The anti-science behind anti-analytics
I don't listen to a lot of talk radio. Well...I don't listen to a lot of talk radio that isn't Howard Stern, but I don't necessarily consider him a normal talk radio host. What I do listen to a lot on my daily morning drive are podcasts. I guess that's essentially talk radio, but instead of being subject to whatever the channel is talking about, I can pick and choose the topics I want to hear.
One thing I didn't miss however was local 610 KCSP morning host Bob Fescoe. I didn't listen live, but instead to the segment posted on the 610 website, which you can find here. On his segment he discusses a fine article written by August Fagerstrom on Fangraphs on the Royals and their 2016 projections.
Now let me get something out of the way first. This isn't going to be my first disagreement ever with Fescoe. In fact, it's not even my first disagreement with Bob this week.
No. How does that site know what anyone should be getting and ur only worth what someone will pay https://t.co/jn0gTh73rN
— Bob Fescoe (@bobfescoe) December 18, 2015
Fescoe has either never heard of intrinsic value or doesn't believe in it. Which is odd... https://t.co/vQrEAYxirq
— Shaun Newkirk (@Shauncore) December 18, 2015
@Shauncore your value is what u get not some made up perceived value
— Bob Fescoe (@bobfescoe) December 18, 2015
I don't know anything about Bob other than the views he's expressed on the few times I've caught his show. I'm sure he's a fine, normal human being. However from what I can tell based solely on his view and approach to discussing baseball, he's a very old-school, traditionalist thinker.
One can tell he doesn't understand or care for the use of even simple analytics like how small the impact of lineup construction is, or that value is provided elsewhere than just home runs.
He's never hit 20 hr and hits 8th smart not to over pay https://t.co/uhBE7ZprUy
— Bob Fescoe (@bobfescoe) December 12, 2015
Leaving aside the trolling by saying Gordon "hits 8th", this statement is wrong. Alex Gordon has hit 23 home runs in 2011 and 20 in 2013.
But this is a different discussion. On his show yesterday, Bob briefly summarized August's article about how the Royals didn't project very well for 2016. He brought up how they beat their projections in 2014 and 2015, how projections are missing variance and luck, and that they aren't 100% accurate.
Fescoe's argument is not necessarily missing information, but he is misinterpreting what projections and analytics mean.
Let's start with essentially the crux of his entire argument. It is a great introduction that lays out essentially what he, and many others like him, either don't understand, don't care or fail to understand, or can't understand about understanding analytics in baseball.
I originally went into this link thinking one thing and then once I saw the numbers that they were looking at I thought...I need a cocktail and I don't really care about a lot of these numbers.
There it is. Immediately out of hand he dismisses the entire approach by Fagerstrom and the advanced (which they aren't that advanced) numbers used. Despite not caring about the "numbers", Fescoe proceeded to spend the next 5-10 minutes talking about those very numbers.
Here's another problem as well with getting into this segment: the preview on the site, which describes this segment as:
Fangraphs comes out with the projections for the Royals 2016 season, we talk with former Chiefs Danan Hughes, a checklist for Alex Smith & Daily Top 10; gifts
FanGraphs didn't come out with the projections just today, or any day necessarily. What these projections are are a daily update of the team(s) for 2016 based off of the moves they have made in the previous days of the offseason. FanGraphs doesn't really come out with proprietary projections themselves at all, nor do they do it on a particular day. What they do do is essentially combine ZiPS and Steamer projections with their own estimates of playing time. This "projection" is essentially accessible at all times throughout the year.
Remember last year Bill Pecota had the Royals winning 72 games
A couple things here. Am I missing a joke or a re-occuring thing that Fescoe does? Bill Pecota has nothing to do with the PECOTA projections other than being the un-eponymous name of the system.
PECOTA of course has nothing to do with FanGraphs, and aren't available on their site. PECOTA is a product of Baseball Prospectus. I'm sure Fescoe would be surprised to find out that Rany Jayzayerli (previous guest of his show and various 610 sports radio shows) worked for Baseball Prospectus.
...all the projections were stupidly dumb for the Kansas City Royals because while statistical analysis has its place there's also room for the personal side of things, the human element that I think doesn't get factored in when you have statistical analysis. I mean you can't always use the eye test and you can't always you stats.
Great point here by Fescoe. He should have stopped talking after this sentence or so. For just a few brief moments, he may have understood projection systems.
Maybe we can try to show you that it takes a good mix. After all, we know that projections are missing normal variance in their projections. Projection systems don't know that 20% of the team will get injured (like the Rangers), guys will have monster break out seasons (JD Martinez), or things that are just beyond knowing really.
Even the "experts" at ESPN, who likely used a good part of the eye test to make their considerations got a lot wrong.
Most of them predicted the Orioles to win the AL East. FanGraphs didn't think so (based on their April 1st projections) and saw them as a .500 team (which they were). In the AL Central pundits were everywhere, predicting every team but the Twins to win. Of course the the Royals won their division while the Indians (projected winners) won 81 games (FG projected 81 wins). In the NL West, everyone was betting on the Mariners and FanGraphs liked them too, projecting them for 88 wins and the division title. It ended up being the surprising Texas Rangers (which absolutely no one saw coming).
For the NL East everyone was picking the Nationals too, and much like the Rangers, the Mets surprised everyone by winning the division. FG projected the Nationals to win 93 games (much like basically everyone in baseball). The NL Central was across the board between the Cardinals and Pirates. FanGraphs projected the Cardinals to win the division and the Pirates to finish behind them with 85 wins. FG got the order right, but under-projected both times by some measure. Finally in the NL West the Padres got some love while the Dodgers were the long favorites. FanGraphs projected the same outcome and was spot on with the win projection.
Division
Expert Pick
FG Pick
Winner
Decision
AL East
Baltimore
Boston
Toronto
Neither
AL Central
Cleveland
Cleveland
KC
Neither
AL West
Seattle
Seattle
Houston
Neither
NL East
Washington
Washington
NY NL
Neither
NL Central
St Louis
St Louis
St Louis
Both
NL West
LAD
LAD
LAD
Both
So both the projections and the humans came out the same essentially. ESPN didn't ask the experts to give a win/loss total for each team so we just have to look at their division winner picks.
Something else is interesting to me too there...a few of the teams that won their division were complete surprises in the Blue Jays, Rangers, and the Mets. Also the Astros went from a terrible 2014 to a playoff appearing 2015 team. The projection system didn't call that one either, but neither did the experts.
That's where variance comes in, heavily. As mentioned earlier projection systems don't know 100% of all the factors that will decide a team's season. It doesn't know what week the Astros were going to call up Carlos Correa, or that the Rangers were going to acquire Cole Hamels, the Blue Jays acquire Troy Tulowitzki/David Price, etc. Once those trades and injuries happen then the projection systems adjust (like humans do) but humans don't know about this stuff either until it happens.
Statistical people say "there's now way you can use the eye test" and eye test people like me look at the stats and say "to hell with the stats, throw them out." There's gotta be something in the middle, that we can all come to an agreement on.
Literally in first sentence there Bob admits to looking at stats and throwing them out, yet he's calling for some middle ground?
So the stat nerds are telling us right now right...right now that the world champion Royals are going to be the sixth-worst team in baseball. Going into the 2016 season. Huh? Huh? Going in to the 2016 season Fangraphs.com as of today, the 22nd day of December 2015, before free agency has really begun, I mean there's so many moves out there, so many moves that can be made, they're telling us right now that the Kansas City Royals are the sixth worst team in Major League Baseball.
Yes, and what Bob is seemingly not understanding is that this roster is still unfinished. This is parallel to the comments I made above that moves are still to be made.
But at this point, the Royals, are an incomplete team that if field right now wouldn't be a World Series contender.
The 2016 Royals are currently without a:
LF
RF
SP
2B who isn't awful
They've got at least 3 major holes to fill, as well as they could use an upgrade from Omar Infante at second, who by all means is probably the worst second baseman in baseball right now that could possibly get 500 plate appearances.
Unless you are really bullish on Jarrod Dyson in right field/center, Paulo Orlando in left field, and Kris Medlen going from okay 50 IP in 2015 to pretty good 180 IP in 2016 then the 2016 Royals aren't the best team even the AL Central.
Maybe Bob wants the projection systems to predict players that will join the team? I don't know, but he admittedly knows the projections are working with an incomplete Royals team, yet judging them and bashing them like they are working with all the information.
...But they've come up with some futuristic prognostications of stats that say the Royals right now are no good...
This won't be the first time you'll hear Bob today either talk down to stats or mystifies something simple like basic projection systems.
Again...he repeats that he knows that the projections aren't working with full information.
The Royals did a whole lot even though Cueto didn't do a whole lot in the regular season for the Royals, right?
Here is where we'll start a string of Fescoe evaluating players 2015 performances, especially those acquired at the trade deadline.
Cueto was worth 1.1 wins for the Royals in 2015. That would be a 2.5 win player (above average) over 180 games.
Johnny Cueto's addition did not get the Royals in the playoffs..
Well sure. No individual player effected the Royals 2015 playoff appearance. The Royals could have traded Lorenzo Cain and still made the playoffs.
However it's completely false to that Cueto didn't have a positive impact on the Royals. He was a good pitcher in his time there, and considering the replacements if the Royals didn't trade for Cueto he was even better.
The roster that won the World Series had an Alex Gordon and a Ben Zobrist in the lineup but this one doesn't. Didn't they go like 34-17 without Alex Gordon? Way over .500...and weren't they a playoff team before acquiring Ben Zobrist?
And this might be the biggest strawman of them all...
Is Fescoe really, really, really arguing that the Royals weren't better with Gordon and Zobrist in the lineup? Citing a stretch of games when Gordon wasn't in the lineup is irrelevant to how the team would have performed with Gordon. In fact it stands to argue that the Royals would have been better with Gordon in the line up.
Alex Rios was one of the worst everyday players in baseball in the 105 games he played. The Royals record when Alex Rios played?
66-39.
Funny how folks are concerned about Holland when Hererra has been the biggest issue in the pen all year
— Bob Fescoe (@bobfescoe) September 12, 2015
The Royals record when Kelvin Herrera pitched? 47-25
@vtachvfib dyson and orlando are part time players. value with them was limited exposure
— Bob Fescoe (@bobfescoe) December 7, 2015
The Royals record with Dyson in the lineup? 63-27
The Royals record with Orlando in the lineup? 48-38
This is an easy trend to figure out...the Royals did a lot of winning regardless of who was in the lineup.
The Royals acquired Cueto on July 26th. From July 26th to the end of the season the Royals went 37-35. The Royals would have been worse if not for acquiring Cueto. They would have been worse if Alex Gordon never came back. They would have been worse if Ben Zobrist wasn't acquired.
It's impossible to say that they wouldn't have made the playoffs, but it's illogical, false, and blatantly wrong to say they weren't a better team with Cueto, Zobrist, and Gordon. That's the point August Fagerstrom is making.
Those guys didn't truly make a difference as to why the Royals were in the postseason
Royals record before Alex Gordon's injury: 46-32 (.590)
Royals record after Alex Gordon's injury: 42-32 (.568)
"...and it sure doesn't look like they're getting Gordon back"... how do we know?
Maybe because 29 other clubs exists and when it comes down to one team vs the entire field, you take the field. The Royals also were reportedly lowballing Gordon too.
Fescoe doesn't believe these rumors though, and that's his prerogative, but they are coming from reliable enough sources. Such as ESPN reporter/analyst Jim Bowden and Royals beat writer Andy McCullough
I don't doubt it. Barring a philosophical change from KC, the expected AAV for Gordon does not fit their budget. https://t.co/IgupyZCBLH
— Andy McCullough (@McCulloughStar) December 16, 2015
Sure we don't know Gordon isn't coming back, but many signs are pointing to that he isn't.
I love how the assumption is the #royals are lowballing Gordon. IF there was such a great offer out there don't u think he would have signed
— Bob Fescoe (@bobfescoe) December 16, 2015
Maybe because the entire free agent outfield market was waiting for Jason Heyward to sign. This is how free agency works. The elite tier guys sign first as they are the teams first options and set the market price. Then the middle and lower tier guys start to sign. Cueto didn't sign before Price and Greinke. Cespedes and Upton haven't signed yet either because Heyward just signed on with the Cubs.
Gordon has undoubtedly received a handful of offers, but he's has every incentive to wait for teams to start getting more desperate for their holes. This is how free agency works.
"Looks like Omar Infante might be the opening day second baseman." Yes and he was and they were in first place when they had Omar Infante
This is similar to the strawman argument above.
There is zero argument that Infante was a good baseball player last year. Zero. This is an indefensible position along the lines that gravity doesn't exist. Omar Infante had a 44 wRC+ in 2015 (66% worse than league average). Even his average defense at 2B couldn't make up for .220/.234/.318 he hit in 455 plate appearances.
However there is another idea inside of here that Fescoe is missing. These projections aren't repetitions of what players did last year. They are projections of what players are going to do next year. Omar Infante will be coming off an injured season, one year old (now 34), where he was a disaster at the plate. I can't imagine how someone could project or predict he'll be extremely better in 2016 after knowing what happened to him in 2015.
"There are plenty of big name outfielders out there but the Royals don't figure to be players for them"...How do you know?
And now we've kind of reached the kindergarten argument of "nuh uh." The Royals aren't signing Yoenis Cespedes to a $100M+ deal, nor are they doing the same for Justin Upton. It just isn't going to happen.
However according to Fescoe's logic earlier, those guys aren't going to get that because they haven't signed that contract yet.
I'll tell you how I feel about it (the Royals not projecting well)...I don't give a rats ass about what these numbers say...
Except for we've now reached minute number eight on the subject, talking about numbers you don't care about.
The Royals have proven, that these are meaningless. The Royals have proven that stats don't tell the story.
This might be the most egregious quote of the whole segment. Fescoe is citing one example of a team outperforming their projection as proof that these projections are irellevant.
What Fescoe should have probably done is read the article a little more as he would have seen the link to the research done on these projections.
If you eyeball it, there's a clear, linear relationship. Worse projected teams have generally been worse actual teams. Good projected teams have generally been good actual teams. That right there is enough to say, yeah, there's value in what's provided. The projections aren't telling you nothing. But you'll notice, also, that there are some sizable gaps between the data points and the line. That's to be expected. Certain things are unpredictable, like injuries or trades. And we don't want to nail this 100%, not that that would even be possible, because then, what's the point?
And it goes back to what we were talking about with the chiefs having those character guys having everybody pulling in the right direction and willing to bunt. To move a runner over willing to sacrifice themselves, for the benefit of the team. That's not a tangible thing that numbers can be put on and that freaks the numbers people out...
The funny thing is...this isn't at all what lead to the Royals success in 2015. Their success came from putting the ball in play, defense, baserunning, and bullpen.
The numbers people love defense and baserunning. They (myself included) think it's a very important aspect of team success. Meanwhile there's been a ton of research done on how effective an elite bullpen can be.
They (numbers people) go ahhhhh call my mom, I'm trapped in the basement, I don't know what to do I don't have a number for this...
This is where the segment goes from at least topical, somewhat directed opinion, to classless... Fescoe is a guy who hates numbers because he doesn't understand them. He'd rather explain a team's success based on chemistry than actually fielding talented players. There's no excuse to be this baseless, reptilian brained, denialistic unless you just plain don't understand something.
This is an insult to people like Rany, Dave Cameron, Bill James, Sky Kalkman, Jeff Sullivan, Dan Szymborski, Billy Beane, Andrew Friedman, Voros McCracken, Kevin Goldstein, Keith Law, Nate Silver, Ben Lindbergh, Theo Epstein, Brian Kenney, Clay Davenport, Gabe Kapler, Brian Bannister, Tom Tango, Dan Brooks, Harry Pavlidis, Doug Thornburg, Russell Carlton, and many others. People who have pushed the boundaries of what we know about baseball in not an effort to gain success for themselves but to better the baseball community. Most who have worked in baseball front offices and have forgotten more about analyzing and understanding baseball than Fescoe has even known on the subject.
All of the above have been more successful in their fields than Fescoe and if analytical minded baseball fans are basement dwelling nerds, then anti-analytical fans are cavemen.
What's even more entertaining is that Bob thinks the Royals don't operate this way despite evidence to the contrary. The Royals the past few years have made a push to be more analytical. Some of their recent hires:
Mike Groopman - Director of Baseball Operations/Analytics; B.A. from Columbia University and once worked for Baseball Prospectus (remember they developed the Fescoe-hated PECOTA).
John Williams - Director Baseball Analytics/Player Personnel; Master's in Atmospheric Science at MIT, BS Geology & Geophysics at Yale University
Daniel Mack - Analyst - Director of Baseball Analytics/Research Science; B.S. Computer Science from Notre Dame, Masters in Computer Science from Columbia University, PhD in Computer Science at Vanderbilt.
Guy Stevens - Assistant to Baseball Analytics; B.S. Mathematics and Economics from Pomona College
The Royals have made a push to try to embrace the new analytically inclined baseball world.
And the Royals and now the Chiefs are doing it as well. They're doing these little things that don't show up in box scores, that don't show up in the standings. They don't directly contribute to wins but if you didn't lose them. They would directly contribute to losses.
So whatever this entity is that Fescoe is describing doesn't show up in box scores or standings (wins and losses). They don't contribute to wins (??) but they do contribute to losses (???). I thought whatever this was didn't show up in the standings? Aren't the standings made up of wins and losses?
Chemistry in the clubhouse was so big for this ball club. Having the right players that are willing to put the team goals ahead of their individual goals was so big for the Kansas City Royals.
Yordano Ventura got some heat for strip club antics. Yordano Ventura was in several fights this year. Alex Rios has had work ethic, chemistry, and will to win problems previously and I quote:
"He's not a winning-type player. He's a guy who just doesn't have that burning desire to win, or be great."
"He was a guy who just didn't go after it. He'd be sleeping in his locker 10 minutes before he had to go out on the field. And not just once."
"He wasn't the same player after he got that [seven-year] contract. It sure seemed like he was a lot more motivated when he was looking for a contract than he was since he got the contract."
Alcides Escobar was noted for cheating on his now ex-wife, getting her pregnant, and refusing to ever see the child. When his ex-wife sued for child support, Escobar petitioned for the settlement to take place in Alabama, a state with one of the lowest child support caps. Keith Law noted that Escobar has "some serious personal issues."
Eric Hosmer and Mike Moustakas are known for being hard to deal with from a reporter perspective and have on record been unwarrantedly rude to a clubhouse reporter.
Moore signed Jose Guillen. Moore traded for Milton Bradley. Moore signed Miguel Olivo.
Edinson Volquez, Yordano Ventura, Lorenzo Cain, and Kelvin Herrera were all suspended this year.
I don't really care about any of the things above, but Fescoe seems like he would. The Royals aren't the epitome of squeaky clubhouse chemistry guys.
I mean the the greatest moment over the last two years this organization was seeing Mike Moustakas lay down a bunt in September and going... He's got it!
THAT was the greatest moment over the last two years? A Mike Moustakas bunt in September!? Forget the World Series title, back to back AL titles, the Wild Card game, the downtown parade...A mike Moustakas bunt in September was the highlight of the Royals past two years...
So we see Fangraphs and all these people projecting for the royals, give them the double bird and say "get bent" because your numbers are meaningless as the royals have proven over the last two years.
Jaw dropping ignorance.
Listen...we know numbers aren't for Fescoe or like-minded folk. However what's even worse than ignorance to analytics is the bashing of analytics by the ignorant. Sometimes you just can't convince people, despite all the evidence saying that projection systems do much better than human predictions, that something foreign to them isn't bad.
We shouldn't expect Fescoe to understand projection systems or use them. That's his choice. However we should expect Fescoe to try to understand them and have some working competency on the subject before analyzing them or dismissing them.
I can understand if you don't have the time, will, or brainpower to understand baseball analytics. However that means you have no room to dismiss them or criticize them.
Bob has his way of doing business, and that's fine. I'll continue my way of doing business in using numbers. Fangraphs and Baseball Prospectus are very successful sites. Sites that have had many of their writers poached to join major league front offices, scout teams, and in the case of August Fagerstrom the MLB.com beat reporter for the Cleveland Indians.
However Fescoe had a choice. He could either go on the air and deliver a hot take about something he doesn't understand, want to understand, or lacks the capacity to understand, or he could let it go by. Clearly he made his choice as wants to be the defender of the anti-analytics.
What other group though is that a positive thing for though? Is there pride in being anti-science? Sabermetrics is considered a data science and is the science of baseball. To be anti-analytics is to be anti-science essentially, anti-reductionism, and ignorant to how front offices actually field their teams. I don't know if Bob has kids or not, but if he does, I hope they grow up to be bright, talent, and successful sabermetricians.