Community Projections: Erik Bedard
The results are in, and now that we’ve collected the data, we’re going to begin publishing the results of the 2008 USSM/LL Community Projections. You all had your chance to be the prognosticator, and now we’re aggregating the results into a blog reader consensus. In order to get through all the reviews more quickly, I’m going to handle the pitchers here and Jeff’s going to do the hitters over at Lookout Landing.
The first pitcher we’re going to review is the shiny expensive new guy from the great white north; Erik Bedard. It’s no secret that the majority of us weren’t particularly thrilled about his acquisition, thanks to the high cost of giving up Adam Jones, but we’ve made it no secret that we’re happy that Bedard is a Mariner. I’ve called him one of the five best pitchers in the American League, and even that may be an understatement of his abilities. Especially if the readers are to be trusted. Here are the results of the Bedard projection.
Average of 74 entries: 197 IP, 174 H, 29 2B, 4 3B, 17 HR, 59 BB, 210 K, 6 HBP, 47% GB%, 3.17 FIP
Optimsitic Entry (wcf51): 211 IP, 161 H, 17 2B, 2 3B, 11 HR, 41 BB, 233 K, 9 HBP, 62% GB%, 2.38 FIP
Pessimistic Entry (panev): 208 IP, 167 H, 25 2B, 2 3B, 22 HR, 30 BB, 115 K, 8 HBP, 54% GB%, 4.02 FIP
Dave’s Entry: 190 IP, 197 H, 34 2B, 5 3B, 18 HR, 44 BB, 184 K, 4 HBP, 48% GB%, 3.25 FIP
Jeff’s Entry: 188 IP, 171 H, 27 2B, 3 3B, 17 HR, 59 BB, 196 K, 6 HBP, 47% GB%, 3.40 FIP
Really, this is a pretty remarkable group of projections. You guys couldn’t love Erik Bedard any more if he was carrying your baby. The absolute worst projection in there has him pitching at the level of 2007 Fausto Carmona or Justin Verlander. The reigning pessimist in the crowd thinks he’s going to set a career high in innings pitched and be one of the 10-15 best pitchers in the league. And that guy’s out on an island of negativity!
More than 1/3 of the projections (26 of the 74) had Bedard posting a FIP of below 3.00. The list of pitchers who have posted a FIP below 3.00 in the American League in the last four years reads like this: 2005 Johan Santana. That’s it – it’s happened once in the last four years, but 26 of you think Bedard’s going to pitch at that level. Nine of you think he’s going to be even better than Santana was that year. That’s nearly as many as people that projected he’d post a FIP worse than 3.50 (11 did that, by the way). A 3.50 FIP would still have ranked fourth in the American League last year (3rd if you don’t include 2007 Bedard, since he can’t finish behind himself).
The people who are optimistic about Bedard think he’s going to have a hall of fame season and stay healthy. The people who are pessimistic about Bedard think he’s going to have an all-star season and stay healthy. And absolutely no one thinks he’s going to get hurt and miss more than a few starts.
You might all hate the trade, but good gravy, you guys love the guy we got back. The next time anyone decides that readers of this blog revel in negativity, I’m just going to point them to this projection. Negativity isn’t even in the zip code. This is an all out Erik Bedard love fest. If he’s as good as you guys think he will be, he’ll be on the very short list of Cy Young contenders at years end.
Comments
38 Responses to “Community Projections: Erik Bedard”
Leave a Reply
You must be logged in to post a comment.
I think that the move into SafeCo has more to do it than anything.
I see Bedard riding in on a purple pony of glory!
Erik:
Voulez-vous coucher avec moi ce soir?
Erik dit:
Est que vous avez une ferme de poulet, julian?
This faux French just isn’t working. Trust me.
Dave, Just curious how many people have Bedard maintaining his K/9 rate of 2007. 10%? 25%?
It makes sense. He’s moving from Camden Yards to Safeco, ie from a slightly hitter friendly environment to one that’s somewhat neutral toward lefties (while Bedard himself is decidedly unfriendly to them) and murder on righties.
He posted a 3.33 FIP last year, and while it may be unreasonable to expect him to drop a half a run off that, some improvement is within reach.
Does this sort of Erik Bedard help our case of winning the AL West at all?
Or, more accurately, people just still don’t grasp regression to the mean. It’s extremely unlikely that 2007 representes Erik Bedard’s true talent level. However, because we’ve been (incorrectly) conditioned to use prior year performance as the baseline for expected performance, people will do exactly what you just did – look at his 2007, adjust for Safeco, and voila, he’s going to get even better.
Except that’s just totally and utterly wrong. I think you guys will see a trend, as we go through these, that the community just fails to grasp proper regression to the mean when it hurts the team. When it helps the team (bumps for Sexson, Lopez, etc…), they’re all about regression to the mean, but when it returns a less rosy result, it gets thrown out the window.
The only part that stuns me is that the pessimistic projection still has him over 200 innings. One more decent start to the season for the team followed by another horrendous late summer collapse should beat this last hilarious vestige of optimism out of us, right?
Well, we kind of had this argument when the call for projections happened. Some people suggested looking at 3-year averages instead of just the past year, and the response was that we wanted community projections, not some variation on a 3 year average we already knew. Since the past year’s data were what people were given, that’s what they used as the basis for their projection. (They probably would’ve done that anyway, of course, but I think the methodology exacerbated it).
Mind you I guess I shouldn’t talk, since I didn’t participate at all. (I guess I should have — I was saying Bedard would probabbly top out at 180 IP when the trade was being discussed, and I still think that. But I’m always the biggest pessimist in the room.)
The discussion wasn’t that people shouldn’t look at more than one year of data; the discussion was that people should do more than just look at last year’s data, but also be thoughtful about where the player is along their career path, park changes, etc.
Maybe someone can help me out with understanding this. I get the general concept of regression to the mean, but Bedard has only 4 full years in the big leagues and it can’t be expected of him to reach his full potential in year one or even year two. So how do we know what his “mean” really is? Can you really regress to a mean when your sample size includes the first four years of your career and your best season was your fourth? Isn’t it just as likely that he’ll spend the next 6 years pitching like 2007, especially with the Safeco Field adjustments, and THAT will become his mean?
derubino-
“I work for the phone company. I have a lot of experience with semantics, so don’t try to lure me into some maze of circular logic.”
Your question is probably valid–it just makes my head hurt. 🙂
French Soldier: Un cadeau.
Other French soldiers: A what?
French Soldier: A present.
Other French soldiers: Oh. Un cadeau.
Other French soldiers: Oui oui.
French Soldier: Allons y!
Other French soldiers: What?
French Soldier: Let’s go!
Other French soldiers: Oh.
Those that adjusted for Safeco should have then adjusted for the M’s fielding.
I get the general concept of regression to the mean, but Bedard has only 4 full years in the big leagues and it can’t be expected of him to reach his full potential in year one or even year two.
The key is to understand that Bedard isn’t an entity unto himself. He’s part of a larger population of players, of which we have far more than four years of data, and of which we can use to draw conclusions about Bedard’s abilities.
I think this link will be helpful in understanding how to properly use regression to the mean in player forecasts.
When Bedard isn’t punching out batters, would it be a fair assumption that he has ground ball tendencies? The thought of seeing either Raul or Wilkerson chasing down balls hit to the corners pains me.
Since the past year’s data were what people were given, that’s what they used as the basis for their projection.
I guess it’s possible that people decided to just refer to the spreadsheet with last year’s data in it, but honestly, I doubt that it’s more than a very minor factor. This isn’t a USSM/LL specific issue – people in general just don’t grasp this concept very well.
Hell, most of baseball doesn’t grasp this concept very well. The Mariners certainly don’t, which is why they never see these inevitable collapses coming. This problem isn’t unique to the people who filled out the spreadsheet – it’s just part of life in general. You’ll see it the next time you hear John McLaren talking about what he expects out of everyone. You’ll see it the next time you read Geoff Baker claiming the M’s are the division favorites. You’ll see it in the community projections.
It’s just part of the deal. In reality, I think we’ll be able to identify the lack of regression in the projections and, if we want to, regress the cumulative projections to come up with something a little closer to reality.
It’s just part of the deal with human prognostication. Quantative systems like PECOTA are missing elements that humans can pick up on (injuries, unique skillsets, etc…), and humans are missing regression to the mean (among a few other things). If we look at them both, we’ll get a better idea of the whole picture.
You know, if you actually regress Bedard’s numbers, he doesn’t really come out looking that different. Strikeouts drop, sure, but only by a little, and walks go down too, as do homers.
Regression is important, and not considered nearly often enough, but I’m not sure that Bedard’s the guy to be hammering this point home on.
Anyone think that part of the very rosy predictions might be out of a desire to subconsciously make losing AJ look more palatable? I for one know I will be very TO’d if AJ lights it up in Baltimore and Bedard struggles or is injured . . .
I wonder how many people mistook FIP for ERA.
They didn’t put FIP in there – we had it calculate based on the raw data they entered.
What were the low and high IP projections?
I don’t really know how to chart this as a Stat but I do believe in the Contender Theory. I really do believe that guys play better and harder when they know they have a chance to make the Playoffs. You take Bedard off of a Team that has been re-building since, what, 1983? Put him in a situation that isn’t perfect but certainly is considerably better, I think his numbers are going to at least match last year. Also, the O’s shut him down in September. Had he been on a Playoff contender, I’m pretty sure he would have at least given it a go.
I don’t really know how to chart this as a Stat but I do believe in the Contender Theory. I really do believe that guys play better and harder when they know they have a chance to make the Playoffs.
So a good player on a mediocre team says “hey, it’s just a difference of maybe $10 million a year between sucking and being good…why should I care?” These are guys who have built an entire life upon striving to beat everyone else, and history is littered with amazing seasons on crappy teams. Pro athletes are very highly competitive, baseball may reign them in with silly unwritten rules (don’t embarrass a team too much, don’t bunt to break up the no hitter, etc.), but the players themselves are always trying.
The premise seems to be “players on contending/playoff teams play well.” Which seems to be a big, fat “duh.” A few players tanking and you’re no longer a contender. A few surprise performances and suddenly a team is a contender. It’s the whole cause/effect or correlation/causation thing.
They didn’t put FIP in there – we had it calculate based on the raw data they entered.
That’s even worse. It means they didn’t.
I don’t know… “Regression to the mean” vs. “Wisdom of the masses”. Both are strong, but I’m going with the masses on this one. There are a lot of reasons to think this might be a true put-it-all-together year for Bedard. Among others:
– Safeco
– new, presumably better, team
– no arm issues last year, yet he got to let his arm rest for the last few starts due to an injury that no one seems to think had any lasting negative effect
– Even closer to getting a fat free agency contract or signing a fat long-term deal.
Here is what may be missing from this particular Wisdom of the Masses analysis, though. There is a real chance – for the sake of argument, call it 10% – that Bedard will get injured and either miss a huge chunk of the season. He’ll also probably pitch poorly leading up the injury, if it’s an arm or elbow thing. In order to accurately predict Bedard’s season, you need to take what you think he’ll do if healthy and deduct a fairly sizeable percentage for serious injury. I wonder if people did this, or whether most folks’ predictions assume a basically healthy season. In actuality, this is a large part of what “regression to the mean” means in this context. If a pitcher is healthy in one year – as Bedard was last year – and you assume no change in talent or luck from year to year, you still should deduct a hefty percentage from his totals in your projections to deal with serious injury.
Ah, but there’s the fun of it: all of that thought on your part results in one projection. Someone else thinks he will get injured, puts in a low projection, and so on, and so on, and so on, and the end wisdom you get contains who knows what… and works. (or doesn’t work)
That’s what I’ve always found so awesome: you can’t really reverse-engineer what people are using to do their projections.
I think the wisdom of the masses thing requires that you get a poll of diverse people, not a poll of people who all hope the Mariners will kick booty.
On the fangraph site, they give predictions of Bedard’s won/loss totals by four “experts” (including Bill James) and none of them have him winning more than 13 games. James has his record at 13-11. It’s not clear if these projections were made before Bedard was traded though.
Yeah, you do run the risk of irrational exuberance. If crowds were always wise, you wouldn’t have stock market bubbles or riots where people trash their own neighborhoods.
If a pitcher is healthy in one year – as Bedard was last year – and you assume no change in talent or luck from year to year, you still should deduct a hefty percentage from his totals in your projections to deal with serious injury.
Bedard wasn’t healthy last year: he missed the final month of the season because of a strained right oblique, which is why he only had 182 IP. And why I don’t expect him to exceed 180 this year, either. That shouldn’t affect his rate stats much, though — unless he tries to “pitch through it” — just the counting stats.
(Though I do like the “Contender Tautology”: contending teams have players that play well; players that play well turn teams into contenders. Lather, rinse, repeat.)
Yeah, but we know that NO ONE projected 50 innings on the season, yet we also know there is in fact a reasonable chance that Bedard will get hurt and pitch only 50 innnings. And that makes sense – in all likelihood, he won’t miss 3/4 of the season with an injury, so no one should “project” that. However, that possibility should, I think be factored into any projection. It’s the same reason people accuse PECOTA and similar systems of being overly conservative. Most projection systems include the non-zero possibility of collapse. For example, Bedard has a 10% “attrition rate” according to PECOTA, which means that by that system there is a 10% chance that his playing time will decrease by 50% relative to his previously established baseline.
Anyway, maybe most people did knock 5-10% off of Bedard’s innings pitched to account for serious injury. But, if they did, they sure assumed a healthy Bedard would pitch a ton of innings.
[this isn’t a “board”, and the pros and cons of the Bedard trade have been rehashed enough]
If we go by this: http://www.hardballtimes.com/main/article/batted-balls-and-park-effects/
, Erik Bedard is going to put up huge strikeout totals this year.
Dave has Bedard with 2.09BB per 9innings. Is my math right? That would be a career best for Bedard.
On the other side of the coin, looks like Baltimore is enjoying their end of the deal aswell…
http://sports.espn.go.com/mlb/spring2008/columns/story?columnist=crasnick_jerry&id=3302059
I’d just like to point out that the community projections are pretty much in line with other projections out there if you ignore the IP.
I just completed an analysis on my blog (start at post 59 for summaries) that shows that Community Forecasts are just as reliable as Forecasting Systems in terms of rate stats.
Furthermore, Community Forecasts are likely better than the rest in terms of depth charts / playing time.
If I had to go with only one, I’d go with the Community.