Stats, Starters, Scouts, And Scales

I don't know if you've heard the news, but Neftali Feliz isn't going to be in the Rangers' starting rotation this season.

Never fear: I'll bring you up to speed. The overall consensus from writers and analysts both local and national (pace a selected few, perhaps most notably Fangraphs' Dave Cameron) seems to be that the Rangers are making a big mistake – that the club is overvaluing closers relative to starters; setting back Feliz's development unnecessarily; giving Matt Harrison an undeserved third (or fourth? or fifth?) chance; and generally making a farce out of pitching evaluation.

Those of you who've bothered to read the bulk of my BBTiA writing are probably expecting me to bring in a bevy of stats at this point. It's tempting, I admit. Feliz's 2011 statistical projections– calculated by Dan Szymborski, and dubbed ZiPS  – came up in an exchange just the other day, and they're pretty striking. According to Szymborksi's system, if Feliz started in 2011, he'd project to win 12 games (against six losses) and boast a 3.43 ERA – with an 82 percent chance of posting a 100+ ERA+^*, and a 1-in-3 chance of a 130+ ERA+.

^*ERA+ is ERA adjusted for park and league. In other words, ERA+ facilitate fair comparisons between a guy starting in the pitcher-friendly league (the NL, thanks in part to pitchers batting) and in a notoriously pitcher-friendly home park like the Padres' Petco, to boot, with  a guy starting in the AL and suffering the slings and arrows of the outrageous Fenway. In the event: ERA+ is scaled so that 100 is average. You can find ERA+ at Baseball Reference, which is a great first stop for all baseball stat needs.

Get this: that predicted 33 percent chance of an ERA+ above 130 outstrips the projections for C.J. Wilson (31 percent).

Now, before anyone gets fired up to post a comment decrying saberism (or to send Dan an angry email), it's important to read his accompanying caveat:

"When I looked at it a couple of years ago, projections for players that had their roles converted had the standard error of their weighted mean projections increase by about 8%. Worse, certainly, but pitchers are damn hard to predict anyway."

For our purposes here, it doesn't really matter if you know what a standard error or a weighted mean is (though they're well worth reading about). The take-home is that not only are starting pitchers' performances difficult to predict, even for the sharpest projection systems, but adding a starter converting from the closer's role makes it an even more difficult task.

This doesn't mean the projects aren't worth calculating. It also doesn't mean they're not worth serious attention. They are, and they are. It's just that no one worth listening to – least of all Dan Szymborski – is going to tell you that the projections should be thought of as unerring, or anything approaching it.

And in fact, I'd argue that when you're looking at a case like Feliz – a young pitcher who has zero track record of starting in the majors; only 53 minor-league starts; a reported need to work on his secondary offerings, one of which is a brand-new cut fastball-ish pitch; and a seemingly mutable dedication to starting in 2011 – that you'd want to refer to the projections with even more caution than usual.

Where else can we turn for information, though? I'd suggest two sources – one fairly transparent, and the other much more tricky to interpret.

The first is a set of resources we're hugely fortunate to have as Rangers fans: guys like Jason Cole and Jason Parks, who not only get paid to watch Rangers' prospects and write about them, but also have the knowledge, skills, and training to do both convincingly. They really doesn't get as much appreciation as they deserve.^*

I'm aware that this may come off as shilling for their websites; they're behind pay-walls, and I know that rankles some readers. But the fact of the matter is that their observations of Feliz – and especially his slurve, slutter, and change – are the closest most of us get to the sort of information that the Rangers get from their own professional scouts.

^*Seriously. For those of you who are new to BBTiA, or who missed them the first time: go back and read Parks' "AOFP" scouting series, and try to find that sort of free, high-quality info for a single other major-league team. I can save you some time on the back end: you won't.

This doesn't mean that I automatically disqualify the firsthand observations of anyone who's not a pro scout, or the de facto equivalent. But when both Coles and El Magico agree^* that Feliz's secondary offerings still need significant improvement, coming out of this spring? That's a major red flag.

^*And no, I'm not going to post their takes here. Go pay for ‘em and read ‘em yourselves, ya cheap crumbs, ya. You can thank me later.

All of which is to say that for all of the advances in advanced stats, scouts are still hugely important. And that leads me to the second resource: the Rangers' decision itself.

Granted, the comments from Ron Washington and Nolan Ryan – focusing, as they did, on the importance of the closer, and seeming to give short shrift to the relative value of a starter – might seem more cause for concern than contentment. I'll admit: I'm more concerned than I'd like. The portrait painted by the media has either shown a united front office, or Jon Daniels painted into a corner – steadfastly advocating for Feliz as a starter, only to be convinced otherwise (or, possibly, simply overruled) by Ryan, Washington, and/or Mike Maddux.

Obviously, much of this is pure speculation. We don't know what happened behind the front office's closed doors, and we've received little definitive information from local reporters.  And so I'm torn. Part of me is dismayed that the Rangers' public pronouncements about the decision seem to fly in the face of saberist logic: the front office seems in some ways to agree with the optimistic statistical projections, but to have decided that despite Feliz's potential for starting success, the team would benefit more from his keeping the closer role.

That said, I'm not entirely convinced that what Washington and Ryan have said publicly reflects the full decision-making process of the front office. It's not that I have any driving reason to ignore or doubt what they've said – it's just that I can't imagine there's any benefit to them admitting that the front office isn't quite as confident in Feliz's starting abilities as they'd like. I also think that it's plausible that the Rangers have come to exactly the conclusion(s) they say they have, using a happy marriage of scouting and stats.

To illustrate this, at least in part, follow me into an admittedly whimsical projection fantasy. Let's say that the Rangers, knowing what they know about Feliz's readiness to start (in terms of stuff and physical and mental preparation and all) have brought together their scouting types and their number-crunchers, and come up with the following five year projection. Feliz's Wins Above Replacement (WAR) as a 2011 reliever are in the left column of each year; the forecasts if he were to have been a 2011 starter are on the right:

The probability-weighted WAR total for Feliz as a 2011 reliever over those five years is 13.9 (an annual average of 2.78 WAR); for Feliz as a 2011 starter, it's 16.35 (an annual average of 3.27 WAR). That's a difference of 2.45 WAR – which, at $5 million per win, represents $12.25 million of foregone production.

But in 2011, Feliz would be worth a probability-weighted 1.60 WAR as a reliever, as opposed to 1.55 WAR as a starter.

I admittedly pulled these numbers out of thin air, and to make a point. They don't even tell the relative story of Feliz versus his substitute as a starter (Matt Harrison, ostensibly) and reliever (most likely Alexi Ogando) – which, of course, is a crucial consideration. I'm not going to go to that deep, here (especially as it's late, and I'm banking on generous readers).

But I'll leave you with this thought: if whatever the Rangers might be passing up in terms of production in 2011 bothers you far less than the setback to Feliz's overall development – his lost WAR, if you'd like – then you should consider what larger issue this might point to in the Rangers' front office and farm system. You should also remember what you were thinking when Feliz was promoted in 2009, and designated the full-time set-up man (and, quickly, the closer) a year or so ago.

Certainly, there was grousing from some quarters about the risks of Feliz becoming entrenched in the bullpen, and the cost to his development. But be honest with yourself: in 2009, when you watched Feliz explode onto the major-league scene, were you really asking yourself whether this was hurting his long-term value? Were you prepared to see Feliz go back to Oklahoma City in 2010, to work on his secondaries in hopes of a brilliant starting campaign this season?

I mentioned this the last time I wrote about the Feliz-as-2011-closer scenario, but I think it's worth repeating: if you think keeping the young gun a reliever this season is a huge, moronic mistake largely because of what it means for his future development, don't you have to seriously question the intelligence behind the way the Rangers have handled him over the last 30 months as a whole?

It's not an entirely unfair question, but I'm not quite ready to press it. Given the Rangers' apparent success in merging stats and scouts – to the extent that at least one saber-type has labeled them "boring," at least with regard to statistical analysis – I'm inclined to give them the benefit of the doubt. It may not be as satisfying as ripping the front office a new one, but who knows? Maybe it'll make you feel more Feliz, anyhow.