Your Opinion of Royals Magic
The other day, I received a tweet saying 72 wins is the new #6org. It was in response to a Jon Heyman article that ripped a Royals preseason 72-win projection. This is just about the best possible time to do that — the only better time, maybe, to write that article could be next week. Now, a difference: the #6org thing is ours. The 72-win Royals projection came from Baseball Prospectus’ PECOTA system. So, this could be their own version of #6org, but it’s not like PECOTA was all that exceptional. All the systems pretty much agreed. Our own systems projected the Royals to go 79-83. That looks better than PECOTA, but it still looks bad. The Royals have made the numbers look silly, is the point.
They were projected to go maybe .500, and they’re close to winning the World Series. A year ago, they were projected to go something like .500, and they came close to winning the World Series. The year before that, they were projected to go something like .500, and they won 86 games, up 14 from the year previous. Right now the Royals are being described every 10 seconds as relentless. They look like they’re unstoppable, and though they’ve come very close to being stopped, and though they were stopped just last October, there’s still something of a special feeling. The Royals, we’re told, do things differently. On observation, it’s hard to disagree.
What I want to know is how the community feels about what I’ll call Royals magic. Some of you don’t like the word “magic” and that’s perfectly fine. It’s not like it’s actual magic. Royals magic is just intended to describe whatever it is the Royals do that the usual numbers have had trouble taking into account. You get into the alleged relentlessness. Related to that is the clutch hitting. And, of course, there’s the run-prevention side, with a strong team defense that helps the rotation get the ball to an outstanding set of relievers. What makes the Royals so interesting is that we couldn’t easily see them coming. We all want to know why, so we can improve in the future. The Royals might be able to teach us a few things.
To review some information: though, the last three years, the Royals have been projected as mostly a .500 team, they’re fourth in baseball in actual wins. They took one World Series to the very end, and they’re threatening to complete this one before half the country knows it started. This is a golden era for Royals baseball. I know exactly how ridiculous that would’ve sounded some years back, but there’s no denying the circumstances.
Over the three-year span, the Royals rank first in baseball in Defensive Runs Saved. Conveniently and unsurprisingly, they rank first in baseball in Ultimate Zone Rating. You know about the defense. We all know about the defense. It was right in the spotlight last October, when the focus was more on what the Royals’ position players did with their gloves. This time around, it’s mostly about the bats.
And, the position players — over three years, they’re 17th in baseball in OPS, and 15th in runs scored. Yet they’re first in batting Clutch, and they’re first by a mile. The 2015 sense of timing wasn’t an entirely new thing. The hitters have stepped up more than you’d think.
On the pitching side, the Royals are 11th in OPS allowed, yet fifth in runs allowed, and fifth in Clutch. By looking at OPS, defense is already folded in — this isn’t like looking at team FIP. There have been fewer runs than you’d think, and some of that’s timing, and some of that’s getting the ball to the right relievers at the right times. It’s amazing how many different strengths have been celebrated over just the past two postseasons. At the moment, it’s all about the Royals hitters making contact. But it used to be about the position players catching everything, and the bullpen being impossible. (The defense is still good.) (The bullpen is still almost impossible.)
This post is a little scattered, but I wanted to get you the information in advance of a poll. What makes the Royals odd is how they force us to question certain reasonable assumptions. On one side, the Royals have clearly over-achieved, relative to team projections. People put some of that on the clutch hitting. They put some of that on the terrific defense. They put some of that on the shutdown bullpen. So much of the Royals’ identity is basically clutchness. Their hitters look the best at big times, and their best pitchers get the most critical innings.
On the other side: baseball history that’s been examined. No one’s ever demonstrated sustainable offensive clutchness, and no one’s ever shown a link between clutch hitting and not striking out. Based on precedent, or the lack thereof, this doesn’t seem like it should be a thing. And then, you know, the Royals didn’t invent the good and deep bullpen. It seems like we have a good understanding of the defense. We have one team showing how it might be something special, and then we have all this information showing how teams aren’t special. Not dependably so. Weird teams, given enough time, tend to become normal teams, and so we’re led to the argument that never shuts up — small sample size. Ultimately, it’s just a few years of one baseball team. Why overreact to that? Why make so much of an N of 2 or 3?
The Royals don’t make anything easy for anyone. And so we’re finally at the poll. I want you to imagine a hypothetical Royals projection. The Royals have all the same players, and in this hypothetical, they’re projected to win 84 games over a full year. Where would you then mentally put them, knowing just the projection, and having observed all the supposed unkillability? How much do you think the numbers are missing? Or are you a skeptic, happy to put them right at 84, as you would any other team? In short — are the Royals special, or is it just that the Royals have been special? The rest of this is up to you.
Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.
I voted in the poll, but the problem I feel is that knowing that there is a hypothetical 84 win projection on the Royals without knowing any other context is not really information to me.
The information is everything you currently know, and the fact that they have the same players, as Jeff mentioned. Not sure what else you really need
I think the hypothetical projections of the other teams would give you some more information about how the projection would be calculated.
Most projections also offer the nuts and bolts of what goes into them as well.
Projections don’t like the Royals because math doesn’t like the Royals.
The probability of stringing three singles together in succession is less than the probability of one slugger in the middle of the Blue Jays or Astros line-up hitting a HR and creating a run by themselves when you look at only balls put in play. It doesn’t really help the Royals much either when you factor in walks since they don’t walk a lot.
But, the playoffs are a small sample of games and chaos reigns.
Chaos reigned for 6 full months when they won 95 games to lead the American League. Nice hypothesis.
Make it to the World Series? Fluke. Division champs? Fluke. Best American League record? Fluke. Return to the World Series? Fluke.
Or maybe the model is flawed and folks should cop to it.
One (or even 3) outlier(s) doesn’t mean the model is broken.
Maybe people should learn how stats work before they demand personal apologies.
He said flawed, not broken. The projection models absolutely are flawed, but I think everyone that doesn’t have their head in the sand is starting to realize they are much more flawed than we previously realized.
I think everyone probably agrees the Royals are a good team. But for any team to make it to the world series takes luck.
Even if you think cluster luck isn’t really luck, you can look at run differentials and see that the Royals have benefited from a tad bit of an unrelated type of luck.
Blue Jays Run Differential: 221
Astros Run Differential: 111
Royals Run Differential: 83
Uh, to take this to the extreme, if I go 161 games and win them all 1-0, and lose the 162nd 162-0, what exactly does my -1 “run differential” tell me?
Make it to the World Series? Fluke. Division champs? Fluke. Best American League record?
The Red Sox did this in 2013, so I don’t know why you’re pretending it’s impossible or even unlikely it could be flukey.
Er, why did you omit the second World Series appearance from the quote?
Ohhh, ok. So once *can* be a fluke. But twice – twice no way.
I bet you’ve put a lot of careful thought into that conclusion.
Outliers are often a source of additional information and not something to just be ignored. Your attitude is what happens when someone thinks what they learned in Stat 101 is cutting edge.
“Uh, to take this to the extreme, if I go 161 games and win them all 1-0, and lose the 162nd 162-0, what exactly does my -1 “run differential” tell me?”
The projection take stats and turn them into runs scored and runs against. What people mean by cluster luck is that they scored more runs and/or gave up more runs then their raw stats suggest. A basic example of this would be if a team gets 4 singles in a game, they would score more runs if they hit all of those singles in 1 inning rather than if they hit 1 single in 4 different innings.
This is the type of luck that the OP was talking about. If you team outperforms their run differential that is a different way of getting lucky. Lets say in a 5 games series The games go 2-3, 3-4, 7-1, 5-0, 1-2. Team B wins the series, 3 games to 2, even though Team A outscored them 18-10. You can believe what you want, and you may be right, but a lot of people will assume luck had to do with Team A going 3-2. Let’s say you know the # of runs each team scored, but you did not know the sequence team B scored there runs. For simplicity sake, if you assume ties are an outcome equal to 0.5 wins. Team B would only be expected to win 1.3 games.
Oh good, more self-important mouthshitting from Blue.
Outliers are often a source of additional information and not something to just be ignored.
No one – anywhere – has suggested that we *ignore* anything. You made that up because you think it makes your position seem reasonable by contrast.
That makes your reference to stats 101 more hilarious, since that kind of broken rhetoric is usually educated out of people around sophomore year of highschool.
Dismissing it as a “fluke” is pretty much the same as ignoring it.
A relentlessly stupid post given the Royals success over the course of over 300 games.
The thing about run differential is that it’s a counting stat. Keeping runs off the board is just as important as scoring them. Winning a ton of games 1-0 or 2-0 or 3-1 shows up differently than for teams that score a ton of runs. I’m not a statistical wizard but it seems there’s some sort of linear weight that needs to be included in the run differential.
You could in theory be a 90 win team with very low positive run differential team if you are very good at keeping the other team from scoring. Especially b/c due to the capricious nature of baseball means you’re going to lose a lot of games regardless of true talent and some of those will be blow outs.
I generally agree. I have never really understood how stat geeks (i use the term affectionately) discount all counting stats but then love run differential which is essentially just a counting stat. It doesnt account for WHEN those runs were scored.
If I win 81 games 10-0 and lose 81 games 1-0 then I have a run differential of +729. By run differential I should be the greatest team in history. But in reality I’m a .500 team because most of the 820 runs I scored were meaningless add on runs. Nowhere does run differential account for that.
This is why you’ll see many people on this site put down run differential in favor of baseruns or similar stats.
Correct me if I’m wrong (and I very well could be) but isnt baseRuns just a Pythagorean expectation that uses an estimated run differential instead of the true run differential?
Baseruns strips out the sequencing/luck to try and determine how many runs a team SHOULD have scored/allowed without said sequencing/luck but at the end of the day its still based on a run differential and would suffer the same flaws.
At least thats my understanding of it. Feel free to enlighten me (i’m not being sarcastic, please feel free)
How low does the home run % need to go before it is more likely to score off three singles than one rally killer?
Ahh, watching people who believe any deviation from projections is only explainable by luck argue with people who think their team is magical and basic math doesn’t apply to them warms the cockles of my heart.
Dayton: For 115 million, you built a playoff team. You lost Shields, Butler, Aoki and both Holland & Vargas to injury and you won more games without them than you did with them. You won 3 more games that the Dodgers won, but the Dodgers spent three point four million per win and you paid one point two.
I know you’ve taken it in the teeth out there, but the first guy through the wall. It always gets bloody, always. It’s the threat of not just the way of doing business, but in their minds it’s threatening the game. But really what it’s threatening is their livelihoods, it’s threatening their jobs, it’s threatening the way that they do things. And every time that happens, whether it’s the government or a way of doing business or whatever it is, the people are holding the reins, have their hands on the switch. They go bat shit crazy.
I mean, anybody who’s not building a team right and rebuilding it using your model, they’re dinosaurs. They’ll be sitting on their ass on the sofa in October, watching the Kansas City Royalswin the World Series.
What is there to emulate? They have 11.38 clutch the last 2 seasons while the only other teams to produce positive “clutch” value (only 5 teams) have compiled 11.42… If there’s a way to do it, only they know the secret. There aren’t other players out there to grab. They’ve cornered the market, if somehow Hosmer, Gordon, and Infante have “clutch” genes or skills.
Incredibly simplistic viewpoint.
The way to think about it is what skills and attributes constitute “clutch”?
As Mariano Rivera said during the WS Game 2 MLB Network pregame show, the last 3 outs of a game, and especially out #27, is much more different than the other 24, or 26 outs. Al Leiter agreed. Rivera claims the players have a greater sense of urgency knowing this is it, so it is harder to get them out. Perhaps guys like Hosmer and Morales can hone in on that sense of urgency better than others and get the adrenaline and focus needed to “keep the line moving” sooner than the 9th. Perhaps the “clutch” skill these Royals appear to have is some sort of a psychological adroitness.
“Perhaps guys like Hosmer and Morales can hone in on that sense of urgency better”
Uh….ok?
I wholeheartedly endorse this rampant speculation, and am discarding my previous theory that clutchiness is all due to titanium jewelry.
Infante does have “clutch” skills. Thanks for noticing. OmaRBI got 7 RBI in 1 game this year. Tied for a Royals record until Moose got 9 RBI a week later.
I think the bit about Holland and Vargas is not nothing. Keep in mind that Yost had both Wade Davis and Chris Young essentially acting as backups, so when we talk about “replacements,” in the case of the Royals a replacement could be better than the starter. Yet, the team generally stays loyal to the above average, but possibly not best, starting player. There is no “Wins Above Starter” stat line and starters usually start for a reason, but heck, look at Rios. You could have replaced him with Dyson or Orlando during the regular season and gotten better production, but that could have disrupted the team above the ears and changed how players batted, given the new guys ahead or behind them in the lineup. Yost waited and was rewarded.
Just so I get this straight: Holland + Vargas (4.4) dropping to (1.2) this year and missing most of the year isn’t much of a loss? A closer with 48 saves and a 1.44 ERA last year was just “replaceable” by Davis?
Then who takes Davis’ spot? And so on and so forth? It’s incredible how they overcame injuries just like other teams who could not.
Yes, that’s exactly what he’s saying. Replacing Holland/Davis/Herrera with Davis/Madson/Herrera was not much of a step back at all.
Yeah, like having Davis was just a random fluke instead of a plan.
No one said that. You are making things up again, you hack.
You guys need to watch more movies. 😉
Was hoping someone would get my Moneyball reference!
Oh we got it, it’s just lame as hell.
no you didn’t Drake. But I still like you
What about lineup stability versus volatility? It’s obviously different but, I used to try to rotate fantasy players in and out week to week but eventually concluded that I was most effective (and it less time consuming) when I simply picked my best above-average group that was solid in most, though not all, scored stats, and rode the ups and downs with very few changes. Maybe similar stability helps the Royals, not just psychologically as players but on average with results?
I think this is a huge factor because existing projection systems simply do not account for the full negative impact of running a succession of barely playable replacements. If I were in an FO and I were looking for a place where private information could lead me to find gaps to gain an advantage figuring out ways to determine which of my starting guys will show up to play is top of my list. And, in fact, I think that is exactly what the Royals have done.
And yet the lineup isn’t all that stable comparing last year’s to this year’s: only four guys (Escobar, Cain, Hosmer and Perez) are batting in the same spot in the lineup. In fact the roster isn’t even that stable: there are only ten guys on this year’s WS roster who were also on last year’s (that core four, Moustakas, Gordon, and Dyson, plus Davis, Duffy, Herrera)
Ventura
I voted higher than 84 not because of the Royals, but because projections are more compressed than results.
Church
If that was the case, then they would simply project higher by x wins or y percent.
That 84, with something like a 6-win standard deviation would be like 72-96 wins 95% of the time, I don’t think 84>81 suggests that the higher end is more likely.
It’s hard to accept there is anything more here than some very small beating the projections around the margins because of their bullpen and contact %. We’ve been down this road recently with the Orioles, and while they didn’t make it to a WS, they had a few year stretch where people were sort of relenting in assigning them as having some way of gaming the system, and, well, lo and behold, it all came crashing to an end. Are there maybe some very small, marginal advantages the Royals have that are sitting outside the projections? Yeah, and we’ve perhaps shown that with contact % and bullpen skill. Is it enough to toss aside their projections? There seems little evidence to.
Well said. We come up with a lot of post-hoc explanations until they don’t work, then quickly forget the whole thing and move along to the next projection-beater that has found a way to “beat the system”.
It’s akin to “I have the failsafe system in roulette! I’ve had three winning sessions in a row!”
Not even close. We have over 500 data points for the Royals the past 3 years, 92-win pace. That’s not 3 lucky hands in poker. THats…..THREE…..SEASONS
You’re right of course; any team beating a projection for three consecutive seasons (THREE WHOLE SEASONS!!!1!11!one!) is a true black swan.
When you have 30 teams and hundreds of sets of 3-season samples, a team beating its projection three consecutive years (or four, or five…) is a certainty.
As a wise man once said, “he who says ‘!!!1!11!one!’ often is making a very good point.”
Ugh, was waiting for the first cheap shot at the O’s in this post about the Royals. The O’s won 81 games this year (a few games below Pythag in fact). In the last four years of “this” Orioles team, they have won 93, 85, 96, and 81 games – that’s an average of 89 wins per year, folks. Don’t see how anything came “crashing to an end.”
Next year could be trouble with everyone we are losing plus the trades to thin out the farm system. But, while a .500 season with a plus run differential may have been disappointing, I fail to see how it invalidates their approach.
There is no magic. What there is are insufficient and mis-specified models. There are two fundamental flaws to the existing projection systems.
1) They use a season’s accumulated stats to model total wins for the season. This is fine as a first order approximation, however baseball records are not collections of stats but rather a summation from a collection of discrete events. To the extent that there is Royals “magic” it is that they have found undervalued ways (especially the importance of a dominant bullpen) to tip a specific set of these discrete events (close contests) in their favor. Emphasizing contact may also play a role in shifting the probabilities for this set of events.
2) Even taking into account the limitation of projecting records from accumulated statistics, the basic statistic used to measure this accumulation of WAR is clearly flawed. It does not accurately measure the value of relievers for one thing (because it relies unwisely on bullpen chaining). It does measure the full value of healthy players because it assumes that any missed games are freely filled by replacement value players (which simply is not true especially at skilled positions like CF and SS). It relies on applying positional adjustment values that are little better than pull-out-of-ones-ass guesses.
So, no, not magic. Just an analytical puzzle to be solved.
This was a great explanation. Well done.
they have found undervalued ways (especially the importance of a dominant bullpen) to tip a specific set of these discrete events (close contests) in their favor.
What evidence is there that they valued a dominant bullpen more than other teams?
we don’t need evidence that they “valued” this more, only evidence that they *have* a better bullpen.
If a dominant bullpen does tilt the odds more in your favor *more* than the projection systems would suggest, it doesn’t matter if KCR valued this more than other teams or just backed into it by happenstance… the outperformance versus the model would be the same.
it doesn’t matter if KCR valued this more than other teams or just backed into it by happenstance
Have you read the thread at all? This matters a lot.
I’m fairly sure WAR isn’t used in the projections at all, so not sure where point 2 comes in. I think team projections are based on projected player rate stats and play time allocations. Depending on the projection, I think they either use this info to simulate games or use something like Base Runs and pythagpat with a strength of schedule adjustment.
Point 1 comes close to begging the question. Yes, obviously the Royals optimally sequenced a season’s worth of discrete PAs and PAs-against. The question is whether they really “found undervalued ways” to do this or if the players simply ended up doing this beyond any reasonable expectation.
One observation: the success of the Royals bullpen is not just how good the pitchers are, it’s how each of those great pitchers (except Herera) individually has distributed his own performance among the different situations in which that individual player was used. That’s what the clutch scores measure.
I mean come on now. This stuff has been getting ridiculous! Your dismissal of “magic” is delivered like some bully telling a young kid there’s no Santa Claus. Even if your “insufficient models” will eventually account for every variable (which is the real belief in magic, no?), right now they do not.
People have ideas that lend towards explanations. Jeff has some pretty good ones. But in the mean time, why can’t people try and figure it out themselves? Or make up other narratives? Narratives that are more fun than numbers. Remember when we were young and thought our favorite players were superheroes? Were we idiots? Did we need Blue to come along and tell us our beliefs were misguided and we should scoop up his instead?
I think there’s been a real freak-out over this Royals team. And it isn’t that their fans are uneducated or overzealous; most WS fanbases are both. It’s that we, people into sabermetrics, have a real problem with change/uncertainty in our own narratives. Not that we don’t purport to welcome it, just when things are uncertain we harp as dogmatically as the crotchety writers that sites like this freed us from. I don’t want to drink a second batch of Kool-Aid.
I for one would be really interested in and really receptive to any rigorous statistical study of the Royals that gave some good, objective evidence regarding what we are missing about them. I don’t think it’s sufficient to point out what the Royals do differently, point out that for the past two seasons they’ve significantly outperformed their projections, and then attribute causation to those things they do differently. I have yet to see anything approaching evidence of causation, rather than just correlation. I eagerly await some evidence. Really, I’ll be the first to admit I’m wrong if someone, anyone can produce some Bill James-level sh– (see: longer than a few paragraphs in the comments to this article) backing up their claims. As Jeff points out, there is a lot of accumulated wisdom about baseball on one side of the scale and three seasons of the Royals on the other. Absent something more, I’m still inclined to think the weights favor what’s been convincingly demonstrated to this point. Honestly, I would love to be wrong.
Royals magic = playing real baseball
The rest of the league is busy promoting rookies and trading long balls for Ks. The royals are one of the few teams going out and playing baseball on a regular basis. The game has not changed, but the model for running a franchise has. Winning is not a priority for most organizations although they may not realize it.
What is “real baseball”? I’m not disagreeing with you but to say the other MLB teams don’t know what they are doing is a bit ridiculous, no?
The entire NL plays real baseball. The designated hitter is fake baseball.
Harrumph!
Grumble grumble!
I’ve never seen the “One True Scotsmen” analogy deployed in this way before. Thank you for opening my mind to new visions of illogic.
Most teams believe that a lead off hitter needs to be patient at the plate, work the pitch count, and a walk is as good as a hit leading off. Escobar typically swings at the first pitch. The Royals actually promote Escobar’s free swinging. The Royals use as justification their record with Escobar leading off is far better than when he is not, and when he swings at the first pitch the Royals record is even better. However, no true lead off hitter would ambush the pitcher.
And to think it’d all be disproven if the end of a game against the Astros went slightly differently…
I kid.
The Royals are good at everything you are taught to do from the age of 5. The team is loaded with top prospects. FanGraphs getting it wrong when it was staring you in the face isn’t my fault.
no kidding, couldn’t agree more. Fangraphs is just spewing out shit right now on Royals luck/magic, while I can quantify the Royals skill. They are #1 in offense by 150 points in the new metric I created called HEWCO. That is why they are winning. And this crap about Alcides Escobar NOT being a good offensive player is bullshit too. He is one of only 18 players that have a hewco rating of 700 or higher. The only one rated 800 is Jose Altuve. Let me know if you would like to see my research, imtigerwords@yahoo.com
Was just tigerwords@yahoo.com already taken or something?
not only was it taken, but the email is a play on words because I play in competitive scrabble tournaments across North America
he had tigerwords@prodigy.net for a long time, but was late to the yahoo! game
I think the thing that many of the projection models take into account is the compounded effect of having excellent defenders at every position and/or quality bats at every spot in the lineup. I suspect there are some scenarios where the team is greater than the sum of their parts. Two defenders who both have 10 DRS for a season individually may account for more than 20 DRS together because of positioning flexibility. I wonder if a defense with two 0 DRS guys and one 15 DRS guy would prevent fewer runs than a defense with three 5 DRS guys.
I think you are actually going to have diminishing returns by chaining good defenders in that you are just now more likely to have balls where to guys can reach them which isn’t that useful. You theoretically want your best defenders placed near your not great defenders (it is better to have a +10 SS and a +10 RF than it is to have a +10 CF and a +10 RF).
But there are NO bad defenders. That’s the point. Rios is the worst on the field and he’s probably avg. Gold Glove at First, SS, Above avg at 3b and 2b (Zobrist or Infante). GG in LF, CF as well as behind the plate. That’s a lot of extra results. Doesn’t matter about diminishing returns, b/c at this point they’re all just making more plays than a replacement player would.
I asked Elo about this in his chat today. He mentioned that it kight be the same as good rebounders stealing rebounds from each other. But think about Escobar’s HR. With two great defenders, the liklihood of that ball being caught is very high. With two average defenders (Cespedes is average in CF) you see what happens. Plus, with good defense all around, your corner OFers can play closer to the line, stealing more doubles. I don’t know. But this is definitely an unanswered question. I think Statcast defense numbers will help with this. But It is my hypothesis as to why he Royals outplay their predictions… their defense is better than the sum of its parts. Perhaps their O is, too, because they don’t have an easy out in the lineup.
have there been any studies comparing the values of a great defensive outfielder or two vs a great defensive outfield (all 3), on projected wins?
if not, can one of you mathematically capable people give it a shot?
Chaining–whether in bullpen analysis or in defense–is a stupid idea. It relies on brittle Min-max calculations and is not robust. Don’t use it.
The Royals are both a good team and a lucky team…why is this so hard to grasp? You don’t have to be lucky or good. You can be both. Our projections can tell us for the most part the good teams but not the lucky ones.
Zips for example this year had 3 teams meet the 90th percentile projections and 3 team meet the 10th percentile. Can the projections be improved? Yes. But there is no evidence that the Royals, or the Orioles, or the Cardinals have found anything that would lead us to think the projections are wrong in a meaningful way.
Stupid. You can be lucky in a playoff series, or maybe two, or maybe three. Hell, 10.
But you can’t be “lucky” over 513 games the past 3 full seasons. That’s 290-223 (.565) or 92-70 pace. 3 full seasons bro.
To be fair, Los did say “good team and a lucky team”.
Yeah!
http://www.fangraphs.com/blogs/the-orioles-and-accepting-random-variation/
http://www.fangraphs.com/blogs/clutch-baseball-teams-arent-clutch-baseball-teams/
Maybe it is the manager. Perhaps Ned Yost is one of the great ones who can take a collective bunch of average talent and consistently produce above average win expectancy. Yost took the smallest market team in Milwaukee who had losing season for 12 straight years and got them to .500 or better in 3 of his last 4 seasons. He then took over the second smallest market team in the Royals who had losing seasons in 15 of the prior 16 years including 4, 100-loss seasons in prior 9 years. The Royals have improved 6 straight seasons under Ned and are now in back to back World Series. Ned also has the highest winning percentage in post season (20 game min.) in the history of baseball. Perhaps Ned should be the one telling us what the secret is to the Royals magic
We’ll that’s sort of the debate here right? The projections were not that they were a good team, so we’re trying to explain both the good and the lucky with variance.
To quote: “Zips for example this year had 3 teams meet the 90th percentile projections and 3 team meet the 10th percentile.”
To restate (I think you and and I agree): I’m pretty sure 3/32 is just about 10%, and 29/32 is just about 90%
So the projections and their guardrails are … within the accepted range that they disclaim?
Why 32?
With 30 MLB teams, then the threes are exactly 10%
I’m not statistical guru but sooner or later doesn’t the clutch hitting/sequencing/magic clock have to strike midnight?
I think the “there is no clutch” thing is a more convincing argument against ex post facto narrative building (“destiny” claims or whatever) but not much else. Otherwise, you have to ignore the experiential and biological realities encountered by successful athletes in high pressure situations, like some Architect in the Matrix. An increasingly complicated model might capture it better, but I think it’s not convincing to say clutch doesn’t exist or that clutch performance will always revert to a non-clutch baseline eventually. For some players, it simply won’t.
Didn’t we just have this conversation last year about the Orioles? Seems to me the only real difference is that the Royals have done better than the O’s in the extremely small sample size that is the postseason. Color me unconvinced.
Also, as an aside, when we’re deciding that a team has found some sort of secret sauce that leads to playoff success, shouldn’t we pick a team that has, I don’t know, maybe won a title already? If the Mets win 4 straight, there’s going to be a lot of foolish-looking articles out there.
The Royals have had very low K%, the Orioles K% has been about average. The Royals have hit relatively few home runs, the Orioles have relied heavily on home runs. The Royals relief pitching has been fantastic, the Orioles relief pitching has been good but not great. The Royals defense has among the best, the Orioles have been a little above average. The two teams are completely different. Just because winning a lot of close games is a fluke for the Orioles, it doesn’t imply that it’s a fluke for the Royals.
I would refer you to the statistics for more than just 2015 if you actually want to be taken seriously with these claims.
Decent/mediocre rotations, great defense, great bullpens – O’s and Royals have had those in common the past few years. Royals offense has been more contact, less power while the O’s have been the opposite but the production has been very similar.
My opinion on this is simply that it’s due to statistical variability, and maybe some “Royals magic” (some characteristic of the team that isn’t accounted for with traditional projections, ie. their propensity for clutchness).
Remember when the Twins were doing well earlier this season primarily because of the unsustainable clutchness, and suddenly they weren’t good anymore?
Or maybe the Royals have some good mental skills coach who has taught the players how to always give it their best in the crucial situations.
Let’s remember as well that, before Dayton Moore and the Royals were Fangraphs Enemy Number One the previous “village idiot was Brian Sabean:
From 2009
http://www.fangraphs.com/blogs/organizational-rankings-19/
“Front Office: D
Brian Sabean’s history of transactions while running the Giants is a series of astounding moves. On one hand, he made one of the worst trades in recent history (Joe Nathan, Francisco Liriano, and Boof Bonser for A.J. Pierzynski), gave out the worst free agent contract ever (Barry Zito), has consistently overpaid for declining veterans (Matt Morris, Dave Roberts, and Aaron Rowand), and has overseen a roster that that hasn’t finished at .500 in four years. However, he’s also done a very good job of identifying veteran players who had some life left in the tank, and has gotten good value from signings that looked like mistakes (Bengie Molina, Randy Winn, Omar Vizquel) and has done a good job of developing and retaining some terrific young pitchers. Given his resources, though, he’s squandered too many good opportunities, and the Giants missed opportunities to win due his unwillingness to rebuild.”
The Giants would go on to represent the NL in 3 of the 6 next World Series, winning all three.
And the Royals, who have now represented the AL in 2 of 6 World Series?
“Front Office: D
Dayton Moore came to Kansas City with a sterling reputation for talent evaluation from his time with the Atlanta Braves. His hiring was supposed to signify a new and better way of doing things for the Royals. Since taking over, Moore has made a long series of moves that have significantly tarnished his reputation and harmed the future of the franchise. The team has taken a spread-the-wealth approach to free agency, throwing a decent amount of money at a lot of average or worse players, wasting precious resources that could have easily been better spent on other players or in player development. Instead of maximizing their return on investment, as other small market teams have successfully done, the Royals have bought mediocrity on the free market, and failed to improve their team’s chances of winning in the process.”
Dude, save yourself the time and just copy/paste this everyday. There’s really no need for you to type it out differently every day.
We get it. You think fangraphs is the dumbest bunch of dummies ever. Maybe…just go read some “smart” people if you hate it here so much.
Why is debate a bad thing? The Royals success should be fascinating to anyone with an analytical mindset. Outliers are cool! That’s why this site keeps discussing it. They absolutely might end up driving the next discovery in analytics.
Don’t fear challenging new info bro.
He didn’t suggest that debate and/or new information were bad things.
He may not have said it, but he almost certainly suggested it.
No, he didn’t. “You are repeating yourself” has nothing at all to do with “new information is bad”. In fact, if anything, it’s the opposite. It’s criticizing information on the basis that it is not new.
In fairness, I still think Sabean isn’t necessarily a good GM. The player development seems to be amazing but Sabean has made a lot of mistakes. Posey and Bumgarner have covered up a lot of that.
Drafting and player development is a pretty big part of being a GM of a pro baseball team, no?
3 world series championships in past 6 years…
you’re a fucking idiot.
Also remember that Dayton Moore and the Royal family, Fangraphs is enemy number one of the first “village crazy Brian Sabean:
From 2009. year
http://vvv.fangraphs.com/blogs/Organizational-Rankings-19/
Front office: D
Transaction History when Brian Sabean giant launched the most amazing series. First, he is one of the worst in recent history (Joe Nathan, Francisco Liriano and Boof Bonser for A.J. Pierzinski), a free agent from the worst (Barry Zito), he refused to recognise the veteran (Matt Morris, Dave Roberts and Aaron Rovand) and make sure your list is not completed within 500 years from four. However, it is also a good job in identifying experienced players had very little life left in the tank, which is a good value for similar defects recording (Bengie Molina Randy Vinn, Omar Vizkuel) and have a good job, develops and maintains a large number of young kr?aga. For its resources, however, he loses too much chance, and the Giants lost a chance to win because of unwillingness to rebuild. ”
These giants would come to represent the NL’s 3 Series 6 in the world, winning all three.
And the Royal family, which is now represented in G 2 of 6 World Series?
Front office: D
Dayton Moore in Kansas City with Sterling Canon Customer talents of his time with the Atlanta Braves. His work, it would mean a new and better way to work for the Royal family. Since taking office, Moore’s series of photographs is significantly undermined its reputation and hurt commodities in the future. It is to spread the wealth, free kicks to a decent sum of money in a large number of mid-and high-end games, wasting valuable resources that would be better for other easy to develop players or players. Instead of the maximum return on investment, as well as small group’s success in the market, the mediocrity of the Royal family bought on the open market and could not improve the chances of their team to victory in This process. “
kinda getting old, bro.
Fair enough. That’ll be my last post.
Serbian to Vietnamese to French and back — just save it for special occasions. Don’t abandon it entirely. I think it’s a funny and effective way to call someone on their nonsense.
Boo, it’s one of my favorite things here (and the closest thing we’re apparently getting to a much-needed gentle nudge in the direction of moderation)
No worries guys. I got it covered.
Doesn’t blue have a good point here?
Dunno if it’s being mocked because it was repeated too much or if he’s a troll or whatever but at this point it’s looking like 4 of the 6 most recent WS will have been won by perhaps the two GM’s and orgs (non Ruben Amaro and Philadelphia division) most vilified by fangraphs and much of the sabr leaning community over much of the last decade.
At this point I think the course of wisdom and of modesty would be for folks to take a long, hard look at the assumptions and models we’re working with, because there’s plenty of evidence that they may be quite flawed and greatly in need of further refinement.
Exactly my point.
This Royals magic or luck is all bullshit, it pure quantifiable skill, all of it.
Three formulas I use for evaluating offense are HEWCO, BSM, and CCR. I parse the offensive data differently too. The Royals are #1 in HEWCO by 150 points over every MLB team on offense and 300 points better than the Mets and that is why they are winning.
I can prove this, plus if someone wants to work with me, getting sequence data we can have whole new formulas and categories to offense that would make batting average, OBP, SLG, and OPS completely obsolete. It is going to require new fields and functionality though. If you want the formulas, email em at imtigerwords@yahoo.com
Eric laughs at all of this silly “debate”. He’s had this figured out all along! Idiots with your feeble “explanations”. Who knew the answers were so simple, and right in front of us all along.
The old guard of OBP, SLG, and OPS are dead. The Royals are far and away leaders in ZORK, BLUMF, and DINGUS. Email me for more info! gotitallfiguredout@geocities.com
imtigerwoods@yahoo.com— don’t tell us, tell GMs so that they’ll hire you to a seven figure deal.
I don’t think it is that complicated.
KC’s elite defense, bullpen, and contact ability helps them excel in high leverage situations. This helps them clamp down and keep a one run lead when they need it. This helps them keep a game close until their high contact, speedy hitters eventually find a way to manufacture another run or two to grab the lead and win the game.
The projections don’t like them because they don’t properly value these skills. Just lump it all in with “luck”, “randomness”, and/or “clutch” and call it a day.
They were ok at “clamping down” in close games, but so were most of the playoff teams. here’s leaders in games in 2015 decided by 2 runs or less: (from baseball-ref)
Rk Tm G W L W-L% RS RA pythW-L%
1 PIT 74 47 27 .635 280 259 .536
2 CHC 81 51 30 .630 312 283 .545
3 KCR 73 44 29 .603 281 257 .541
4 STL 85 50 35 .588 269 248 .537
5 LAA 80 47 33 .588 274 264 .517
6 TEX 82 46 36 .561 345 330 .520
7 NYY 76 42 34 .553 274 257 .529
8 MIN 75 41 34 .547 266 253 .523
9 NYM 80 43 37 .538 272 261 .519
10 SDP 78 42 36 .538 274 263 .519
maybe the padres are on to something!
It’s true that the Royals have led the majors in Clutch the last two years in a row, but the question is whether thats a fluke. Looking at career statistics, the only player on the Royals who looks like he may be one of the rare players who hits significantly better in high-leverage situations as a real ability rather than a fluke is Hosmer, whose career wOBA is .388 in HL situations (#12 in the major leagues) as opposed to .304 in LL situations (#211 in the major leagues). Perez and Cain also have HL/LL differences of 20+ points (.331 vs. .309 and .341 vs .321), which might be significant but isn’t so likely.
Contact hitting, defense, and relief pitching is also why the Giants have done so much better than projections in recent years.
I think it’s a combination of they’re an okay team, playing in a crappy division, and being quite lucky. We had this exact conversation for like 2-3 years about the Orioles. Their fans couldn’t get enough of rubbing it in everyone’s face about how wrong we all were, and now they’re back in last place about to lose half their team. A team like the Royals? I’d REALLY enjoy what you’re watching right now while it lasts, if I were a fan of that team.
This seems unusually self-proving, where teams predicted to be good but aren’t are unlucky, and teams predicted to be bad but end up good are lucky. And then when a successful “lucky” team turns bad, somehow that establishes they were bad all along.
You may want to revisit how each division performed against the others this year. The Royals had a winning record against the AL East, AL Central, AL West, and in Inter-league play. There are other divisions in baseball that fall below the AL Central’s performance.
Is “the Royals are lucky” the new “David Eckstein is gritty”?
No it’s the next “Put any statute at 1B who walks a lot and has power because defense doesn’t matter (cough, we can’t measure it well)” sabermetrics conventional wisdom circa 2003.
Good players are performing well, and the site that is supposed to explain the why and how is asking us to provide free labor?
I hate the Royals because they cause the anti-intellectual fans to come out of the woodwork and repeat the “Hurr Durr, put down the calculator and watch the game” memes.
Well, as a Royals fan who would prefer that you not hate my team, consider this: Any time someone or some team performs contrary to projections — or somehow appears to mystify stats-based analysts — those genies pop out of bottles and spew nonsense. Why? Because page clicks. So really you should just hate baseball. Or ignore the nonsense and enjoy the game we all love.
Seems like an odd thing to hang on the Royals as people make those statements literally every game. Even when a team isn’t consistently beating projections.
“no one’s ever shown a link between clutch hitting and not striking out.”
Maybe I’m mistaken, but I thought it was understood that there actually was a relationship between clutch hitting and not striking out. If I recall (this might have been discussed in the chapter on clutch in The Book, or in a thread on The Book Blog), players with high clutch scores tended to be contact hitters like Tony Gwynn, and guys with low clutch scores tended to be high K sluggers like Mike Schmidt and Sammy Sosa . . .
Projection systems got in wrong because the Royals have been a team of underachieving top prospects and a strong farm system for so long that it was tough to predict that Hosmer, Moustakas, Cain, et al would take a big step forward at the same time. Luck, as they say, is the residue of design. And when the team got lucky (Edinson Volquez with 2.6 fWAR? Come on…), they were able to take advantage. Then, they went all in and picked up the best all-around player on the market (Zobrist) who just happened to fill their two biggest holes down the stretch (LF while Gordon was out, and 2B thereafter, pushing out one of the worst regulars in the game, Omar Infante). Add to this a great bounceback season from Kendrys Morales, and you’ve got a team that had easily 6 wins greater performances than anyone could’ve anticipated based on prior evidence. Add to that some positive variation, and a lot of batted ball luck in key situations, and here we are.
I don’t think you need to reassess everything when a team outperforms expectations. You just need to look back at your expectations, and ask, “were they rational?” And I think, in this case, they were. So many things broke just right for the Royals this year, from the rise of Paulo Orlando on down. And they managed to address the things that went wrong at the trade deadline, using the last of that strong farm system, and now they’re healthy and peaking at the right time. Kudos to the Royals, but the projections were fine (maybe a *little* low, but close to enough) based on what we knew.
“I don’t think you need to reassess everything when a team outperforms expectations. You just need to look back at your expectations, and ask, “were they rational?” And I think, in this case, they were. So many things broke just right for the Royals this year, from the rise of Paulo Orlando on down. And they managed to address the things that went wrong at the trade deadline, using the last of that strong farm system, and now they’re healthy and peaking at the right time. Kudos to the Royals, but the projections were fine (maybe a *little* low, but close to enough) based on what we knew.”
When you’re on year 3 of pretty grossly under-rating the same team with the same core group, you have to at least evaluate it. The Royals do a lot of things at the extreme…high contact, low K rate, superior defense, great bullpen with high usage. I mean, from an analytical view, it’s pretty likely that they actually DO break the model in some way. The question is how.
It reminds me a lot of the super low tempo college basketball teams in Ken Pomeroy’s system. He specifically wrote about Wisconsin, because he acknowledged that, even though he knows they break the system, it’s not clear how to fix it.
And as I’m alluding to in other posts, it’s not just constant underestimating of the Royals over a three year period, but constant underestimating of a quite similarly constructed Giants team over the last 5-6 years.
We should probably stop questioning the Giants and Royals at this point and start questioning why we were so wrong as to question them so harshly and incorrectly in the first place.
If the projections were fine, I’d hate to see what bad projections look like to you!
Let’s see your better ones!
This is so tired.
Things only “broke right” for the Royals if you ignore all the things that didnt go right for them. Things like;
-Omar Infante becoming the worst position player in baseball
-Alex Gordon exploding his groin and missing 2 months
-Greg Holland falling off a cliff and having TJ surgery
-Alex Rios being pretty much exactly what he was projected to be (replacement level)
-Escobar regressing back to being one of the worst hitters in baseball.
-Guthrie being the worst starter in baseball and ~2 wins under what was projected from him.
-Yordano Ventura struggling for most of the year.
-Sal Perez regressing and being worth about half as many wins as he was the past two seasons (and half what he was projected for this year)
There is a hunger and intensity to succeed, to overcome any obstacles…that cannot be figured in with a stat model.
“Luck” and “randomness”, in the end, are weasel words, like “team of destiny” or “magic”: they don’t explain anything and are just used as shorthand to say you don’t know (or care) what’s really going on. I went off on this topic on FanGraphs already, but I’ll repeat: 1.) if you have a bunch of living, breathing human beings engaged in an activity, then that activity is not going to be completely quantifiable, because people are not robots, computer programs, or the laws of physics. Period. End of. What is so hard to grasp in that? Mathematics can only go so far, and beyond that you have to deal with emotion, irrationality, interpersonal relationships, and all that messy stuff that can’t be captured in algorithms. It’s neither magic nor fate nor random, just the human experience. Baseball is more quantifiable than most things, which is why analytics go so far in helping us appreciate it. But they can never capture the whole story. (Yes, I used the word “story” and I’m not taking it back.) 2.) Again I ask you: what am I supposed to believe, my own two eyes watching the play on the field, or your projections? What is more real in the end? If your projections have consistently failed to capture the empirical reality they are supposed to predict, then change your damn methodology; don’t complain about the realia that the methodology can’t handle. I find it ludicrous that so many allegedly rational people seem so angry about the Royals’ success.
The more angry ones seem to be Royals’ fans who just can’t understand why the projections “underrate” and “disrespect” them so. Just like Orioles’ fans from prior years, and Giants’ fans, and…
I also enjoy the railing against projection systems that aren’t 100% right, 100% of the time. If you have a better model that is a better fit for all teams, feel free to share! (Not saying the PECOTA projection or FG projections are infallible; it’s just a lot of carping after-the-fact with little “let me show you my better system” BEFORE the season.)
I can promise you that Royals and Giants fans are some of the happiest fans in baseball right now.
Seeing your favorite team raise WS and pennant flags is a nice consolation for not being a Fangraphs darling.
Ho hum.
You would think, but there’s *so* much butthurt about how a system could dare expect them not to win 120 games and 8 out of 10 world series!
I’ve spent a lifetime doing research and associating with others who do the same, in a variety of fields. It gets drilled into your head from early on that, as a scholar, your job is to be objective and deal with the reality in front of you, whatever that is in your case. True, we all come up with theories and analytic models as a way to make sense of what we study, and every one of us has been guilty at some point of trying to jam the square peg of recalcitrant data into the nice, round hole of our theory. But whenever that happens, someone will let you know: your professor, an editor, colleagues at a conference; and when that happens you go back to the drawing board, and tweak your theory; and, if you find you’re tweaking it all the time, you scrap it and start over. True, there are always crazies who believe that their model trumps reality in some way and can never be wrong, but the rest of us assume that our theories are good if they account for most of the data and we have a good way to account for the rest. However, we always have to be on the look-out for convenient “garbage cans” into which we can safely dump all the stuff that doesn’t work–that is, essentially meaningless categories that are defined vaguely and only generate more questions than answers. In my experience, such garbage cans are particularly prevalent in the social sciences, whenever there are quantitative models that try to explain human behavior: these always work up to a point, but beyond that point, “people be people”, things get messy, and the elegant bonsai gardens of theory get overrun by the urban vacant lot of human imperfection.
In this particular debate, “the Royals are lucky” and “oh, that’s just randomness” are two such garbage cans. I would argue that the railing which you mention is, in the end, largely directed at them, not at the projection systems themselves. Well, that, and the interconnected phenomenon of sabermetricians getting so defensive that they lash out at the realia that has bolixed up their models instead of simply trying to perfect their models, to the extent that you wind up having a forum like this with well over 100 posts, with real invective directed toward Royal fans, Ned Yost, etc.
There’s something wrong with that. Just like there is something wrong with essentially dismissing the results of the past two AL seasons as “random”. Again, if the models you use to make sense of baseball can’t account for what the Royals have done, then change your damn methodology, and stop deriding them as “essentially a .500 team”, because they clearly aren’t.
In the meantime, I’m going to enjoy the rest of the Series, because the Mets are pretty good and the Royals are an exciting team to watch.
Very, very well stated. Wish I had written this!
I would have no problem if folks were like “Yeah, we have a prediction. It looks like it has some validity and is a good starting point but it is clearly missing a lot. Oh, and here are the error bands around my point estimates.”
Instead it’s “See! Science!”
So this is what nags at me:
Did Dayton Moore and the front office really, truly see this coming, or have they back-ended into something special and maybe fluky or fragile or unsustainable or you-pick-it description of mostly lucky.
I say this because, well, I am among the countless who have panned nearly all of his major moves, and lamented plenty of his minor ones as well. And I say this because someone above compared DM with Sabean, citing so many heavily criticized trades and signings from both general managers, using the success of both to bludgeon stat-based analysts.
But DM and Sabean differ in the very thing that was mentioned: nearly all of the trades and signings for which Sabean was roundly criticized were failures, and his team’s have succeeded despite those moves. Dayton Moore’s moves, on the other hand, have mostly worked well for the club, and the team has succeeded because of them.
Not everything has hit for Dayton, but his ledger looks solid right now, and, really, for me it goes all the way back to the Gil Meche signing, which nearly everyone said was a massive over pay, and about which the Royals made mention of some mechanical change which they thought would improve Meche, and, well, the Roayls were right, until they sacrificed Meche at the alter of over-use.
So, again, have they really seen these things in real time, or are we praising luck in hindsight?
Sorry, this is long, and maybe the same thing Jeff and everyone else is pondering at the moment, but it kind of sends me in circles.
Really, you’re going to claim that most of Sabean’s debateable moves have been failures?
If the GM of a team that’s won 3 of the last 5 (6) World Series has mostly had “failures” when he’s made moves that anyone else disagreed with, then what on earth could you say about the moves made by the other GM’s who haven’t even sniffed a WS in that time period?
It seems that you’re either grading things on a ludicrously steep curve, or are desperately trying to stick to a clearly failed narrative, that the Giants somehow have won 3 World Titles in spite of Sabean, which is beyond ludicrous.
I do agree with your thoughts on Moore, but I think this whole situation should simply lead to the conclusion that we don’t know as much as we think we do, about anything.
To clarify, I’m by no means a Giants fan. I’m a Mets fan in fact.
But that some folks are still on here expressing doubt about both Sabean and Moore after seeing how the last 5 years have turned out seems to me to be the height of delusion.
We’re not just talking about one or two fluky WS appearances from the SABR ‘bad boys’ at this point, rather an incredible run of dominance from Sabean’s Giants that the Royals appear to be in a position to equal or at least challenge.
I think that for most rational people, the ship has to have long ago sailed on the idea that either of these teams or orgs are winning in spite of themselves.
At this point there should really be some massive soul searching from the fangraphs and SBR communities as to just what’s wrong with our assumptions, biases and models.
What could you say about the other GMs? You could say that they don’t play the games. You have to evaluate on process not results when there is as much variance as there is in baseball.
That assumes we know, objectively, what the process should be. But we, clearly, do not.
Royals are a fairly young team and many of the players haven’t established a mean to regress to, as Perez, Hosmer, Moustakas, Escobar, and Cain continue to improve. Add to that that the projections thought ’14 was Morales’ new normal and ’14 was Volquez’ outlier and projections weren’t even in the correct area code. There’s a big difference between dirt and spread sheet.
I admit not being entirely clear on how the Team Wins projections work, but I am looking at the 2015 ZiPS projections for hitters I downloaded on April 5, and some of the Royals numbers are damn close (Hosmer) and others are not (Cain). What seems to be fact is that the position players with the most PAs (my cutoff was 500) were over projections or at projections with the exception of Escobar.
Hosmer – accurate (damn near perfect projection actually)
Escobar – under projections (BA/OBP/SLG all under)
Morales – over projections
Moustakas – over projections
Cain – over projections (power numbers were significantly up)
Perez – at projections (some under some over)
For giggles, the 2014 ZiPS from March 24, 2014 look like this (same 500 PA cutoff yields everybody)
Gordon – accurate
Escobar – over
Perez – at projections
Butler – under (lost his power)
Infante – under
Aoki – at projections
Hosmer – under (power down)
Cain – over
Moustakas – under
So it looks like the hitters projections were averaged out for 2014, but low in 2015.
I’m too tired to look at pitchers projections vs actuals tonight, so maybe somebody else is interested enough to do it.
As much as I hate to say this, being an Angels fan, the Royals right now are looking a lot like the Angels’ teams from 2002-2009:
Good contact hitters
Decent starting pitching
Excellent bullpen
Consistent outperformance of their win expectancy
Go back to 2002 and look at how the Angels obliterated teams with their offense and shut them down with their bullpen, led by Troy Percival, with a late-season assist from K-Rod.
This formula netted the Angels five division titles and a World Series in eight years, so (unfortunately) I would not be surprised to see the Royals’ run last another five years–if they manage things just right.
And the same is true of the Giants since 2010, same formula.
A big part of the problem, why the Royals outperform statistical expectations, is that WAR doesn’t give enough value to great relieving. Wade Davis’ WAR this year was only 2.0. Out of the 141 starting pitchers with at least 100 IP (so basically the top five starters on every team), that puts him even with Mike Pelfrey, #71, right in the middle (and less than Wade Miley (2.6), Bartolo Colon (2.5), Andrew Cashner (2.3), Jesse Chavez (2.1), and Robbie Ray (2.1), just to name a few). Does anyone seriously think that Wade Davis’ value to a team is the same as that of average major league starting pitcher? On the other hand, Davis’ WPA value was 4.01, #6 of any pitcher, starter or reliever. WAR relies heavily on FIP, and FIP doesn’t consider when a PA takes place, so it’s problematic for relief pitchers, whose appearances vary wildly in terms of leverage, so that it doesn’t make sense to weight them equally (this is obviously much less of a problem for starters). For example, from the point of view of a closer who starts the 9th inning with a 2-run lead, it doesn’t (and shouldn’t) matter whether the first batter hits a home run or a single, and they will generally pitch that way, but the difference between a home run and a single in terms of FIP is enormous. WPA is intended to deal with this problem, by relying heavily on situational context, but the problem with WPA alone is that it doesn’t account for luck in terms of BABIP, which is especially important for relievers because they face so few batters. The best solution to this problem, I think, is to value relief pitchers with a combination of FIP and WPA. For example, WPA could be understood to be the basic evaluator, but whereas home runs, walks, and strikeouts would be calculated as WPA is normally calculated, balls in play would be calculated according to some standardized expectation for what a ball in play will produce (that is, a certain number of outs, double plays, singles, doubles, triples, etc.). This would obviously be pretty complicated, but it would give a much better idea of what relief pitchers are really worth to their team.
I am anyone. I think an slightly above average starting pitcher is more or less exactly what his value is. Basically war plus roster effects. You can only have so may starting pitchers, so adding more slightly above average ones has greatly diminishing returns, and there is a lot of roster value in having your best few RPs be very good.
But the projections already take this into account.
There is not even close to enough evidence that the models are getting something wrong here. We have a dice with an unknown number of sides and an unknown numbering system, and our model is pretty good that most of the time when we roll it we can make ok predictions.
Now we just rolled three 8’s in a row on our mystery die when the mode says that is extremely unlikely and everyone is like “model broken”. It is idiocy.
There simply is not enough data to think that the Royals represent some definitive flaw in system without some actually compelling argument about how and why. And so far the cases people present are mostly garbage and amount to “the system undervalues the things the Royals do”.
You sound like someone who thinks actual baseball is a pale imitation of fantasy baseball.
??? No just someone who doesn’t accept conventional wisdom on stuff like this because the more we look into it the more it is obviously terrible.
Having an ace closer is great. Worth several wins a year. SO is having a slightly above average 4th starting pitcher instead of a below average one. So are a lot of things. The idea that there is one way to build a baseball team is stupid and have been invalidated by this very season, as well as many past ones. Some teams have good bullpens, some do not, some hit for power some do not.
Let me get this straight. By your reasoning, if you were in charge of the Royals, if the Mets were to offer you Colon in exchange for Davis for the duration of the World Series (assuming trades like this were allowed), you would accept (2.6 WAR vs. 2.0 WAR)? Seriously? It sounds to me like you are blindly accepting conventional thinking in the sense that you are assuming WAR is the answer to all statistical questions. I was the one who was suggesting something new here in terms of statistical measurement, not you.
See, this is the type of wooly thinking that confusing season aggregated WAR with actual marginal importance leads to.
FanGraphs WAR includes leverage in WAR calculations for relivers.
http://www.fangraphs.com/blogs/war-and-relievers/
It still uses chaining, which basically invalidates it.
The Royals are a long standing, well-crafted work in progress…I’ve watched most Royals games over the past 8 years….They’ve added wins every year since 2007!….that’s a progression that isn’t figured in projections at all. It’s a very rare feat.
This one is easy. From 2013-2014 the Cubs greatly underperformed their projections, pythagorean wins, base-runs wins, whatever.
A fangraphs article pointed out other teams that did this for two years in a row, and found that the third year was completely uncorrelated.
So likely, the Royals will not overperform again next year.
It should also be considered that the Royals are an extreme. They don’t strike out much. Projections are an approximation, and approximations like this are often inaccurate at extremes of a variable.
So maybe call the inaccuracy of projections based on the extreme value of the variable of strike outs???
They also are somewhat etreme on outfield defense.
Or they are just very unconventional and not likely to be accurately projeced based on models that fit conventional teams.
There was an X Files episode once, about a MiLB player who was actually an alien. He loved baseball, and was very good at it (because … alien!), but wanted to keep a low profile. So every time a scout was in the stands watching, he’d strike out or drop easy flies, or some other crap.
Clearly the Royals have a roster full of aliens. Except for Escobar. He’s a real live magical wizard.
https://en.wikipedia.org/wiki/The_Unnatural_(The_X-Files)
I just find the whole expectation that “THE SABR PEOPLE HAVE TO DO SOUL SEARCHING” because a projection systems get a team wrong, absolutely ridiculous.
Big deal, the models are wrong about something. That’s why they’re models, and not baseball games. Find me a person with a fixed and repeatable methodology who will do a better job of projecting the performance of a thousand players and 30 teams than PECOTA does.
At the end of the day, you play the games for a reason. You don’t just run projections, or poll a bunch of sportswriters. They’re playing the games, the games are awesome, and the Royals are winning. The fact that you have to play the games doesn’t invalidate the success the projections have had in the aggregate.
Point to these “successes”, please.
Could there be amplification effects from the Royals’ team being built counter to current baseball trends? In other words, something like (just an example, not saying this is what is happening), strikeouts are up so defense is less emphasized but the Royals strike out less and so exploit that opportunity, teams steal less so catcher throwing defense becomes less important and the Royals exploit that opportunity by emphasizing speed and stealing more. In other words, the advantages that they have loom larger in results without showing up in the underlying data, which is contextualized in a larger context that de-values them.
This kind of thing definitely happens in other sports. The 49ers in the early 80s set up the run via the pass against 50+ years of football history and exploited a huge advantage. Now every team does that and it’s not an advantage. Similar thing happening in the NBA right now in terms of stretch 4s. I think the argument has always been that baseball is an individual sport masquerading as a team sport and so not subject to effects like this. Maybe that’s worth another look and the Royals are the case study?
This thread is divided into two groups, one of which argues that the Royals are outperforming expectations because the existing model undervalues contact hitting, relief pitching, and defense, and the other of which argues that the Royals are just getting lucky and we can’t draw again conclusions from these two years. Well exactly the same factors were characteristic of the Giants in their three World Series titles in 2010, 2012, and 2014, and in all three cases they also outperformed predictions. The people who say that this is all just luck will have a hard time explaining all of this, unless they simply claim that everything is luck and they care more about the model than the results.
Well at least that didn’t misrepresent and grossly oversimplify the arguments being made!
The problem is that you are making the World Series out to be a referendum on the model.
There are thousands of players and 30 teams. A model’s job isn’t just to tell you how good the Royals are. Or to answer the single question “Will the Royals win the World Series – Yes or No?”
The objective with a model isn’t a crystal ball, it is to find a consistent and methodological way to guess at thousands of player and team variables and predict the performance of 30 teams, while minimizing average error across the whole data set. If people think this same task can be accomplished differently, all they have to do is do it themselves and their numbers will be on these pages instead of Dan Szymborski.
You’re downplaying the model to the point it’s made meaningless. Saying that most teams will be average, and some will be above average, and some will be below average, and there’s chaos too so leave the model alone… suggests there isn’t really a “model” to speak of. A model predicts, and if it’s consistently predicting incorrectly, that’s a problem with the model, not the subjects the model is trying to measure.
No, models should certainly not be evaluated by minimizing “average” error. They should be evaluated by minimizing squared error.
I have a feeling that the trading deadline (trades in general really) throw a wrench into the whole preseason projection thing. If a mediocre team spikes the first half, picks up reinforcements at the deadline, then plays like a true talent good team the rest of the way, the projections are going to undershoot them. Likewise, a good team can underperform in the first half and punt the second half.
Good point.
By the way, are we even having this conversation if the Royals were eliminated in the ALDS?
Are we having this conversation if the Royals were from New York or Los Angeles?
If there is one thing I am confident of in this whole discussion, it’s that geography has nothing to do with it.
We may be having it, but about Astros rather than the Royals.
Often, the best models we have don’t work especially well on some cases. Shouldn’t throw out the model; should just try to see what it is about the cases the model is missing on and see if there’s an opportunity to update the model. I don’t believe most of these projection systems are necessarily “bad”, I think they may not fully account for something the Royals do well (perhaps not handling extreme contact hitting skills correctly?) So I guessed this hypothetical Royals team performs a few wins better than the projections. But I don’t think we should fault these projections (or projections in general).
“No one’s ever demonstrated sustainable offensive clutchness”
Maybe I’m misunderstanding this. Plenty of players have a positive career clutch value?
I believe that magic is real–for one season. A team catches fire and even the fans cannot explain it. But they regress and worse the following year.
The Royals are not that. Their model is not only different, it sometimes violates sabermetric rules, which at one time demolished conventional wisdom but now is conventional wisdom itself.
A walk is as good as a hit only if you hit bombs. If you are on a budget that is not a good model. If you are on a budget, paying for aces is not a good model. Relief pitching is in the budget. So apparently is defense. The Royals are playing their own brand of Moneyball, as are the Astros.
-Projections say Royals will win 72 games.
-Royals win 95 games.
-The majority of FG commenters continue to insist that the models are fundamentally sound — instead, it’s the Royals who have gotten historically lucky. Not just for the past week. Or the past month. Or the past season. But for the past three years.
-Said commenters cannot come to terms with the overwhelming probability that the Royals truly are undervalued by most sabermetric projection models.
Models are definitionally imperfect and flawed. They are an incredibly useful way for us to approximate reality based on the best information at hand. Occasionally, unconventionally built teams will expose weaknesses in the models, precisely because there are few precedents for these models to draw upon. This is completely expected behavior and is in no way an indictment of PECOTA or FG. Going forward, the models will continue to be refined to better reflect the unconventional team’s previously-unaccounted-for strengths. Rinse and repeat.
Royals fans have the right to feel aggrieved by the talking heads on this site who do not understand this dynamic.
The Mets were only projected to go 82-80 by PECOTA, but I have heard considerably less complaining from their side, despite there presumably being more Mets fans and my geographic proximity to them.
I think that’s just a reflection of the reality of how these team are built.
The Mets are an entirely different team from pre-season. Probably half the roster are guys who came over in season or rookies who had no good baseline at the beginning of the year.
The Royals are basically the same team as they were in pre-season, with the (fairly major) exceptions of Cueto and Zobrist.
I hope everyone in baseball tries to mimic the Royals and then I win a zillion-dollar lottery.
The first thing I’d do is buy the Rockies. Then I’d move the fences in, throw out the humidor, and assemble a roster of 25 Adam Dunns.