or hit more than 12 home runs. And yet, if the Padres had sort of fluked into a division, which they could have done if, you know, the pitching had come together and they'd had an Orioles-like run in extra innings, he would have won it, I think. And so it's not that the unlikely is actually not

all that unlikely as a group. I mean, each individual is extremely unlikely, but as a group, it's not all that unlikely. So I was just glancing at MVPs from years past. And you actually have to go back quite a way to find, I would say, a real out of nowhere MVP. And by out of nowhere, I would sort of define it as not a guy who had received MVP votes previously and not a young guy like Trout if he had won it last year. Obviously, he hadn't won MVP votes before, but he was 20. votes before, but he was 20. So you could maybe, maybe make a case that Justin Morneau fits because he had never won MVP votes. He had never been an all-star and he was coming off a season in which he had a 741 OPS, but Morneau was also, you know, an elite prospect just like two years earlier. elite prospect just like two years earlier so it really wasn't that out of nowhere at all so i wouldn't count him so if you don't count him you have to really go all the way back to 96

with ken caminiti who was 33 had never received an mvp vote had only been an all-star one time but also you probably would have considered him more likely than Coco Crispy. He had hit 26 homers, driven in 94 the year before, had a 900 OPS, and was a leader kind of guy. So Caminiti probably doesn't quite reach Coco status. And then you could go to Terry Pendleton in 91, who had received MVP votes one time five years earlier. He finished 18th. He had never been an All-Star. In fact, he wasn't even an All-Star the year that he won the MVP award.

And he's a great example of how a guy like Coco Crisp can be an MVP because he was so narrative-y and so clubhouse guy. His numbers were not nearly those of Barry Bonds that year. And yet he won it, and then he finished second the next year. And he was 30. So he's a good example, I think, of Coco Crisp-like ability. And in 89, Kevin Mitchell won it at the age of 27, had never received a vote, had never been an all-star before,

was coming off a year with... He was a good hitter. He couldn't... I don't even know who... I think Kevin Mitchell winning it would be like Ryan Ludwig winning it, probably, which almost happened once. So not super unlikely, but very unlikely. But yeah, I mean, Crisp is a good example of a guy

who has absolutely no shot. You would easily name, I think, 100 people before you got to Crisp, or you would quit naming people, I guess, before you got percentage, scores 135 runs, has a sort of like Tony Phillips kind of year and steals a ton of bases and just becomes the known leader on a team that cruises. I think it would have been easier for him to do it last year because the A's aren't quite sneaking up on people this year. i still can't imagine it you still well i mean obviously really imagine him playing a full season even just like a pretty good season it's kind of

hard to imagine him playing 150 games because he's never done that i don't think um so there's that and yeah i don't know it i don't know that it seems any more plausible to me today than it would have a few weeks ago oh it doesn't seem any more plausible to me today than it would have a few weeks ago it's just that like i can see a non-zero likelihood of it in there. It's not a good chance, but I can sort of accept it. I mean, you know, somebody hits 350 for 300 plate appearances for no good reason at any given time. And if that were Coco crisp and you know it could happen tori hundred by the way is hitting like 360 over his

last 500 plate appearances right now it's kind of crazy yeah so is there any does anybody jump out of you as a coco crisp alternative though no uh i meant to do a post at some point about what pakoda percentile the typical mvp winner is maybe i still will do that because i kind of wonder because i mean every every mvp winner is better than than he would have been projected to be that season really there's no there's no one who is i i would i guess I'd be surprised if there were an MVP winner who Pakoda projected or would have projected retroactively to be that good. Do you think that that has happened? Or is it, I mean, every MVP is kind of exceeding his projection to some extent, I would think.

Yeah, I think you'd have to go back to the crazy days of Pakoda, where I remember looking recently at a story that ran in 2002, 2003 maybe, and Jason Giambi was projected to be 12.5 wins. So you have to go back to those days. so you have to go back to those days to i mean otherwise it's not that like like even uh like albert pools when he was winning him i mean it he was still outperforming his 50th percentile to some degree just because it's such a conservative kind of approach and uh i would i mean you if you found one it would be one of those sort of terrible years of pakoda where like like ryan howard i could see have having won it with a sub 50th percentile

yeah coco chris 90th percentile this season was 288 351 431 with 12 homers, 53 steals. So even that would get him nowhere near surprise MVP level, so he would have to be much, much better than that. All right, I'm going to name some names, and you tell me whether you think there are better or worse contenders for this award, for this Coco Crisp kind of category. Okay.

Jed Lowry. I guess better just because more power and we've never really seen him stay healthy, but he's had some very good or at least one very good partial season where he, I don't know, I guess I'd say better. Howie Kendrick. Probably better because he's always been someone that people thought would hit for a really high average and he could have some crazy bad up year where he hit 350 or something and won a batting title and the angels were good and maybe that would do it

uh dion vciado um i guess where he is on the aging curve he would be more likely just because it would be so strange if crisp who's what 10 years older than Vecchietto, suddenly became an MVP. So I guess he is at an age at which it would be less surprising if he took a massive step forward suddenly. So I guess I would still say Crisp. Less surprising. Alexi Ramirez. I think that would probably be more surprising. I think that would probably be more surprising.

And I'm trying to find one last one for you. You know Escobar. Great clubhouse guy. Yeah, I guess in the wake of last season and his general reputation, maybe I could see him being a better player than Coco Crisp, but can't see him being MVP material. Okay, wait, one more. Nolan Reimold. I like Nolan Reimold, but I guess I'd say Crisp

would be more likely. Okie doke. Yeah. Great question, Brad. Great, great, great question. You guys aren't bringing it like Brad is bringing it. Sorry. Yeah.

Do you want to do your earned run ratio one? Yeah. Just quickly, I just want to give a nod to this. Steven says, in regards to kind of ERA and sort of basic run stats, considering pitchers do not average nine innings per start, I've been thinking it might make more sense to use an earn-run ratio rather than ERA. If we use a ratio of average innings per start colon two, average earn runs using the formula, let me give you the formula, which is going to be hard to say

and have you understand it, but it's simple enough. Innings pitched divided by games started to earn runs divided by games started. Looking at one ratio would provide a frame of reference for what the pitcher usually does in one game. For example, in 2012, Cliff Lee had an ERA of 3.16, which we intuitively know is good, but his earned run ratio was 7.03 to 2.85, which is to say he went an average of 7.03 innings per game and allowed 2.85 runs per game so his average line basically pitching line would have been 7.03 and 2.85 which gives us a clear indication of his average performance on a game by game basis so um this is interesting because i think that he i think that era's uh i mean when era was kind of invented and became popular, pitchers were pitching the full game.

And so it essentially was a way of saying how many runs do you give up every game. And the way that it's used now is a little bit more complex and a little bit more intuitive. Sorry, unintuitive. And it only makes sense to people because they grew up on it and know it and so they know which numbers are good um but it's not all that intuitive especially uh because it doesn't really tell you what happened even though it does explicitly tell you what happened um the problem is that ratios are hard to say they're hard to write around um and uh this is like you just heard me trying to explain it in a lot of words it's not something that can be sort of stated real quickly neither

can era of course if you actually try to explain it real quickly but uh like i said since we all know it it's pretty easy it's ingrained um but i I think that it's the idea that phrasing it almost like his line looks, seven innings on average, three runs on average, it gives you a really good sense of what the pitcher does. And I don't think enough probably is made of how many innings per start each pitcher goes. made of how many innings per spot per start each pitcher goes um i think that's probably an underreported statistic among kind of mainstream places it seems like if you were doing a broadcast and you were showing the basic statistics of a pitcher even if you didn't want to get into advanced stuff innings per start is very simple very, and very important. More important than showing a lot of the stats that they show.

So I do like the idea. I do think that if it could be cleaned up and made simple, it would be great, and it wouldn't bother me at all if it caught on and ERA was replaced completely. Yeah, it'd be nice if the earned run could be cleaned up and made simple and replaced completely maybe true yes true so if we are gonna get rid of era maybe we should just do away with the whole earned run but yes i like the idea makes sense i like the idea too uh okay this maybe will be a

quick one from daryl uh i'm playing score sheet for the first time and part of the fun is weekly determining batting order for my team. I initially used Baseball Prospectus' projections and the rule of thumb from the book to determine my lineup, which led to my batting Nolan Reimold first against left-handed pitchers, even though the Orioles had him ninth. But now that I have some 2013 data, I'm tempted to base it on that, even though it's certainly a small sample. I'm tempted to base it on that, even though it's certainly a small sample. Do you recommend I continue to use the projected stats and risk putting slumping players at the top of the order and not riding streaks? Or should I use what little 2013 data I have and risk missing out on slumping players regressing positively to the mean? would say that the answer is if you are choosing between the preseason projections and the three weeks or so of 2013 data that we have,

then you should use the preseason projections. I think if it's a binary choice like that, either or, I think that it actually makes sense to use the preseason projections all the way into the end of the year. If I'm not mistaken, I remember seeing Nate say sometime way back when that the projections remain more predictive of a player's true ability, even after a full season than that season does. So yeah, I mean, I've always gone by that as a rule of thumb. Of course, it's about balancing such things. So the good news is that hopefully you will not have to make that binary choice. I think that the plan is for in-season Pocotas to be

rolled out early next week, maybe by Monday or something like that. The last I heard from Colin Wires, the keeper of Pocotas. So you will not have to choose between preseason and 2013. You can get preseason plus 2013. So that will be on the player cards, I believe, at some point next week, updated every day. And, yeah, you see that, I mean, when you start looking at those, there's not a huge swing. Even if you look at whoever the hottest hitter will have been

the first day that we have in season pakodas uh his projection will not be dramatically different from what it was before the season just takes a while so don't give in i guess stick with the the steps that got you here that's's great, Ben. Yeah. All right. Any others you want to get to or are we done? That's all. That's all for me. Okay.

We will be back with two new topics tomorrow and you can start emailing us for next week at any time. Actually, we just received an email as we were talking, and maybe we'll use it next week. So podcast at baseballperspectives.com is the address. We will be back tomorrow.