The Dord of Darien

Musings from the Mayor of the Internet


This is the best ever

I swear I’m not making this up. Over at Breitbart Sports — which is such a hilarious idea I can’t believe I haven’t written about it before — some clown called John Pudner has just written… this. Let’s not beat around the bush here: this man has just revolutionised baseball analysis.

Breitbart Sports Introduces Value Add Baseball

When the name of your metric is incoherent, you know you’re off to a good start. "Value Add Baseball?"

Breitbart Sports today introduces Value Add Baseball (see the top 100 pitchers here), a much more accurate measure than WAR for evaluating the value of starting pitchers by analyzing every start.

Apparently WAR only considers every third start. Who knew? Incidentally, I was very disappointed by the list of the 100 top pitchers; it turns out that this crazy metric doesn’t generate results so balls-out lunatic that I can really laugh at them. I mean, no, Patrick Corbin is not anything resembling the best pitcher in baseball, but he’s at least a good pitcher. I was really hoping it would pick somebody completely weird, like Kevin Correia.

A solid starting pitcher has only a 21 percent chance of getting his team a win if he needs an ERA of 1.00 to 1.99 for the game, but if his team gets him just one more run he has a 63 percent chance of winning.

I — what? I think that’s an incredibly weirdy-beardy way of saying that it’s easier to win a game if you score three runs than if you score two. Which, I mean: thank god we have Value Add Baseball to tell us this, because otherwise, how would we know?

The simple "Runs Support" models simply do not work.

Apparently they only "work" if you rephrase them in some incredibly complex fashion involving target ERAs.

While WAR (Wins Above Replacement), like Value Add Basketball, is an excellent measurement of other players on the field, the pitcher’s position is unique.

Value Add Basketball? Did he just say "Basketball?" Let’s look again.

While WAR (Wins Above Replacement), like Value Add Basketball, is an excellent measurement of other players on the field, the pitcher’s position is unique.

I harbour a deep suspicion that Value Add Basketball is unsuitable for evaluating pitchers. Never mind that. Here’s a brief list — just off the top of my head — of ways in which WAR is not like Value Add Baseketball:

  • Not created by a crazy man
  • Reflects actual value rather than crazy-man things
  • Is a real thing that actually exists and people actually use

I could keep going, but the list sort of peaks there.

The starting pitcher is the one player who has responsibility each game for getting his team the win.

All those position players? God only knows why they’re there. Probably just to keep the starting pitcher from getting lonely. And relief pitchers? Fuck them, I guess. They’re only responsible for getting their team the "save," unless they’re middle relievers, in which case extra-fuck them.

He is the most important player on the field whenever he pitches, and yet he sits out most of the games.

Who cares? I’m happy that you get paid by the column-inch, but get to the point already.

To win, he needs to stay below an "ERA Needed" in each game, which is the total of: His team’s offensive runs that day MINUS unearned runs allowed by his defense MINUS relief runs allowed, DIVIDED BY his innings pitched TIMES nine.

Fuck the hecking heck? That is the most insane thing I have ever heard. All that hopeless bullshit math amounts to absolutely literally nothing. It is an unstoppably weird way of rewriting the following formula:

(runs scored) – (runs allowed)

If it’s positive, you win. If it’s negative, you lose. Baseball! Or, alternatively, we could use this weirdo formula that — for some reason — is attempting to piece together an approximation for runs allowed by adding together a bunch of other stuff.

Of course, this is all notwithstanding the fundamental crazy conceit behind this whole "system," which is: this metric, which is touted as a definitive stat for valuing pitchers, is derived almost entirely from run support. He’s done like the exact opposite of FIP; he’s discarded practically everything the pitcher does control. Unreal.

We ran this formula on 2,157 starts this year by pitchers who were part of their team’s four man rotation, so these numbers do not include the poorest pitchers (No. 5 starters, spot starters, or fill-ins).

Why the hell not? I’m serious. What is the point of leaving out a bunch of data? Did that break your "formula" and you left it out so you wouldn’t look bad? Or are you just lazy?

For reference: WAR totals are available for all those pitchers unworthy of being Value Add Basketballers.

Here are the team’s records based on the ERA Needed they gave their pitcher.


Then there’s this chart that just proves that teams win more games when they score more runs. Which is, and I cannot sufficiently stress how incorrect this is, credited to the pitcher by this crazy person.

A pitcher is credited with a "victory" if he either gets the "Win," or if he pitches at least five innings and his team wins.

We all know how much I love pitcher wins. If I’m reading you correctly, you’re just adding wins and "wins lost," the existence of which goes a long long way toward justifying why I hate wins.

Obviously, a pitcher cannot get his team a win if the team never scores in the game, or the defense or relievers give up more runs than the team scores. Therefore, these pitchers did not win any of these 509 games in which their ERA Needed was 0.00 or less.

This is not so obvious, apparently, that Value Add Baseball doesn’t still consider it a failure of the pitcher. Seriously, if you throw nine perfect innings, allow a baserunner on a dropped third strike, and then that guy scores on three consecutive passed balls because Russell Martin was adjusting his mask and forgot to catch the ball, you’ve had a pretty damn good game. WAR will tally it accordingly. Value Add Basketball? It says you did a shit job if your offense — you know, all the guys on the team who aren’t you — didn’t score.

In 239 other games, pitchers had to throw shutout ball to get the team the victory, and they were successful 49 of 239 times, with two pitchers accomplishing it three times. Justin Masterson has guided the Indians to three 1-0 wins this year, throwing complete game shutouts when the White Sox visited April 12 and when the Yankees dropped in May 13. When the Rangers came July 27, Masterson needed a little relief help after leaving in the eighth inning of a 1-0 win. Jorge De La Rosa is the only other player to accomplish the feat three times, but those were all games in which his relievers gave up just one less run than the offense scored.

So… okay? I’m lost. What does that meandering mess have to do with anything?

These are the truly hard wins.

Ah. Your presumptive 2013 Cy Youngs: Justin Masterson (150 IP, 7.1 H/9, 0.6 HR/9, 3.3 BB/9, 9.2 K/9, 111 ERA+, 2.7 WAR) and Jorge De La Rosa (126.1 IP, 8.7 H/9, 0.5 HR/9, 3.1 BB/9, 6.1 K/9, 138 ERA+, 3.7 WAR). Why? Because they just win, baby! When it counts!

Presumptive Cy Young runners up: Jack Morris and Jack Morris.

However, when teams get good pitchers just two runs to work with – the winning percentage jumps incredibly to over 60 percent.

What? No! That is the opposite of incredible! That is very, very credible. When you score twice as many runs, you’re considerably more likely to win. I mean, look at games in which teams give their pitchers thirty runs to work with — can you believe it? They win 100% of the time! The funny thing is, according to this crazy system, either those pitchers are awesome, because they won, or they’re shit, because that win was so "easy." No way to tell!

Aside: Wes Littleton earned a save in that game. The Rangers won goddamn 30-3, but Wes Littleton still recorded a save. I’m sure we all stopped caring about fucking saves when that happened, right?

Replacement level pitchers do not fare as well of course, so the right hand column gives the chance a replacement player would have had to get the victory if given the same ERA Needed.

Wait, what level? Replacement level? You mean, the sort of thing you determine as part of WAR? Hmmmmmmmmmmmm.

The actual formula used to pinpoint this approximate curve is the square root of the ERA Needed to win up to a 4.99 ERA Needed. From 5.00 to 6.99 the formula is the square root plus one, and from a 7.00 ERA up it is the square root plus two, with a maximum Replacement Chance of 8.00.

Hahahahahahahaha what? That is blindingly arbitrary. I mean, my goodness. I can think of lots of terrible methods of determining replacement level, but that’s very very close to the worst. Why did you even bother?

Also: I call bullshit on a replacement-level pitcher having a 50% chance of winning a game in which his team scores nine runs. I think I probably have about a 50% chance of winning that game.

The flaw in simply using “Runs Support” is that three equal pitchers could all get 20 runs to work with over 10 games. A pitcher who had two runs to work with every game would likely win six games, while a pitcher who got four runs in half the games and none in the other half would likely win three games, and a pitcher who received all 20 runs in one game could only win that one game.

Well, no; the flaw with using run support — aside: has anybody ever seen it written as "Runs Support" before? — is that it has nothing to do with the pitcher at all.

Calculating Value Add

This same pattern has played out – with slight adjustments during the high scoring years – since I introduced it in the New York Post more than 20 years ago, and it continues to measure the one position player that WAR cannot.

You’ve been doing this madness for twenty years? Twenty years? The fact that absolutely fucking nobody has paid attention at all in that entire time is probably a sign, John.

Also: WAR seems to have done a fine job of measuring pitchers. Here are your 2013 WAR pitching leaders: Kershaw, Hernandez, Wainwright, Sale, Harvey. Pretty good. 2012: Verlander, Price, Kershaw, Harrison, Cueto. All-time: Young, Johnson, Clemens, Alexander, Nichols. CANNOT BE MEASURED.

Also also: pitchers are not "position players," you crazyass. Are you really sure you’ve been writing about baseball for twenty years?

To determine each pitcher’s Value Add, he gets credited one Victory for any Win, or when he pitches at least five innings and his team wins. For each game we then subtract the likelihood that a replacement player could have won with the same ERA needed (see table for basic guideline). The result of those two figures is a pitchers Raw Value Add.

Any time a pitcher fails to go five innings he is given a “Blown Game,” and the best score he can receive is a -0.6 in Raw Value.

A player’s Value for that game is then adjusted by one of two figures. First, if his ERA Needed was 0.00 so he had no chance to win, his Raw Value Add is 0.0, but his Adjusted Value Add is +2.0 – a figure that continues to be as accurate as it was when first introduced when running all starts for all pitchers.

In any other case, the pitcher is given the ballpark adjustment for where the game is played. The biggest adjustment by far is a +0.09 a player gets any time he pitches in Colorado

And this is why people complain about stat wonks. Because dipsticks like you invent these bizarro mathematical mazes for no real purpose. I mean, all you’ve done is take wins and exaggerate the effect of run support on it. Actually, you know what you’ve done? You’ve created a metric that’s like a way, way more complicated, fiddly version of "quality starts." QS is a metric I don’t love, but which has some value; a pitcher gets credited with a "quality start" if he pitches at least six innings and allows no more than three runs.

So, in conclusion, I’d just like to reprint the very first comment on the bottom of this article:

This is by far the dumbest thing I have ever read in my life. I am impressed.

July 31st, 2013 Posted by | Baseball | no comments

No Comments »

No comments yet.

Leave a comment

You must be logged in to post a comment.