clock menu more-arrow no yes mobile

Filed under:

Snow Daze Links

With part of the country under the snow, here are some sabermetric links from the past few weeks.  Throw a log on the fire and check them out!

Graphing the MV3 " Play a Hard Nine
Steve takes a look at three Cardinal greats with a WAR plot. I was surprised at how well Edmonds and Rolen look.

Chase Utley: SB Wizard?
Neat, simple way to look at SB efficiency by Tango.  FWIW, following this approach, I have Rickey Henderson's break even point at 135 SB attempts (95 successful steals), Barry Larkin's at 58 SB attempts (41 steals), and Michael Bourn at 87 attempts (61 SB).  I'm not sure if those mean anything or not, but it's fun. :)

Advancing by ground
Nice work by Harry showing that GB%, like E%, changes steadily with minor league level.  Therefore, I think we can argue that both are at least indicators of league quality.  If we only had batted ball data going back to the 1800's...

Baseball Prospectus | Getting Out of the Zone
Colin is taking inspiration from SAFE, and perhaps PMR, in his new fielding metric for BPro.  I really like the use of continuous functions to fit players vs. typical fielders, so I'm optimistic about what he will come up with.  It also has the added benefit of coming from MLB Gameday data, which will give a nice check against UZR (which is based on Baseball Info Solutions data).

FanGraphs Splits | FanGraphs Baseball
FanGraphs has lots of splits now.  When used cautiously, this can be helpful, though in general I think the availability of splits just facilitates people making judgements based on small sample sizes.  That said, people LIKE to make judgements on small sample sizes, so who am I to stop them?  I do like that they have a "bunts" split, which is AFAIK unique.  Check out Willy Taveras!

Baseball Prospectus | Introducing SIERA
Matt Swartz & Eric Seidman roll out SIERA, a new DIPSy statistic that intends to replace QERA (see also this post, which has a more approachable discussion of the construction of their metric). They report it will outcompete FIP at prediction, and I believe them.  I wonder by how much (they're scheduled to show those data on Thursday).  Intuitively, aside from some second-order interaction terms, I do find it to be not blindingly obvious as to why it is clearly better than FIP...beyond just being fit to the thousandth decimal place. Compare that to tRA*, which uses much more data, a very different approach than FIP (estimated linear weights based on all batted ball statistics), plus appropriate regression, to create a new stat.  Then again, tRA (and especially tRA*) hasn't been subjected to the sort of performance testing they've apparently subjected SIERA to, so tRA has to be considered somewhat untested at this point.

Evaluating the 2009 Forecasts: Chone/ZiPS + Fantistics Win
Tango walks us through a nice study by J. Cross and his high school students (don't hold that against them). Great work by all involved. Punchline? CHONE and ZiPS are really good at rate predictions, while PECOTA was broken last year. Badly.