Thinking Through the $38 Yankees Bullpen

Author’s Note: The article you see below was completed before Andrew Miller was injured by a comebacker on Wednesday. I believe the thought process and information below is still relevant and useful, even in light of this injury.

I love this time of year. Sure it’s fun to be doing my own draft prep. But a much underrated part of draft season is all the talk by others in the industry about their own processes. If you’re trying to learn about other strategies and improve at this game, the expert draft season is a treasure trove of goodies. A lot of the strategy and thought provoking discussions will dry up as we get further away from March.

I gobble up the draft recap articles and podcasts that come out of the LABR and Tout drafts because I’m such a junkie for the strategies and decision making processes others use.

And there were some interesting ones this year. Ron Shandler took his new BABS approach to Tout Wars. Steve Moyer attempted the Labadini plan. And there’s now a Tout Wars head-to-head draft with some interesting rules that will lead to interesting strategies (as Jeff Zimmerman wrote about on Rotographs (free!), Rotographs again (free!), and Rotowire ($$)).

But the one outcome I want to discuss wasn’t so much an overall strategy as it was a specific plan or series of decisions. Chris Liss, of Rotowire, drafted Aroldis Chapman, Andrew Miller, and Dellin Betances for a combined $38 in the AL Tout Wars draft (click here to see the entire results of all the Tout leagues).

Was this a good move? Or not?

The Projections

Before we analyze this approach, let’s take a look at the projected stats for the players involved. The projections in the table below are an average of the Steamer, Fans, and Fangraphs Depth Chart projections (I performed this aggregation using an Excel tool I created).

Now that’s interesting. For comparison, Chris Sale is projected by Steamer to have 15 W, 274 SO, a 2.78 ERA, and WHIP of 1.03. Sale went for $32 to Jason Collette.

So there’s an argument to be made that Liss “built” a better Chris Sale WITH 40+ saves and paid only $38 for him.

In fact, if I create a mythical pitcher with the total stat line from the table above (and remove Miller, Chapman, and Betances), that mythical beast comes out with a projected value of $45.

Seems pretty straight forward, right? This was a great move. $45 of value for only $38?

Not So Fast, My Friend

Said in my best Lee Corso voice… There are some logical flaws in the information I just presented to you. Do you know what they are?

If you read Chris Liss’ recap, he alludes to the biggest potential problem with his strategy:

While they’d be taking up three roster spots instead of one, it’s not a big deal because your last two pitching spots in an AL-only aren’t worth that much and can sometimes have negative value.

The fact that the strategy requires three roster spots to implement is worth noting. An interesting thing happens if you break the three-headed mythical creature into individual bullpen monsters…

Whoa! How’d we go from the $45 of value I mentioned above to just $33 and change? That’s the effect of replacement level stats being taken away. The problem with my $45 presentation earlier is that replacement level statistics were only removed ONCE. This artificially inflated the value.

For reference, my approximation for replacement level in a 12-team AL-only league looks something like this:

* Before winning 5th starter job

As you may know, a player’s value is determined by the statistics they provide above and beyond those of a freely available replacement level player.

We can quibble over who replacement level is, but it’s not going to make much of a difference here. It looks like something in the neighborhood of four wins, 65 strikeouts, one save, an ERA of about 3.4 and a WHIP of about 1.23. Outside of wins, they’re all measures that the Yankee trio easily surpass.

Oh, So You’re Saying this was an Overpay?

No, not yet. There’s something else we need to look at.

Nearly all closers in the Tout Wars auction were going at a premium. So overpaying projected values was simply the cost of doing business in the saves market. You can sit there and walk out of the auction without a closer, or you can overpay slightly, like everyone else, and adjust your values elsewhere.

Player	My Projected Value	Actual Auction Value	Difference
Wade Davis	$17.96	$23	$5.04
Craig Kimbrel	$17.79	$22	$4.21
Cody Allen	$17.55	$20	$2.45
Ken Giles	$15.42	$18	$2.58
David Robertson	$15.38	$18	$2.62
Zach Britton	$15.33	$20	$4.67
Sean Doolittle	$14.15	$11	$(3.15)
Shawn Tolleson	$13.88	$14	$0.12
Glen Perkins	$13.22	$12	$(1.22)
Aroldis Chapman	$12.53	$18	$5.47
Huston Street	$11.85	$14	$2.15
Francisco Rodriguez	$11.58	$17	$5.42
Andrew Miller	$11.47	$10	$(1.47)
Dellin Betances	$9.05	$10	$0.95

If you look at the prices Liss paid, I don’t see that he overpaid in relation to what everyone else did.

Here’s the Biggest Reason I Like the Move

You can see that I’m warming to the idea of this being a good series of moves… But there’s more.

As I talked about some in my article last week (to which Alex Chamberlain responded), I think we can benefit from accounting for reliability in player projections. Projections are weighted average probability calculations that are all mushed into one line. This example is extremely oversimplified, but Mike Trout’s home run projection might really be composed of a 5% chance of hitting 25 home runs, 50% chance of 30, 35% chance of 40, and 10% chance of 45. You run the math on those probabilities and you get a projection for about 35 homers.

With that probabilistic example in mind, look again at the prices for the closers listed above.

Something doesn’t make sense to me.

Sure Chapman is suspended. But there’s no way Perkins should have a projected value surpassing Chapman. Ratios and strikeouts aside, there’s a pretty good chance Chapman surpasses Perkins in even raw saves. Same for Tolleson and Doolittle. I can’t find a reliable article to point to, but it seems like we hear all the time how the closer turnover rate pushes 50% each year. The prices above don’t seem to reflect that possibility.

In fact, if you look at the Depth Chart or Steamer projections, guys like Perkins, J.J. Hoover, Tolleson, Doolittle, and even Fernando Rodney are all projected for about 35 saves. The risk of these guys losing their job does not appear to be factored in correctly (if there’s a 50% chance a guy like Hoover loses his job, he should be projected for 18-20 saves, in my opinion). And we’re probably lying to ourselves if we think all of Jansen, Davis, Kimbrel, Allen, Britton, Familia, and Giles escape the year unscathed. There’s still risk there.

And that’s what I like most about Liss’ strategy. Certainty. Elimination of risk.

I can’t imagine a scenario where he doesn’t have 95%-100% of the Yankees saves this season.

Errrr… Ummm… I can now imagine some scenarios. But you get the idea.

So if he overpaid on Chapman, he has guaranteed insurance policies in place were something to go wrong. Those that overpaid on Davis, Kimbrel, Allen, Giles, Britton, and K-Rod have no such policies. They’ll have to claw tooth and nail and expend vast amounts of FAAB to secure a backup closer that will likely have inferior skills.

While this insurance policy does require a total of three roster spots to execute, a good portion of the value spent is insulated from problems. Imagine if we could pull off a similar strategy to insure Mike Trout or Paul Goldschmidt…

Can’t Pull the Trigger on $38?

It’s a tough pill to swallow. And now that this strategy has been pulled off in a high profile Tout League, it may not be so easy to execute.

When I was floating the idea for this article out to some of the Rotographs staff, Brad Johnson commented that you could try the “budget version” of this plan with the Baltimore Orioles (Zach Britton, Darren O’Day, and Mychal Givens).

But This Got Me to Thinking…

After going through this thought process, I think I’m sold this was a well-executed strategy. But it has got me thinking about the value of a roster spot. This goes back to Liss’ allusion that he wasn’t thrilled about the aspect of taking up three roster spots to do this.

I don’t think we give enough thought to opportunity cost. In this situation that would be, “What else could be done with those two roster spots?”.

Having such firmly cemented assets in the pitching lineup is nice. But what does it cost you in terms of roster flexibility? It would be hard to justify pursuing another middle innings shut down reliever that might be easily had for some cheap FAAB during the season. It also becomes a little harder to pursue saves in season. Do you really want to allocate four spots in your pitching staff to relievers?

Hmmm… Interesting to think about. This is a topic for another day.