'Moneyball'-ing the Western Conference Playoff Teams
Saw this post over on reddit.com/r/hockey (and sourced from his blog originally) where a guy tries to 'Moneyball' the Western Conference playoff series based on certain stats. Although I don't think it goes in depth enough to provide an extremely insightful analysis, it does give some idea of relative strengths and playoff experience going into the post-season. edit: to clarify, these are career playoff stats
First off, "Moneyball" itself was not really about statistical analysis as much as it was about maximizing a player's value and utilizing statistical analysis to maximize player value and find players who were undervalued by the market and who could produce the most benefits for the least cost.
Second, using statistical analysis as a way of assessing player value was not new to baseball, or any other sport for that matter. Teams had been doing it for more than a century. The difference brought to light by Sabremetrics and brought into the mainstream by "Moneyball" was the fact that teams had just been using the wrong statistics.....or at least, a less accurate set of statistics. Sabremetrics showed that OBP was a far more accurate predictor of the future and component of team success than batting avg., HR's, or RBI's.....or ERA, saves, etc.
Third, just as goals are the key to hockey success, runs are the key to baseball success. A player can score runs in baseball just like a player can score goals in hockey. But as the Sabremetric analysis showed, merely looking at a player's statistic run totals does not give a truly accurate prediction of the player's benefit nor his likely contribution to team success. In baseball, the trick was not in figuring out which player scored the most runs (a stat that has been tracked since well back into the 19th century), but in figuring out what specific component of the game most directly correlated to producing those runs for a team (and/or keeping them off the board on defense). This guy's stats are just telling you (sort of) who has previously scored the most goals in the past. It does nothing to tell you how to best generate those goals (which is where Sabremetrics struggles a bit outside of a statistically predictable game like baseball) or which players are the best at doing so.
Fourth, even in "Moneyball," there was some important focus on the defensive side. On defense, it was all about who could keep runners OFF base, regardless of how they looked while doing it. Most of the book/movie focused on the offense because that was the biggest hole on the A's roster needing to be filled at the time, whereas the starting pitching was largely well set. But the defensive side (preventing opponents from scoring) was just as important to the A's.....and I see no mention of any defensive or goaltending stats among that guy's numbers. Kind of a rather blatantly obvious omission.
Finally, even ignoring all that other stuff I posted and ignoring any of the merit of this guy's numbers......this guy didn't even do his own calculations right. 87 / 602 does not equal 6.92 goals per game. That last column doesn't show goals per playoff game for each team.....it shows playoff games per goal.
That's not "Moneyball"......that's just "Math Fail"
Last edited by SJGoalie32: 04-09-2012 at 07:00 PM.