Over the last three years, the Orioles have consistently walked away with more victories than models would predict (whether based on forecasts or observed game action), but Dave Cameron of Fangraphs argues that random variation is still the most likely explanation. You’ll need to read the full piece, but in essence, Cameron says that the O’s outperforming streak is probably not attributable to some skill or special insight, but is rather an outlier that falls within the expectations of the models that predict win-loss record.
More from around the game:
- Cubs GM Jed Hoyer indicated that the team is focused on building out its big league staff in the near term, as Mark Gonzales of the Chicago Tribune reports on Twitter. “We know we have to have balance,” said Hoyer. “That’s going to be our main area of focus.” With several of Chicago’s touted young position players beginning to make an impact at the MLB level, many have suggested that the organization could become a big player on next year’s free agent market — especially to fill out a rotation that is now without Jeff Samardzija.
- Designated hitter Billy Butler reiterated recently that he is still hopeful of remaining with the Royals, Jeffrey Flanagan of FOX Sports Kansas City reports. In spite of a recent hot streak, his $12.5MM club option for 2015 seems a bit steep. “After the season, we’ll see what happens,” said Butler. “We’ll know five days after the World Series what will happen. But even if they decline, it doesn’t mean they won’t offer me something else. I hope that’s the case.”
- Phillies manager Ryne Sandberg’s comments about Darin Ruf’s playing time reveal a continued flaw in the organization’s decisionmaking, argues David Murphy of the Philadelphia Daily News. “The situations he’s been in the last couple of years here, not being able to have a string of at-bats, it’s hard to really get a gauge still,” Sandberg said of Ruf. But while consistent playing time would appear to offer a means of evaluating the outfielder/first baseman, Sandberg said “that’s the tricky part about making lineups and also trying to win a game.” As Murphy opines, this line of thinking suggests that the organization is still focused primarily on winning meaningless games this year, rather than setting up the organization for future success.
Just make it a split. A platoon split. Straight. Pinch hit Ruf against lefties. Start him against lefties. Drop Howard in the order
But Howard is hitting lefties better then righties this year
Regression to the means. He’s not hitting both particularly well anyway. Like saying this apple is less rotten than that one, both are still rotten.
He’s hitting LHP to the tune of an .800 OPS (with a massive .252 ISO) in over 100 PAs. That’s not great by Ryan Howard career standards, but it’s much better than “not particularly well.”
Spray charts also show him going to the left side more this year than the last two seasons combined, and he’s got more homers to CF, LCF, and LF this year than both 2012 and 2013 combined as well.
I hope the royals decline the 7 million option on Wade Davis
The Orioles get no respect. In midst of all of these deep statistics we forgot human will. Not everything can be explained by math. The Orioles have a great manager who manages above the statistics. Human will, determination, and focus could potentially explain this “outlier”. These aren’t computers playing, they’re real people.
So the Orioles are just more determined to win than the other twenty-nine teams? While I’m not completely sold on BaseRuns as a predictive engine, that is a much less compelling argument than Cameron’s.
I think the Orioles have figured out how to win in the current game. The game changes, so it may not last, but they have found something that other teams have not. That is why they are the outlier. Statisticians should be required to have played the game to understand the human element and feel for the game. Buck knows what he’s doing right now.
I played baseball for years, and I have been in my share of games where one team started to tilt and lose focus, causing them to play poorly. So I understand what you mean. But I think part of what makes these guys professionals is their ability to stay focused, so unless you could show me some concrete proof of the Orioles specifically targeting players that are somehow quantifiably more focused, I can’t just give that edge to the Orioles over other teams. Also, as Cameron pointed out in his piece, the Showalter argument is weak because his Texas Rangers teams actually under-performed their BaseRuns records.
I encourage everyone to read the article.
If Buck has it all figured out what happened to last years team? The one that underperformed in “clutch” situations. Or the Angels team that beat expectations for 8 straight years, but has fallen short the last three?
I encourage you to consider things other than numbers.
I read the article, I understand. I agree with the statistical model. Its not about finding a way to predict it or understand why the stats don’t back up the win total. What I’m saying is that there is no denying that they are in fact winning. Maybe this team’s successful model, unique to these players, this year, and only in the current environment (including every other player and team and coach in the game), is based on what they think will work. Consider things like match-ups, lineups, hot and cold guys, nagging injuries, etc. Buck has a feel for what will work because he has developed an image of what each of his guys brings each day. He feels things with his gut. I know that sounds crazy because there is no “gut stat”, but I can tell you there are a lot of baseball men who go with their feelings, and it works. Don’t ask me where it comes from or why its only this year. But if you want to actually explain the outliers (including the Angels), you would have to get inside the brain of the manager.
Orioles get no respect. 4th best team ERA since the ASB, one of the best bullpens in baseball, leading the MLB in home runs, and a #1 defense, and they’re just an outlier? Please, this super sabremetrics of “regression to the mean” and “outperformance” is nonsense. They’re a great team.
You’re not talking about the same thing they’re talking about. They’re talking about pre-season predictions.
They are a good team that is sequencing like a great team.
They are (in the AL using FG out of 15 teams):
wOBA – 4th
wRC+ – 3rd
Baserunning – 10th
Defense – 2nd
Overall positional player WAR – 3rd
Starting pitching
ERA – 8th
FIP – 14th
xFIP – 14th
Starting pitching WAR – 15th
Basically you have a team that is solid with positional players (right behind the Angles/A’s) with mediocre to poor starting pitching. Their bullpen ranks right in the middle in terms of WAR.
Sandberg is in a no win spot. Can you really just sit Howard and his giant contract? It’s pretty hard to do, especially if the FO knows it can’t trade him at this level of production. Howard still had an ops+ of 115 and a positive WAR last year, in about half a season. That doesn’t mean he’s good. But it does lead one to fear that if you just cut him loose he might start hitting again.
The problem isn’t with playing/sitting Howard. The problem is that Sandberg has insisted on playing Grady Sizemore – a player who has 0 future with this team. And in doing so, sacrifices playing times for Brown/Revere/Ruf. And while none of those trio will probably ever be stars, they can all be contributors for the next 3-4 years (and cheap ones at that).
I have no problem with playing Howard – because you are stuck with him. You can’t just cut him and swallow $60 M over the next 2 seasons. That’s not a feasible decision. Like you said, he’s nowhere near the player he was before, but he can still be a useful piece.
My opinion as an outsider; don’t worry about what Howard would do if traded, he will never play to that contract. Forget about brown (can’t field or hit). Revere is playing like an all-star. Your right about sizemore, no matter what he does he’s not the future, play Ruf everyday.
I’m not worrying about what Howard would do if traded – just the fact that we are locked into $60 M for him over the next 2 years (unless they find a way to trade him and not take on all the salary). What’s the point in cutting bait with him if the only reason to do so is to NOT have him on the team? I don’t think Ruf upgrades the position much (if at all) from the current level of production Howard provides.
As far as Brown goes, it’s funny that everyone says he’s done. He was an All-Star just last year. I’m not saying he’s going to be a perpetual one – just that I’d rather they not give up on a 26 year old kid who was a top 5 prospect 3 years ago.
I think Sizemore is being played for the same reason Boston took a flyer on him. When he’s fluid, he looks like he can still be a plus player, and he works hard, which has credibility with Sandberg. The problem is that the Phillies have a lot bigger fish to fry than whether Howard is playing full time or part time.
You can cut Howard. Angels did it with Wells. I haven’t heard an Angels fan yet wish that they held onto Wells. Free up the roster space and move on. Now, if there is no one better to take his place that’s another story.
The Angels didn’t cut Wells, they traded him for relatively nothing (and got the Yankees to take on some of his salary – actually, Yankees paid over $10 M of his remaining $40 M). The Yankees then cut him this year.
Like I said in a later post, I don’t think Ruf (the current option for an “upgrade”) is even an upgrade.
The reactions on the Ruf comment are a bit dramatic IMO. I get what Sandberg is trying to say. Over analyzing has become the norm.
In other words: the Orioles are just lucky. Ugh…
Apparently you didn’t read the article. That is not the conclusion that Cameron reaches.
I have a background in higher math and read the Cameron piece with interest. I even re-posted it. Cameron does not analyze, purposefully or otherwise, the factors that may predict the Orioles now-consistent performance as a statistical outlier.
Between roster makeup (Dan Duquette) and game strategy (Buck Showalter), quite possibly there is a factor that predicts the Orioles outlier performance now in two of the last three years.
SAT scores are also normally distributed, but there are factors that may predict that certain students consistently do well on the tests (innate intelligence, quality school district, good test-taking training, etc).
Observing that a student gets 99s does not go to the “why” and there can be a “why” more compelling than dumb (good) luck. Cameron should have done more.
100% agree. After 3 years its no longer an outlier its a trend so the model not not measuring all necessary variables to make a accurate prediction.
They are still an outlier among their competition. They have outperformed where other teams have not found a way to outperform. I’m not even questioning their predictive models. I believe the Orioles have positioned themselves as the outlier using a bettter strategy, they win more games, and the mathematicians can’t accept that it is not a random event. There is nothing random about finding a winning formula.
Haven’t we been hearing the “Orioles are lucky!” argument for 3 years now? I could believe that more in 2012 when the point differential wasn’t very good and they had that incredible run in extra inning and 1 run games. But last year and this year are a little different than that.
That being said, some people take the models as the end all be all of baseball prediction. Which it isn’t meant to be in the first place. It seems the O’s are more on the good side than the lucky side this year.
It’s not 3 years. It’s 1.5 years out of 2.5 years. They underperformed last year. And if it’s all Buck then why wasn’t his success repeated before coming to Baltimore. And if it was Duquette why was the team so poor before 2012?
Cameron acknowledged that there may be something to this other than sequencing. However, it cannot be explained and is most likely just noise.
I think Cameron’s point was that, while an interesting case, the Orioles beating their BaseRun predictions in two out of three years is very likely to just be a result of variance, and that they have not consistently beaten the projections on a yearly basis enough to suggest otherwise. There just isn’t enough evidence yet to say that the model is bad when it comes to the Orioles.
Despite what the nerds in the corner will lead you to believe, baseball isn’t just a giant equation. There is so much more that goes into winning games than the numbers side of things.
I don’t agree with most of the advanced stats unless is a very large sample size but even then I still like to go with the straight up raw numbers unless the the stat adjusts for ballpark or league. Like some people rely on the Pythagorean record for win-loss (sometimes it works) but its not 90-100% accurate.
I thought Cameron’s article was fair. The AL East is a nightmare this year. Someone’s got to win it. The O’s have been impressive.
The AL east has the best winning percentage in baseball.
True but the talent in AL West is concentrated with Oakland, Seattle and the Angels. Each team gets to play far lesser teams ( Rangers and Astros) at the bottom 38 times.
I’d say the AL East isn’t top heavy (although the Orioles do appear to be beating up on a lot of good teams lately). But not being top heavy doesn’t mean the division as a whole isn’t good.
It’s an off year for the Sox and the Rays while the Yankees continue to puzzle teams by remaining in the picture at all.
That’s not to take away from what the O’s have been doing, but both them and the Jays still have big question marks surrounding them. The East is normally better than this.
It usually has more elite teams than this. But I don’t think it’s a “nightmare.” It still has the highest winning % in the AL. Just because you don’t have the frontrunner for the WS, doesn’t mean you aren’t, from top to bottom, the best division.
I like Dave Cameron’s work and appreciate Fangraphs. The one flaw in their projections is not accounting for the leadership’s effect on the team. In the O’s case, Buck Showalter is good leader in decisions on the field, instilling good habits/training new players and to inspire better performance from his players. It would be interesting to look back and see if Showalter’s teams have routinely outperformed projections. It’s one of the elements hard to quantify, but can be important in a team’s overall performance. I’d love to see a rating system for management calculating how much a group may add or subtract from a team’s wins.
From the Cameron article:
“From 2003 to 2006, [Showalter] managed the Texas Rangers; in three of those four years, the Rangers lost more games than expected, and his overall average winning percentage in Texas was 16 points lower than the BaseRuns model…Of the nine seasons managed by Buck Showalter in the years for which I have BaseRuns data, his average bump in winning percentage amounts to 1 point of winning percentage per year.”
It’s funny seeing people getting all riled up about Cameron’s article. Mostly it’s people who haven’t read it or don’t understand the article. It’s the same folks that still believe that wins are an important stat or that saves mean something. Unfortunately, those folks are pretty hard to reach.
There was a time in my life that I felt like the Angels had figured something out. Shrewd baserunning and bullpen management led the Angels to outperform the models for 8 straight years. That’s a heck of a track record vs what the Orioles have done. Yet, as Cameron points out, they have underperformed the models for the past 3 years now. While I *think* that there might be something that the Angels were doing that was giving them an edge, that edge is probably pretty minor. The reality is that they simply had things break their way pretty consistently for a fairly long period of time. That doesn’t take anything away from what the club accomplished because at the end of the day wins count.
“Meaningless” is broad. You can’t totally discount the average 30,289 people watching the game, focusing on Howard’s name brand. But they should dump Howard and play Ruf. He was projected at the beginning of the season to have one of the higher OBP or SLG (can’t remember) in the league. He could be good.
Couldn’t disagree more.
Sabermetrics are by no means worthless. They can uncover interesting trends. The A’s and Brewers are using statistics to exploit opponents with platoons and all sorts of stuff. Every manager …well most… use some basic form of these statistics to try to gain an edge. The deep stuff, like I said, can uncover some interesting trends, but they cannot truly predict anything. They just don’t take into account that people play the game, not computers.
They’re far from completely worthless, but they’re not perfect either.
Which areas would you like to see them investigate or change because you don’t feel they are accurate measures or predictors?
Ok, maybe completely worthless wasn’t the right terminology. I just feel that everything I see or read about baseball now has to do with sabermetrics, and that no one is actually watching the games or the players. They look at the stats and make predictions and write articles based off numbers. There a people that have a knack for finding great baseball talent, and they don’t need all these statistics to choose the players for them. Sabermetrics don’t take any of that into account, which is the reason I feel they shouldn’t be taken as seriously as they are.
And what are the odds of losing five games by 10 runs and then winning 20 games by three runs in a 25 game stretch?
No, the models are not perfect. What they do is attempt to model reality, not fiction intentionally plucked out of air to break them. Over the course of a 162 game season a team will have some really bad games, and some really good games, but most of their games will be near some level that represents their actual level of talent. That’s what a model like run differential is trying to measure.
Is it 100% accurate? Of course not. Is there room for improvement? Of course. Does that mean “look at how my completely made up example representing a sample of only 15% of a season makes this whole idea look bad!” is an insightful critique?
I actually agree with you that we have shifted too much into the realm of sabermetrics, especially from a lot of people who don’t have a good understanding of what they even are or measure but want to be on the bandwagon. They’re treated a bit too much like gospel rather than (moderately) complex mathematical models for my taste. Likewise I don’t think you’ll have many people disagreeing that there aren’t scouts out there with an eye for talent who don’t need the numbers.
But I still believe they are still the better way to evaluate things. Human beings are just too prone to their own biases, and to being influenced by isolated events. Confirmation bias (where we give more weight to evidence that agrees with what we’ve already concluded) is a very real thing that has been studied and measured. I much prefer methods that are completely unbiased, even if they sometimes get them wrong. That’s why it’s important to find the cases where the models are failing and identify if the model itself needs to be adjusted.
Andrew – I’m not sure that you are understanding sabermetrics.
First, on things like defense and base running there are people watching every play. They are carefully calculating distances, running stop watches, etc. They will watch and rewatch every single interaction far more than a casual fan. They then translate their findings to numeric values.
Second, nobody is suggesting that scouting or watching isn’t necessary. Even the most ardent defenders of advanced metrics will tell you that visually watching is an important component to evaluation. The question becomes, which is more important – how much weight do we give to different things? What has the most predictive value? Look around to the best run organizations and you will find teams that have great scouting departments AND great analysts.
The BaseRuns model, the one that says the Orioles should be closer to .500 than where they are now, is built on more factors than just run differential, basically for the reasons you described.
I actually stopped to read the FanGraphs article at this point, since I hadn’t read it yet.
I think the basic problem is that you’re expecting something different than the model is trying to produce. From the article, Dave says “[c]learly, we have to acknowledge that it is possible to consistently win more games than the model suggests, at least over a five year stretch. It’s happened, and not all that long ago.”
At this point, he goes on to examine the model: Is the model getting something wrong? Do we need to adjust the model? (His conclusion, incidentally, is no: The data still fits a normal distribution, which is what is expected.)
By contrast, you go looking for why the team over- or under-performed and identify things like a young pitching staff and struggles against only a couple of teams, or a strong bullpen.
Neither of these are wrong, but they’re clearly different approaches looking at completely different things. Maybe that’s a source of your hesitations.
I get that and don’t argue… more often than not the model will work… more often than not a pitcher will return to his FIP after having a better year or subset of a year.. more often than not a player with a high BABIP will come back to earth. My problem is that every time the O’s start to have a little success we start throwing out words like “Luck” and “Fluke”… completely not the case. Sometimes we need to just let the model be right 90% of the time and say this is one that the model got wrong… the O’s are a 69-50 team this year and right now they might even be playing better than that.
Frustrating to see the home town team having great success but not being able to find many national writers willing to paint it in a positive light.
But isn’t that also a RA problem? If you think your team has so many great players that are worth so much on the trade market, you must also be thinking that you are one hot streak away from being in contention. It’s hard to just strike the set and say “ok, time to do a teardown.” Unless you are a team like Houston, I suppose
GOOREOS – It is luck or a fluke unless you can demonstrate that the Orioles are doing something systematically that invalidates the model. The first thing you would look at is whether it is repeatable. Cameron does this. He points out that they had similar success in 2012 but underperformed in 2013.
Statistics is an attempt to strip away the noise and get down to what is really happening. The model predicts that the Orioles will own a winning % lower than 580 when the season concludes. Those that ignore the model and go with gut instincts think that they are as good or better than their current mark. We’ll see. The season’s only got another month and a half. Me personally, I’ll take the under all day long.
Problem is we aren’t talking about invalidating the model… we’re simply talking about people who have their head so far into a stat book that they can’t appreciate truly good baseball. The model is not invalid or incorrect… it’s simply not 100% and there are teams out there in the past (or players for stats like FIP, BABIP etc..) that out perform these statistics simply because their true value isn’t as easily measured as some people wish.
I’ve mentioned what the Orioles have done that makes them better than what the model predicts but I’ll summarize as well for you.
Great coaching
Improved performance of young pitching
Best defense in baseball
Great bullpen
Great depth
It’s been mentioned earlier in this article but the O’s have the 4th best ERA in baseball since June 1. Chris Tillman gave up 20 runs over 14 1/3 innings in the end of may (I still contend he was fighting through an injury and should have sat down for the good of the team but I’ll never bash a guy for gutting through it) but has pitched to a 2.76 ERA in the 143 1/3 innings outside of that brief period of stinkage. Things like that inflate the statistics and make the team seem on paper like they are worse than they actually are… Tillmans inflated numbers make you think he’s an average AL pitcher but in reality you’re more likely to get 7 innings and 1 or 2 runs from him than anything else. The last time he gave up more than 3 earned runs was 14 starts ago on June 5th, in those starts he’s pitched into the 7th inning 9 of 13 starts yet his stats tell you he struggles to get past the 5th. That is the improved pitching I’m talking about…
Another example is Miguel Gonzalez, FIP is constantly used to say Miguel will eventually become a below average pitcher yet 68 appearances and 62 starts later with 380 innings under his belt he still out pitches his FIP by more than a full run (3.64 career ERA to 4.68 Career FIP). That’s not so say that FIP is “Wrong” or “invalid” as some people try to argue but simply that it’s not 100% and the way Miguel pitches (Not a lot of K’s, a few walks, pitch to contact) doesn’t represent well in FIP calculations. You can always argue that without the O’s defense behind him he would struggle but we’re talking about how good this team is overall right? And defense factors into that?
I appreciate your knowledge of the stat book and how the stats show that this team will under perform but I watch this team every day and being completely objective I can tell you this team really is THAT good. They don’t have a weakness, they have strong starting pitching and a top 5 bullpen and with the power they have in the lineup 1-9 they are one swing away from the lead in nearly any game.
As for 2013 just to address that for a moment I always find the argument funny that the 2012 team was “lucky” but the 2013 team wasn’t “Unlucky” Jim Johnson blew 9 saves in 2012 costing the team significantly in the race for the wild card (Boston won nearly 100 games in the east). If we can play shoulda woulda coulda for a moment and Johnson (or whatever closer should have replaced him when he struggled) blows around a league average of 4 saves on the season the 2013 O’s end up 90-72 on the year right in the thick of the wild card hunt with a chance to win it as they wouldn’t have gone into late September development mode playing guys like Chris Snyder, Jason Pridie, and Dan Johnson the final weeks of the season.
I realize this is just shoulda woulda coulda and not reality but the point is if you want to argue “luck” in 2012 you have to acknowledge them being “unlucky” in 2013. Or perhaps unlucky in the 2012 playoffs when Jim Johnson blows a save to a pinch hit game tying home run to Raul Ibanez then Brian Matusz gives up a walk off game winning home run in the same game to….. Raul Ibanez. I actually don’t call it fluke or luck but attribute a lot more of the credit to great coaching and management by Showalter as 2013’s failures can perhaps be blamed on Bucks’ inability to give up on his struggling closer. Coaching decisions that can swing the Won/Lost record by 5-7 games by the end of the season.
Just realized my earlier comments were for some reason deleted…….. they explained a lot of this and broke down what I feel demonstrates what the Orioles did that threw off the statistical models. Not sure why it was deleted but it would make my comments make a lot more sense.
Likely they were considered disrespectful or too long.