Here is part 3 of the 4-part interview with Derek Carty from The Hardball Times:
Fake Teams: When evaluating hitters before a draft, I like to look at their previous year's stats and decide if the hitter can improve upon those stats or not. How do you evaluate hitters when preparing for a draft?
Derek Carty: My approach is a bit different and comprehensive, but here's the general gist of it. I look at a lot of different things for hitters. For power, I've created a statistic using HitTracker data which I call True Home Runs (or tHR). What this aims to do is take some of the luck out of home runs. You'll often hear people say something like "a home run is a home run", but that couldn't be further from the truth. A guy who pounds a 500-foot home run is much more likely to hit more home runs than a guy who squeaks one over a 320 foot fence, ceteris paribas. He's stronger, and if he can hit the long ones, he's only guaranteed to hit some short ones. tHR first neutralizes park and weather (obvious influencers ) and also accounts for the distance of home runs (not so obvious but very important influencer). The logic behind doing this might not be readily apparent, so you can read more in-depth explanations here and here (the articles are older and from the preliminary stages of development, but the logic is still relevant).
For batting average, I'll focus on the three things that it is comprised of: how often the batter puts the ball in play (strikeout rate), how often those balls clear the fence (home run rate), and how often the remaining balls in play fall for hits (BABIP). Strikeout rate is a very stable stat, so not much needs to be altered with it. For the HR component I'll use tHR and for BABIP I'll use xBABIP. This is not the xBABIP that some refer to, calculated as LD%+.120. I've found this to be worse than actual BABIP in predictive ability and should always be ignored for these purposes. Instead, I like using the xBABIP created by Chris Dutton and Peter Bendix, which I've found to be very predictive (relatively speaking). In fact, I'm actually working with Dutton now on further improving the formula. That work should be published later in the off-season.
As I mentioned earlier, I'm also currently working on a form of CAPS for hitters, which I suppose will be called CABS (Context Adjusted Batting Statistics), so that'll be thrown into my evaluations. Eventually, it'll all boil down to my projection system spitting out a projection for every hitter, which will encompass all of the concepts I've described above and a few more. For some hitters, I'll probably supplement these projections with my scouting perspective as with the pitchers.
More after the jump:
Fake Teams: I have won a few NL-only 5x5 leagues by drafting just middle relievers and closers and focusing on drafting hitters early. Other than your LABR draft strategy, what other draft strategies have you used, and how successful were they?
Derek Carty: I believe it's crucial to create your strategies based on the specifics of the league: the settings, the opponents, and what you know about your own tendencies within the context of these things. In some leagues, I'll refuse to draft closers until the end. In others, I'll take a pair of them early. In some, I'll embrace risky players. In others, I'll play it safe for the first half of the draft or longer. There are some general rules for these things (i.e. the shallower the league, the easier it is to take risks) but every league truly is different and must be evaluated individually.
As one example from this season, I played in an expert league run by KFFL with some heavy competition (partnered up with fellow-THT writer Eriq Gardner, though he replaced Victor Wang after the draft, who left THTF to do some work for MLB clubs). It was a 12-team mixed Yahoo! League. We focused on hitters early, taking mostly guys with good health records and stable skill sets with just one pitcher in the single-digit rounds, then grabbed several SPs in the early teens (including Javy Vazquez - in 12-team mixed leagues, I've found that there's a treasure trove of good pitchers to be had in this area). After that we went for some high-upside guys (like Matt Wieters and Nyjer Morgan) or undervalued guys (like Cody Ross and J.D. Drew), and eventually took a couple of closer candidates towards the end.
One smaller strategy we really looked to employ focused on deliberately getting a few injury risk players. I feel that the market does not value these players properly, leaving excess value ripe for the picking. I explained last year that if you combine the stats of a high-skill, injury-prone player with those of a replacement level player (when he's injured), you end up with a stat line well-worth the reduced cost of the so-called "injury risk" player. This is a strategy that will only work in shallower leagues like this one, though, because the static production of the replacement level players is much higher than it will be in deep leagues. In an AL/NL-only league, this would fail miserably because replacement level is someone like Juan Castro. In KFFL, we looked to use it with Chase Utley (whose status was unknown at draft time), Chipper Jones, Rich Harden, Matt Wieters, J.D. Drew, and a couple others. We were also very active on the waiver wire (over 200 transactions), keeping several rotating bench spots and streaming extra hitters on Mondays and Thursdays and elite relievers on the other days. We didn't start this until June, though, which likely cost us the league. We ultimately finished 3rd, though it was very close and we were in 1st just a few days before the end of the season.