Tangotiger Blog

The AL Cy Young was the more confusing one.  Here’s how it looks if we go with 75% FIP-enhanced, and 25% Classic:

Pred Cy Pitcher
1 1 Gerrit Cole
2 2 Sonny Gray
3 3 Kevin Gausman
4 4 Kyle Bradish

5 6 Zach Eflin
6 5 Luis Castillo

7 8 George Kirby
8 9 Framber Valdez
9 7 Pablo López

As you can see, we get the top 4.

Now, here’s how it looks if we go 50/50:
Pred Cy Pitcher
1 1 Gerrit Cole
2 3 Kevin Gausman
3 2 Sonny Gray

4 4 Kyle Bradish
5 5 Luis Castillo
6 6 Zach Eflin

7 8 George Kirby
8 9 Framber Valdez
9 7 Pablo López

We get Gausman/Gray flipped, but we get 4 thru 6.

And if we go 75/25 toward Classic over FIP:
Pred Cy Pitcher
1 1 Gerrit Cole
2 3 Kevin Gausman
3 2 Sonny Gray

4 4 Kyle Bradish
5 5 Luis Castillo
6 6 Zach Eflin

7 9 Framber Valdez
8 8 George Kirby
9 7 Pablo López

This is the worst of the three configurations, with Valdez going too high.

So, we’re at the point where the higher you are in the rankings, the more it’s FIP weighted, and the down-ballot votes are more Classic.

https://twitter.com/tangotiger/status/1724942770844885120

The NL is more straightforward, other than Strider: 100% FIP-enhanced

https://twitter.com/tangotiger/status/1724932999836205167

You can go to 80 / 20 for FIP over Classic and you get this.

Pred Cy
1 1 Blake Snell
2 4 Spencer Strider
3 2 Logan Webb
4 3 Zac Gallen
5 5 Justin Steele
6 6 Zack Wheeler
7 7 Kodai Senga
8 8 Corbin Burnes

That’s it though.  Give any less weight to FIP, and Gallen goes ahead of Webb.

So, the NL is firmly on the FIP side, though Strider is a huge miss. Since Strider is ahead of Webb in BOTH the FIP and Classic predictor, this points to a structural issue with the predictor.

Since Strider is a K machine, this points to the likelihood that the strikeout is simply overweighted.  Or that maybe Walks should be considered.  But that would really hurt Snell, so you can’t give the walks too much weight.

Definitely need to look at the structural issue here.

Now, let’s look at individual voters.  Starting with the NL, the focus is on Senga and Webb.  Both had identical scores using the Classic Predictor, while Webb was way ahead using the FIP-enhanced.

Senga got 7 votes. Four put him ahead of Webb.

Webb got 25 votes. 21 put him ahead of Senga.

So, the Webb voters are overwhelmingly FIP-based, while Senga voters are Classic voters.

Here it’s pretty clear.

***

How about Gallen v Webb?  Gallen is ahead by 2.6 points on the Classic while Webb is 0.7 ahead on FIP-enhanced.

Webb is ahead on 21 ballots, while Gallen is ahead on 9 ballots.

So, again, it’s the FIP-enhanced that leads.

Now Strider.  Strider is the FIP-vote to end all FIP votes.  But, he did not do so hot in the voting, certainly not compared to Webb and Gallen.

In a head-to-head of Strider v Gallen, Strider was ahead by 15 while Gallen was 14.

So here, we end up with a split-vote.  If we relied just on these two pitchers, we’d therefore think that both versions had a strong role.

How about Wheeler v Senga?  Wheeler was behind Senga by 7 points in Classic, and ahead by 6 points in FIP-enhanced.

Wheeler appeared on 15 ballots.  He was ahead of Senga on 14 of them.

Senga appeared on 7 ballots.  He was ahead of Wheeler on all of them (on six of them Wheeler was not even listed).

Again, FIP-enhanced is preferred.

***

So, I think we can say that most voters prefer FIP-enhanced to Classic, with some head-scratching on Strider.

And therefore, we are now are on a march toward that paradigm shift.  FIP is in.  Maybe not by 2024, but by 2025 or 2026, we may end up ditching the Classic predictor.

Well… let’s turn our attention to the AL before I make that pronouncement…

Castillo was ahead of Eflin in the Classic Predictor by 7 points, while Eflin was ahead in FIP-enhanced by 7 points.  So, we have a clear choice here.  How did the voters go?

Castillo was preferred by 12 voters, while Eflin was preferred by 8 voters.

So, this is the opposite of what we found with Wheeler v Senga!

***

How about Bradish v Eflin (clear Classic pick) and Bradish v Castillo (clear FIP pick)?

Bradish is 17-7 over Castillo, while he was 17-5 over Eflin.

So again, it’s basically split evenly.

And a good sign is that only one voter had all three on their ballot, again showing a clear pattern that they don’t consider both at the same time.  On a 5-player ballot, with six viable candidates, there’s no room for confusion.

***

So, the NL ballot makes it seem as if FIP is holding great sway (and that was the case in 2021 with Burnes and 2022 as well), things are a bit more murky in the AL, with both having pretty equal impact.

For Gausman v Gray, the point difference was 1 or 2 points either way: Gausman on the Classic side, Gray on the FIP side.

However, Gray was preferred at a 21-9 clip.  For two very close candidates under either method, that Gray won so handedly suggests that the point system is not really being handled the best.

Gray went 8-8, while Gausman was 12-9.  It’s possible that the number of Wins, which is already being given little weight, is bearing too much weight still!  They had only 1 inning difference.  Gray was ahead in both ERA easily and by FIP slightly.

Gausman being way ahead in K again probably also carried too much weight, which is something we saw with Strider.

***

I think at this time, given what we’ve seen with 2021-2023, that it’s time to come up with a unified Cy Young predictor.  One that both gives more weight to FIP while at the same time underweighting K and wins.

If you want to have some fun, go thru each voter, one at a time, and see if you can create a voting system to reflect their pick.

The first one I tried had these picks:
Cole
Gausman
Bradish
Kirby
Bassitt

So, these pitchers did NOT make it:
Gray
Castillo
Eflin
Valdez
López

The ONLY reason not to vote for Sonny Gray is his 8 wins.  So, this voter gives a good amount of weight to wins.

Castillo has a high FIP, so the voter gives a good amount of weight to FIP.

Eflin led in wins and was 3rd among the gang of 10 in FIP.  But his IP was bottom 2, so a heavy weight is giving to low IP. Though Bradish was last among the 10 in IP. Eflin did have a high ERA among the group.

Valdez, Lopez are not strong candidates, so it makes sense they do not get listed. But, how does Kirby get in there? Very low walk rate is what helps him.

And Bassitt?  Led in wins and 2nd in IP.

So, if I had to create a model for this voter: this voter likely players who are among the leaders in the key categories, and the voter like to discard players who are trailers in those categories.

So, not so much try to balance out everything, just alot of: add/drop based on key categories.  This kind of voter is very difficult on trying to create a model.