Extended Annual: Why "Net"

This is the first in a series of pieces that expand on material mentioned in the Indians Annual. Most of this material was excluded for lack of space, and/or because it was a little more sophisticated than a broader audience might appreciate.  There is essentially the second half of the Antonetti Q&A, some more insights from Mirabelli and Grant, a humor piece, maybe even some all-time pitching seasons that didn't quite make the cut.  For today, there's Net.  In the "About the Stats" sidebar to Adam's Top 20 Prospects piece (page 87), I promised that the never-before-published stats Net and Progress Score are explained further right here at LGT, and today, I'll start keeping that promise.

Net was inspired directly by the Indians front office.  I can't remember the exact source, but it was reported at some point that when the Indians are evaluating hitting prospects, one quick-and-dirty check they do on his stats is to see whether he has more walks and extra-base hits than strikeouts.  It's an intriguing idea.  It seems simple, but it's actually a mashup of measuring a player's raw production, and the robustness or durability of that production.  It isn't three-true-outcomes — which at times is so purist as to be ridiculous in many ways — but it is rooted in the basic idea of focusing on the elements of performance that are more directly controlled by the player, and less by other players and random variance.

Well, I figured, plus-minus stats are all the rage lately, so why not make a plus-minus stat out of this concept?  And why not also do it for pitchers, with the same variables in reverse?

Net (for hitters, draft version) = BB + XBH – K

Net (for pitchers, draft version) = K – BB – XBH

The NetSB Tweak (for hitters)

But then, I thought, for the hitters, why not throw NetSB in there as well?  That would give some credit for having "impact speed," without overwhelming the basic production we're trying to capture.

NetSB = SB – 2*CS

I came up with this years ago, and then I was crushed to find someone else also using a stat called NetSB — with the exact same formula — so maybe this isn't all that novel.  NetSB springs from the idea that the break-even success rate for base-stealing attempts is 70% — if you're gaining the extra 90 feet closer to home more than 70% of the time, then you're helping your team, but if it's less than 70%, then you're hurting your team.

What that suggests is that if you steal seven bases successfully but get caught in three other attempts — this is the key part — then you haven't helped your team at all, it's just a break-even.  This is important because it's tempting to think of a guy with 32 SB and 13 CS as a guy who can help your team with his speed — "he can steal a base for you" — but in reality, his base-stealing is more or less insignificant, because he gets caught too many times, eliminating a potential run while squandering an out.  A guy with 9 SB and 2 CS may seem like much less of a threat/asset on the basepaths than the 32-13 guy, but NetSB says that the real difference between the two is exactly one base — insignificant — and to be blunt, NetSB is right about that.

The NetSB formulation above puts the break-even point at 67%, which may seem a bit low, especially considering some would say the true break-even point is closer to 75% than 70%.  I personally am very comfortable with the formula reflecting a 67% break-even rate.  I don't doubt that the "true" break-even point is between 70% and 75%, but a lot of CS occur in very specific situations where the cost is not nearly as high.  Consider, for example, the classic situation where you have a man on first, two outs, and a very good hitter at the plate.  The very good hitter gets two strikes on him, and the runner takes off on the next pitch.  If he succeeds, you've put the runner in scoring position for a very good hitter.  If he fails, then you start the next inning with a very good hitter at the plate — and with a fresh count, rather than two strikes behind.  The break-even on that type of situation is far less than 70%.  Strategic elements are not well accounted for in a 70% or 75% estimate, so it makes sense to stick with the simple formula, correlating to 67%.

Anyway, you add this in to the first-draft Net formula by just, well, adding it in:

Net (for hitters) = BB + XBH + SB – K – 2*CS

One final complication that we realized later was that we were going to be showing Net for split lines like vs-LHP and vs-RHP, and the NetSB part of the formula doesn't really make sense as splits.  I will tell you candidly, Adam and I decided just to distribute each player's NetSB total into those two split lines, roughly in the same proportion to the number of PA in each split.  A more appropriate way to split it would be to have three lines — vs-RHP, vs-LHP, and as-baserunner — but that would have been more confusing way to present the stats overall.  The way we did it, it's easy to see at a glance how much damage a hitting prospect is doing against righties as compared to lefties, and that's the important thing.  The distortion is minimal, and viewing splits for Net is undeniably interesting.

The HR Tweak (for pitchers)

Having done a small tweak for hitters, I decided to do a small tweak for pitchers, too.

Net (for pitchers) = K – BB – XBH – HR

As my Dad strenuously objected and you no doubt will notice, for pitchers we are essentially double-counting home runs.  That is, while for hitters a home runs is just another extra-base hit, for pitchers a home run allowed is scored as twice as bad as a double or triple allowed.  With hitters, you don't necessarily want to see home runs as much as you want to see extra-base hits; some would even say that a prospect who hits more home runs than doubles is in some ways suspect.  Younger prospects, including high-ceiling guys like Brantley (and formerly Sizemore) haven't really developed their full power yet.  What we can say with some confidence is that a prospect's home runs should not be considered twice as important as his doubles.

But it's different for pitchers.  When a hitting prospect hits one out, he's often capitalizing on a mistake, and since he'll see fewer and  mistake pitches as he climbs the ladder to the majors, his ability to hammer mistake pitches a country mile (rather than just hitting a solid line drive) is not that important.  But when a pitching prospect gives up that same homer on a mistake pitch, we want to capture that, as it says something about his ability to get to the majors.  All extra-base hits allowed indicate a pitcher's inability to dominate the hitter, but home runs are also an indication of outright meatballs-served.  To be sure, we are slightly over-penalizing the pitchers and under-crediting hitters when it comes to home runs, but this is the nature of a deliberately simple formula.  We could introduce fractional factors, but at that point it becomes a very different creature.

This seems like a good spot to emphasize that Net is intended to be a useful way to look at prospects.  I have no real interest in looking at it for major leaguers, although you could make the case that it would be worthwhile hitters under 25 or 26.  As presently constructed, it's about emphasizing the peripherals and secondary skills that will help the player eventually succeed against major league competition.  The basic formula is about that, and the two tweaks are about that, too.

Ultimately, for all we could talk about whether to add in HR for pitchers and NetSB for hitters, Net remains very much about the basic idea of comparing K totals with BB + XBH totals.  Note that for each of the two tweaks, (a) it doesn't make much of an impact on the final number in the great majority of cases, and (b) in the few cases where it does make a big difference, it's in cases where you'd want to take note of an extreme situation.  Such as, a player stealing 40 bases and only getting caught five times.  Or a player getting caught 10 times in 18 attempts.  Or a pitcher giving up large numbers of home runs.  For those same reasons, it might make sense to throw GIDP and HBP into the mix as well, both for pitchers and for hitters -- neither one would tend to make much of a difference, and when they did, you'd want to know about it.

The Meaning Of Net

With any new stat, there's a period of discovery.  What do the numbers look like, what's good, what's bad, what's typical?  What is it doing, what is it not doing?  What does it mean?  Let's take a look at how our best prospects who played in 2008 fared.

Let's focus first on what Net is not doing, to keep the faint-hearted among us from freaking out about certain players:

• It's not aware of differences between levels, or whether a player was age-appropriate for his level.  Santana and Rivero were at the same level, but Rivero is two critical years younger.
  
• It's not aware of positional differences; Rodriguez is a middle infielder, Brantley is a corner outfielder.
• It's not giving much if any credit for contact hitting ability, even though batting average in many ways more significant for prospects than for major leaguers.  That is, younger prospects with prodigious contact hitting abilities often develop significantly more patience and power as they develop and reach the majors, but Net would see those hitters as simply not having much of a story to tell in terms of secondary numbers.

A few guys jump out right away, especially the second guys on both lists.  For those who glance at Brantley's numbers and couldn't see what the big deal was about this guy ... well, this is the big deal, along with the advanced age/level status.  He combines great footspeed with great bat control and a very smart approach both at the plate and on the basepaths, and Net really shows off those talents.  They can't strike him out, and they can't throw him out.  He's good at getting on base and likely to get even better.  If his power develops -- a big if -- he could be a game-changing player, but he looks formidable already, and Net shows why.

Josh Tomlin — absent from anybody's Top 20 list — jumps out even more.  Yes, he was old for his league, turning 24 just after ending the season in Kinston.  Yes, he's only a reliever.  But the other way to look at that is, he racked up a higher Net than almost anyone else in the organization, and he did it in only 110 innings. Then again, Scott Lewis racked up almost as high of a total in only 97 minor league innings.  (Wait a sec, do we need to come up with Net Rate?)

Net also boils down to a single number each of the two best seasons any of our prospects had in 2008, those by Santana and Huff, who both produced numbers that were both superficially impressive and fundamentally solid under the hood as well.  I'd grown tired of noting Santana's obscene RBI totals, and it's nice to have a stat with some real meat on it to look at — his +71 is twice as large as Kouzmanoff's breakout 2006, or Victor's breakout seasons in 2001 and 2002, all of which were more fueled by batting average than Santana's 2008 numbers were. What kind of hitters those two players turned out to be in the majors has been dictated largely by whether they could continue to hit for high average, but Santana's stunning +71 suggests that he has more than one avenue to developing into a major-league threat at the plate.

As a final note, of the top six hitters ranked above by their 2008 Net total, the Indians acquired five of them in the past nine months.  Anybody think that's a coincidence?