The Recent History of Teams Like the Royals

As you probably have heard by now, our projection systems don’t like the Royals, again. Our Playoff Odds page has them forecasted for just 77 wins, with only an 8% of reaching the postseason, the lowest of any team in the American League. ZIPS and Steamer just aren’t that high on the team’s individual players, and since the projections are context-neutral, there’s no adjustment being made for the fact that the team has won more than expected in recent years. The Royals have relied heavily on context-specific performance to reach the postseason, and projection systems assume that’s not a sustainable skill, wiping it away at the start of each season.

On Friday, I posed a question to you guys, based on the crowd’s overwhelming response that they believe the projection systems are underrating the 2016 Royals. The response to my question was also overwhelming; you guys believe that the Royals are going to significantly outperform their BaseRuns record once again. The top four answers selected in the poll were the four options that had them beating their BaseRuns record, with 78% of those voting selecting one of the options that suggested the Royals have an inherent skill that BaseRuns isn’t accounting for.

Overall, by weighting the results by the proportion of people who voted for each option, you guys project that the Royals will beat their BaseRuns expected record by 3.4 wins in 2016, accounting for about three-fourths of the difference that Jeff Sullivan found when he polled the crowd about expected record versus the projections. Given that the Royals have beaten BaseRuns by an aggregate 25 wins over the last three years, our readers believe that there’s some real skill there. You don’t expect that they’ll get the same type of bump as they have the last few years, but you’re willing to assume that, at this point, BaseRuns is just missing something about how they play, and the forecasts are low by 3.4 wins because of it.

To follow up on that expectation, I wanted to look at how other BaseRuns-beaters have done, and not just teams that have had an outlier year here or there. The Royals are now on a three-year run of significantly outperforming their context-neutral expected records, and so I wanted to see how other teams who had put up similarly strong performances over three-year periods performed in the fourth year. To do that, I took our BaseRuns data — which stretches back to 2002 — and created three year rolling totals for every team, which gave us results for the 11 seasons from 2002 to 2015; 330 three-year stretches in total. Of course, the ones that run 2013-2015 don’t give us any information on the 4th year yet, so we’re really looking at 300 data points that tell us how the team did in the year following a three-year run of crushing their BaseRuns record.

Of course, because what the Royals have done is pretty amazing, there aren’t that many teams that have put up a +25 win difference over a three year stretch in our sample. In fact, teams had only eclipsed that mark four times, and three of them came from one team; the mid-2000s Angels. Over a four-year stretch from 2007 through 2010, the Angels beat their BaseRuns record by a whopping 46 wins (topping out at +17 in 2008), so the rolling averages that include those years all show up at the top of our leaderboards. So, if we just tested the Royals against other teams that won 25 more games than their BaseRuns record over a three-year period, we’d really just be testing how the Royals stack up against the Angels from a few years back, plus one Astros team that managed to go +30 over 2008-2010.

That’s not a particularly big sample, and wouldn’t really give us much of an answer, so instead, let’s break things down by deciles. Taking the 300 data points we have, we can look at them in blocks of 30, and compare how they did in their three-year run to the fourth year. The results of each deciile are in the table below.

BaseRuns 4th Year
Decile 3 Year Average 4th Year
1 to 30 19.0 3.2
31 to 60 11.2 1.1
61 to 90 6.7 -0.5
91 to 120 3.2 1.0
121 to 150 0.2 -0.1
151 to 180 -2.0 -1.4
181 to 210 -4.3 -1.8
211 to 240 -7.0 0.2
241 to 270 -10.6 -0.5
271 to 300 -16.5 -1.4

I can’t imagine a much better result for the crowd; the average BaseRuns differential for the top decile was almost an exact match for what you guys projected the 2016 Royals to beat their context-neutral numbers by. The 30 teams who beat their BaseRuns record, by an average of 19 wins over the prior three years, do indeed show that there may very well be a sustainable trait that these teams had in common that allowed them to continue carrying their BaseRuns-beating ways into the future. Certainly, there’s some pretty strong regression to the mean, as the top decile lost about half of their per-season ability compared to the prior three years, but they didn’t regress back to zero.

Interestingly, the results aren’t symmetrical; the teams that underperformed BaseRuns by the largest degree didn’t continue to underperform at the same rate as the overperformers. While the fact that bottom deciles continued to underperform shows that there is some deficiency that these teams weren’t able to fix entirely, they regressed much closer back towards neutral than the top decile did. These results suggest that teams that dramatically underperformed their BaseRuns last year — the Reds, A’s, and Astros being the three most prominent examples — shouldn’t be expected to diverge by more than a win or two this year. And while the Reds are rebuilding, so they might not get a big boost in the standings from this regression, the A’s and Astros could both appear to be significantly better this year simply by not being quite so terrible in clutch situations.

But this post isn’t about the A’s and Astros; it’s about the Royals, and the crowd’s remarkable ability to eyeball something very close to what the data confirms. Based on the last decade or so of teams who have beaten their BaseRuns performance in a substantial way, we should probably expect the Royals to do so again in 2016, and the +3 to +4 win margin that you guys suggested appears to be in the right ballpark. They won’t get as big a boost as they did in 2015 — only four of the 30 teams from the top decile put up a +10 or better season in the 4th year — but these results do suggest that these teams have found something that allows them to outperform by a decent margin after they’ve been identified as teams that might have this skill.

What might that trait be? There have been lots of suggestions, mostly centering around contact ability and bullpen makeup, and so tomorrow, we’ll investigate the 30 teams from that top decile and see if we can find some commonalities. We might not find anything, and this might remain a bit of a mystery for the moment, but least this initial look at the data suggests that you guys are right to be skeptical of the idea that the Royals are going to perform at their BaseRuns record in 2016.



Print This Post



Dave is the Managing Editor of FanGraphs.


Sort by:   newest | oldest | most voted
RedsManRick
Member
2 months 20 days ago

You cite the Reds, A’s, and Astros as notable examples of teams that underperform their BaseRuns. Their team offensive K% were 12th, 2nd, and 29th best (lowest) respectively while their pitching K% were 17th, 21st, and 12th best (highest) respectively. While I would not be surprised to find that contact rate is a factor, clearly it’s hardly deterministic.

I took a quick look at the standard deviation of batters with 300+ PA and found the Royals were not more “consistent” throughout the lineup than the underperformers as some might suspect (they were actually comparable to the Vottos… err Reds!). I also looked at bullpen performance in high leverage and while the Royals were good (2nd), the other 3 teams were all over the map (11th, 23rd, 30th). It will be interesting to see what you come up with!

Richie
Member
Richie
2 months 20 days ago

Thank you for your own quick look here, RMR. Very helpful. Well, maybe not for Dave. :-)

tz
Member
tz
2 months 19 days ago

I think the reason why there’s not symmetry between the Royals and the BaseRuns underperformers is that they’ve constructed a team going against the prevailing grain by focusing on guys whose value comes mostly from non-TTO talents. So for the underperformers, the results are mostly random variation, but for the Royals it’s a combination of random variation and intentional design.

Baseball4ever
Member
Baseball4ever
2 months 19 days ago

I have said it before and I’ll say it again – TTO baseball is failing baseball on the offensive side. I researched it, I proved it, and I created HEWCO, CCR, and BSM to replace the slash line, BA/OBP/SLG/OPS. Its a better predictor of what is going on and how to set your lineup.

Serbian to Vietnamese to French and back
Member
Serbian to Vietnamese to French and back
2 months 19 days ago

I have said before and I’ll say again to attack the cancellation baseball baseball. I was wondering, do I have to prove myself and HEVC, which would reduce the time and BSM, three times / / SLG / OPS instead I created. It is a better idea about what is happening and how to adjust the settings.

Jason B
Member
Jason B
2 months 19 days ago

I mean, he ain’t lyin’. Alcides Escobar had a league-leading 16.8 HEWCO’s last year. Did you ever hear about that on any baseball broadcasts? HMMM?

Jason B
Member
Jason B
2 months 19 days ago

John Fogerty is still the all-time leader in CCR, though.

Shirtless Bartolo Colon
Member
2 months 19 days ago

You should have seen me getting out of bed this morning. Now that’s a bad moon rising.

Cool Lester Smooth
Member
Cool Lester Smooth
2 months 19 days ago

I think part of the Royals’ craziness is the combination of a lineup full of contact hitters with a lockdown bullpen, not just one or the other.

KCDaveInLA
Member
KCDaveInLA
2 months 19 days ago

May sound simplistic, but many blogs and thinkpieces assert that ‘there has never been a team like the Royals’. So, how would you go about predicting a team like that?

Ernie Camacho
Member
Member
Ernie Camacho
2 months 19 days ago

Why would there be any special synergy there?

Ernie Camacho
Member
Member
Ernie Camacho
2 months 20 days ago

I’m pretty surprised there was such a big bump for that top bucket-very interesting stuff. As a skeptic of the “it’s a skill” hypothesis, I really look forward to the next piece.

Oh, and maybe nitpicking, but those buckets are “deciles” not “quintiles.” For those of us who naturally zero in on the tables first, that minor error adds a bit of unexpected confusion.

vecnyj
Member
Member
vecnyj
2 months 19 days ago

It doesn’t have to be a skill, it could just be a gap in the BaseRuns formula that over/under estimates certain teams.

Ernie Camacho
Member
Member
Ernie Camacho
2 months 19 days ago

In theory, yes. And that just underscores the importance of Dan Rausch’s request below that FGs disclose its BaseRuns formula.

jscape2000
Member
jscape2000
2 months 19 days ago

I wonder about the baseruns consistency of other teams year to year. Dave mentions that underperformance and outperformance don’t regress symmetrically.

We expect teams at the top to take steps back as their best players age or leave. We expect teams at the bottom to rise by acquiring talent through the draft or through free agency. But nothing says the two movements are equal.

Given the pressure on teams to contend, versus the natural progression of aging & service time, can’t we build a wider “contention window” graph, and then control the results above again our window? Is the sample size just not big enough yet, or is there a layer of complexity I’m oversimplifying?

Damaso
Member
Damaso
2 months 20 days ago

Great stuff.

I wonder if there is any easyish way to factor in roster turnover and/or age into the analysis. KC keeping the same roster largely intact and largely prime aged might indicate even more reason to assume they can continue to beat baseruns, compared to some previous baserun overachievers who may have lost key personnel by that 4th year. Or not.

Richie
Member
Richie
2 months 20 days ago

Prime age sure shouldn’t matter, as the projections fully take that into account. Roster turnover now, maybe.

Cool Lester Smooth
Member
Cool Lester Smooth
2 months 19 days ago

The projections don’t take overachieving BaseRuns into account at all.

Which is why we aren’t talking about the projections, here.

tz
Member
tz
2 months 19 days ago

I think Damaso is on to something here.

How many times have you followed a young player whose role gets jerked around, loses development time, gets frustrated, etc.? Dave L’s Sunday column mentioned Matt Murton, who’s a good example.

So if you believe (as I do) that it’s possible to derail a player’s development by changing their status a lot, shouldn’t there also be some value in letting guys go through their growing pains and sticking with them the way the Royals have with Moose, Hosmer, etc?

And this would show up as a bias towards beating their projections, even factoring in age curves etc.

Jim
Member
Member
Jim
2 months 19 days ago

Yes, I remember chatting with a KC front-office type at Surprise, AZ 2-3 years ago and mentioning that Moustakas hit only .181 against lefties the season before. And his response was, well, he’s young and we’re going to let that play out.

Ernie Camacho
Member
Member
Ernie Camacho
2 months 19 days ago

That might help explain players beating their individual projections, but I don’t see how it necessarily gets at the BaseRuns question being asked here.

evo34
Member
evo34
2 months 19 days ago

Cause vs. effect. Players who do well tend to keep stable roles more than the other way around.

Paul22
Member
Paul22
2 months 19 days ago

Excellent point. This fact may make it hard to find out much about these teams unless its controlled in some fashion..

scooter262
Member
Member
scooter262
2 months 20 days ago

My best guess is that there is something about this group of players — specifically the lineup — that is doing something “magical” (e.g. difficult to quantify).

The 2016 Royals lineup is about the same as the 2015 version, and my belief is that the sum is greater than the whole–again, in some way that is difficult or impossible to measure. So, yes, I do think they will again significantly out perform there BaseRuns and overall projections.

Of course, it could all be noise, and the 2013-2015 Royals were very fortunate in their sequencing.

kevinthecomic
Member
Member
kevinthecomic
2 months 20 days ago

It is articles like this that need to be cited whenever some blowhard starts going off about how FanGraphs is ‘against their team’ or ‘made up of a bunch of nerds who don’t understand the importance of chemistry’, etc.

As evidenced by this article, FanGraphs is more than willing to challenge their current assumptions in light of new information. FanGraphs is not made up of ideologues who espouse only their current thinking. FanGraphs is in pursuit of the ‘truth’ and if the ‘truth’ goes against their current assumptions, then they are more than willing to challenge (and if warranted change) their current assumptions. This is what makes FanGraphs so gosh darn interesting and informative. Thank you and keep up the good work.

MLB Rainmaker
Member
Member
MLB Rainmaker
2 months 19 days ago

C’mon, I’m 100% not a Royals fan, but Dave has consistently bashed the Royals for three years now, starting with this gem:

http://www.fangraphs.com/blogs/royals-mortgage-future-to-be-mediocre-in-2013/

Which he followed in 2013 and 2014 with at least a handful of pieces essentially saying, “Sure, the Royals have done better than I said, but they will revert to the mean” Except the Royals didn’t….they went to the WS back-to-back and then won the 2nd time. So now 4 years later….Dave is finally, sheepishly, admitting he was wrong.

And its 100% OK to be wrong, you can still write great stuff and miss plenty, but go back and read those articles and tell me there isn’t a pompous tone to them that doesn’t benefit this site.

Serbian to Vietnamese to French and back
Member
Serbian to Vietnamese to French and back
2 months 19 days ago

At 100%, I’m not a big fan of the royal family, but Dave still break the king a series of three years, from this little jewel:

????://???.?????????.???/?????/??????-????????-??????-??-??-????????-??-2013/

Then in 2013 and 2014, there is at least a few songs basically says, “Of course, Royal did better than what I said, but it will be high,” the royal family. .. To go to the VS line, and then won 2 hours. And now four years later. Dave finally admits it was wrong visezih.

And it’s 100% goes wrong, you can always write wonderful things i did not have much, but go back and read the article and let me know if not arrogant tone to them do not use this site.

Shirtless Bartolo Colon says in Serbian to Vietnamese to French and back to John Elway
Member
Shirtless Bartolo Colon says in Serbian to Vietnamese to French and back to John Elway
2 months 19 days ago

Just broadcast names, names, names, names. The seals will clap?

Beer
Member
2 months 19 days ago

lolwut?

Jordans_80_Hit_Tool
Member
Member
Jordans_80_Hit_Tool
2 months 19 days ago

To be fair there was a lot of negativity throughout the baseball media to the trade analyzed in the linked post and the comments on the post detail the collective distaste for it among the Fangraphs community. IIRC, the highest profile media figure that defended the trade (that I followed) was John Manuel who liked that the Royals got a #1 starter for a prospect he had questions on due to Myers’ high K% in the minors.

KCDaveInLA
Member
KCDaveInLA
2 months 19 days ago

I admit that I have a hard time reconciling my appreciation for advanced stats with reliance on them for prediction models. The real fact is: it’s impossible. Every year there will be outliers that beat individual players’ perceived capability (or don’t). Every year there are many Lorenzo Cains who demonstrate double the value of what was predicted, and just as many Hanley Ramirez’s that completely crater out. Sabermetrics give you a pretty good idea of what to expect from a given player or team, but they’re just not always right, that’s all, there’s always variance.

Jason B
Member
Jason B
2 months 19 days ago

And, at the end of the day, let’s not forget – WHO CARES WHAT ANY PROJECTION SYSTEM OR PREDICTION SAYS ABOUT YOUR TEAM. You can predict my favorite team to go 0-162. If they win 100 games and the World Series (or even 84 games and the World Series)

whether by sheer skill or incredible luck, or any combination thereof…

whether with a +300 run differential, or a -30 run differential…

whether with a $30MM payroll or with a $300MM payroll…

whether picked to win their division by every outlet or by none of them…

whether with the talking heads and prognosticators heaping praise or fury upon them…

WHO CARES. Your opinions about my team and its players (or Dave’s, or FanGraphs’) don’t impact what actually happens on the field a bit. You win the World Series or have an otherwise successful year? Just bask and enjoy it.

Lawd.

*hops off soap-box*

Jason B
Member
Jason B
2 months 19 days ago

Amended to add, that’s not directed at KCDave at all. Just piggybacking on his (correct) assertion that every year will see some outliers, and it seemed a good a place as any to decry the fans of (insert team here*) decrying that they don’t get “any respect” from some outlet or writer.

*Insert team here has been the Royals, Orioles, and Giants most often over the past five years on FanGraphs.

t
Member
Member
t
2 months 19 days ago

“…the belowest of any team in the American League. ”

The belowest! I like it but I would have gone with “blurst” there. Because it’s been the best of times for the Royals and projections say the blurst of times are coming.

MLB Rainmaker
Member
Member
MLB Rainmaker
2 months 19 days ago

This article is another cop out by Cameron. Instead of saying, “hey, maybe our weighting of variables is wrong in computing WAR and we need to re-assess our view as a community”, he is again saying the Royals are just lucky, as “context-specific performance” is a poorly-veiled euphemism for “luck is a skill”. That view is counter to the whole basis of this site….you can’t ignore data because it doesn’t fit your model….you need to go back and review the model.

The Royals do three things at an elite level — play defense (#1 UZR), make contact (#1 Contact%/#4 lowest SwStr%), and bullpen pitching (#2 ERA)– and its clear that the formula for WAR doesn’t factor defense appropriately, doesn’t account for contact rates at all, doesn’t value innings pitched based on context, and only provides value on a linear basis, not accounting for marginal value.

The reality of complex models is that as you build in more and more complexity, the windows where the model break become larger and larger, and that is what the Royals performance is pointing out about the prediction models. Calling it “context specific skill” isn’t helpful for the process of understanding those outcomes and attempting to address the weaknesses in the prediction models.

zkello
Member
zkello
2 months 19 days ago

I believe there is an article around here somewhere that explains that the the biggest difference between the Royals projections and their performance was not related to defense or pitching, but timely hitting. Compared to MLB average the Royals were able to get hits in important leverage situations at a higher rate. Because this has not proven (yet) to be a repeatable skill, this facet of their game is attributed to luck.

Serbian to Vietnamese to French and back
Member
Serbian to Vietnamese to French and back
2 months 19 days ago

This is Cameron’s second policy. Instead of saying, “Hey, maybe we should go to the weight of the computer war, and we need to re-examine their positions as a community,” he said, which is the royal family and the family is blessed “the specific context of the play” is a euphemism for poor coverage “luck is a skill.” This view is consistent with the fundamentals of this site … you ignore the data because the model does not fit … you have to go back and check the model.

The Royal Family do three things in the defense of the elite of the game (# 1 UER), contact (# 1 below SvStr Contact% / # 4%) and heavy at the meeting (# 2) – and all the obvious, this is a recipe for war are not protected by inappropriate factors , general contact rate, not the value of the contracts, and only in the context of value based on a linear, rather than limit value.

The fact that the complex patterns until you build the complex, which shattered windows form becomes greater, and this is how efficient is to look for models of Royal forecasts. They call context-specific jurisdiction “is not helpful in the process of learning about the successes and efforts to address weaknesses in the models predicted.

Snarfle
Member
Member
2 months 19 days ago

One thing I remember about those Angels teams: super aggressive. Swung a lot (Vlad!) and ran a lot. That’s also true of these Royals, especially the latter. I wonder if there’s some synergy around running and putting the ball in play that Base Runs is missing–especially if your team has to really be built for that to actually benefit.

Robert L
Member
Robert L
2 months 19 days ago

Those Angels teams from 07-10 had a BB% of 8.1% (league average was ~8.8%), and had the 20th lowest Swing%. They stole a lot of bases (3rd) but were only ok in BsR (15th). In both ERA and FIP they were 12th.

But wOBA in High Leverage:

Hitting – T-5th

Pitching – T-3rd

Mike
Member
Mike
2 months 19 days ago

Dave, do you have any significant tests for those figures? i.e. how many are those are statistically significant from zero? or is the number of observations just too small?

Bryon
Member
Bryon
2 months 19 days ago

One challenge of Baseball projections though is the relative proximity between “good” and “bad” or 1st place and last place teams within a division as compared to other sports. For instance, the worst record in the NFL was shared by the Cleveland Browns who went 3-13 while the 1st place team in that division went 12-4 (Cincy). The Bengals had 400% more wins than the Browns with a win pct. spread between the two of 0.562! The best record in the AL in 2015 was KC at 95 wins and the worst was DET at 74. The win pct. spread between the two was 0.126. Now I realize that a 16 game season has the potential for a much higher deviation due to the smaller sample, but accurately projecting a baseball teams win total to within 10 games is almost numerically equivalent to projecting an NFL teams win total to within 1 game. Again, I know these two are very different in many ways, Im just saying that the degree of accuracy required to nail down every teams win total is impossible and tweaking the model to account for the Royals (who I am a fan of) isn’t warranted since the model does well projecting most teams.

Antonio Bananas
Member
Member
Antonio Bananas
2 months 19 days ago

On top of the contact and bullpen hypothesis, I wonder what breakout seasons do. The Royals core all kind of made strides in single years. Hosmer, Moustakas, Davis, Cain, etc. projections seem to have smoother progression. If a guy jumps from 1 WAR to 3, that’s probably not expected in projections.

Antonio Bananas
Member
Member
Antonio Bananas
2 months 19 days ago

In other words, age of players. Guys 23-26 going from 0.5-3 WAR or 2-5 WAR. Get enough of these guys together and you get huge swings.

BigChief
Member
Member
BigChief
2 months 19 days ago

This article is about out performing Baseruns, not projections.

amaahs
Member
Member
amaahs
2 months 19 days ago

Qualitatively the over-performers have a non-sabr reputation while the under-performers are more sabr inclined. It reminds me of an ongoing debate in basketball where analytics loving teams attempt 3 pointers and dunks/layups at greater rates because these shots are “more efficient.” But, of course, the type of shot that matters is one that goes in, so simply forcing shots deemed to be more efficient on paper don’t necessarily help the team. So teams which should have a higher shooting efficiency under perform while teams that take the dreaded long 2 can thrive if the shots are uncontested.
I’m not 100% sure how to relate this back to baseball but I’d be interested in seeing if any on-field strategies baseruns typically frowns upon aren’t actually hurting the Royals (or perhaps helping their sequencing of events). Equally interesting would be if there were some way to show the A’s were playing simply to meet context neutral goals at the expense of real life runs.

TL;DR – The teams used as examples over performing seem to have a non-math reputation, while the under performers are more on the math side. Coincidence?

MGL
Member
2 months 19 days ago

For your historical dataset, what is the baseline? Are they actual pre-season projections based on Steamer and ZIPS or are they just the teams “projected wins based on components?” What components do you use to project team wins? Surely you’re not just using s,d,t,hr, and bb for an against, while ignoring base running, wp, pb, stolen bases, etc.?

In other words, what does this even mean?

“Given that the Royals have beaten BaseRuns by an aggregate 25 wins over the last three years.”

I know what a BaseRuns formula is/does, but how did you go from BaseRuns to a w/l record? Again, BaseRuns normally does not include base running, stolen bases, WP, and PB, or even ROE for that matter.

Ernie Camacho
Member
Member
Ernie Camacho
2 months 19 days ago

From answers in past chats, Dave has confirmed that FGs uses a BaseRuns formula that includes SB and CS, but I’m not sure if it’s something like Tango’s complete formula or one of the intermediate ones that Smyth or Patriot came up with. To be honest, I’ve tried a few and have never been able to reproduce FG’s BaseRuns estimates.

To get a w/l record, I’m pretty sure FGs just applies PythagenPat to the R and RA estimates that come from BaseRuns. I should really let Dave answer these, but not sure if he’s still reading comments.

johnforthegiants
Member
johnforthegiants
2 months 19 days ago

One really striking statistic about the 2007-2010 Angels is that their AVG against was .265 in low-leverage situations (#27) but only .233 in high-leverage situations (#4). Their pitchers’ K% was 17.6% (#20) in LL situations but 19.9% in HL situations (#7). While this may have been luck, it may have also reflected a particular strong pattern of using their most effective pitchers in high-leverage situations. During this time they also had by far the most saves of any team, 199 (the second highest was 177). Their closer in 2007-8, Francisco Rodriguez, had the highest pLI, InLI, gmLI, and exLI in the majors in those years, indicating that he was used extremely effectively. In 2009-10, though, it’s less clear. The main closer was Brian Fuentes, and although he did get 71 saves, and the team was tied for 5th in saves with 90, his LI numbers weren’t nearly as high (8th, 3rd, 3rd, and 17th).

Paul22
Member
Paul22
2 months 19 days ago

After the 5th inning, the Royals out scored their opponent 319-248. From innings 1-5 they outscored them only 405-393. So close to a 500 team innings 1-5 and a 625 team (100 W) in innings 6+

Going into the 6th inning they were 84-10 when tied or ahead (894 WP). League average was a 753 WP. That’s 14 wins right there

When scoring 3-4 runs, they were 31-9. League average was a bit below 500 WP

For whatever reason, clutch, bullpen, luck, whatever, they simply are or were a much better team after 5 innings. The SP was not great but could keep them in the game for 5 innings. They also had the highest OPS in late and close situations in the AL and were 3rd in R/PA in the AL, and the bullpen gave up the fewest R/PA in the AL in late and close situations.

tz
Member
tz
2 months 19 days ago

Great info Paul22! The Royals’ massive improvement in performance over the later innings is something that is not captured at all by BaseRuns, but can easily be worth several wins on its own. And, rather than being a byproduct of luck, much of that difference can be explained by their deep and terrific bullpen (and perhaps a bit by the contact hitters in their line-up faring better against power relievers than most clubs).

Ernie Camacho
Member
Member
Ernie Camacho
2 months 19 days ago

Note, though, that looking only at actual runs scored misses half the point of BaseRuns. It could very well be that the Royals’ opponents had more raw offensive than the Royals did in innings 1-5 (i.e., hits, walks), but because of sequencing, it didn’t result in a negative (for the Royals) run differential. If so, that would mean it was innings 1-5 that explain most of the BaseRuns discrepancy. That would mean the bullpen had nothing to do with it.

Paul22’s stuff is a great first step, but it barely scratches the surface.

tz
Member
tz
2 months 19 days ago

Agreed, it could be driven by sequencing luck. My gut reaction to Paul22’s stats is that this seems to be too dramatic a difference to be just sequencing luck or other random variation, but like you said we’d need to dig deeper into those numbers to see what really caused them.

Dan Rausch
Member
Dan Rausch
2 months 19 days ago

Hopefully without stating the obvious, it seems that the discrepancy in the BaseRuns calculations is either due to something that is being calculated incorrectly or due to something that is being omitted. Based on my cursory understanding of BaseRuns, here are some possibilities for the former:
– BaseRuns treats all outs the same – there is no difference between strikeouts and other outs. However, outs based on balls in play have the opportunity to be productive: move runners ahead, score on sac flies, etc. The Royals avoided strikeouts like crazy last year, close to 300 missing strikeouts relative to an average team. That is a lot of extra opportunities to move runners that BaseRuns is ignoring.
– I’m not sure how the BaseRuns coefficients are calculated, but my guess is that they are specific to the current run environment. If that is the case, then it is possible that the Royals bullpen plus defensive replacements could alter the run environment significantly enough that the coefficients used for the Royals aren’t necessarily appropriate. You probably need a pretty significant difference, but I would speculate that a great, deep bullpen could contribute enough innings to create some divergence.

And for the latter:
– As MGL stated above, BaseRuns ignores baserunning. I’ve seen some implementations that use SB, CS and GIDP, but nothing that incorporates the ability to go first-to-third, tag on sac flies, etc.

Of course, without being able to look under the hood and see the actual implementation that Fangraphs is using, this is just speculation.

Ernie Camacho
Member
Member
Ernie Camacho
2 months 19 days ago

I would love to see the BaseRuns formula FGs uses.

I’d also love to see some more analysis about how, mechanically, the Royals have beaten BaseRuns. The “clutch” stats offer some key clues, but the pitching clutch conflates so many different factors (individual pitcher clutch; bad mop up guys; awesome back end), so is much harder to interpret.

tz
Member
tz
2 months 19 days ago

I think a team with the kind of early/late game scoring splits that the Royals have (see Paul22’s summary above) should mechanically produce a very strong team “clutch”. I’ve poked around at this a little bit, hopefully will find some conclusions.

Ernie Camacho
Member
Member
Ernie Camacho
2 months 19 days ago

For sure, and the Royals 2015 batting clutch was extremely high. I believe Jeff mentioned in the comments to one of his posts that it was in the 96th percentile of all seasons going back to 1974. That immediately suggests that defense and lock down rotation may not be the main forces at play here. [Though defense was always a red herring, because it should mostly be accounted for in BaseRuns estimates.]

People casually mention “contact hitting” but Jeff’s work showed an extremely weak tie between contact and clutch. And in any case, this should be fairly easy to investigate. Did the Royals gain an unusual number of runs through contact events, like sac flies, groundouts, and ROE?

Ultimately, I would love FGs to commission a simulation ninja to test some of these theories in a cleaner way.

Ernie Camacho
Member
Member
Ernie Camacho
2 months 19 days ago

Oops, that should be “lock down *bullpen*.”

KCExile
Member
KCExile
2 months 19 days ago

Ernie’s comment is interesting. Looking back on some of the ways the Royals scored runs it was incredible. It was never more obvious in the playoffs but they would get like two runs with one hit that left the infield.

In one notable instance in 2014 while fighting to qualify for the Wild Card they beat the White Sox by scoring the tying and winning run against with both runners scoring from second base without a ball leaving the infield. One of the craziest things I’ve ever seen back to back.

http://m.mlb.com/video/topic/6479266/v36295001/cwskc-dyson-steals-third-ties-game-on-wild-pitch

http://m.mlb.com/video/topic/6479266/v36295093/cwskc-gore-races-home-on-cains-walkoff-single

tz
Member
tz
2 months 19 days ago

The Royals got about 1 extra win of batting clutch last year just because of how the “clutch” stat is constructed. It’s not all from contact, but about equally from the bias in clutch favoring low K%, low BB%, and low HR%.

Ernie Camacho
Member
Member
Ernie Camacho
2 months 18 days ago

That’s a great point. I’m still wrapping my head around the exact contours of the clutch stat and those biases, but I’ve read enough about them from kincaid and Tango to trust them.

I don’t think it changes the overall point, though. Just compare the 2015 WPA of the Royals lineup to that of their most likely 2016 rival, the Indians. Identical team wRC+; team wOBA within 3 points, yet a 12 game delta in batting WPA and 7 game delta in “clutch.” And the Indians were also a good contact, low power team, if not quite to the same degree as the Royals. It’s hard for me to believe that vanilla batter sequencing didn’t do the heaviest lifting in their separation last season.

matt
Member
matt
2 months 19 days ago

Royals remind me of the Braves of last year

Joe Blow
Member
Joe Blow
2 months 19 days ago

Remember when Rany Jazayerli was propped up as someone who knew what he was talking about?

I mean, you can try to come up with projections, but it’s really clear who actually understands baseball as it’s played on the field, and theoretical garbage..

evo34
Member
evo34
2 months 19 days ago

The guy’s a baseball genius compared to the current BP staff. Then again, so is my pet iguana.

matt
Member
matt
2 months 19 days ago

I like the BP staff with the 1 obvious exception

matt
Member
matt
2 months 19 days ago

Projections have a far better history projecting W/L than people

wpDiscuz