tag:blogger.com,1999:blog-18284688.post2994230914790864342..comments2013-09-29T22:34:44.486-04:00Comments on Bluefish Canoe: Baseball Games Not Independent: Streaks Maybe Real!Bluefish Canoenoreply@blogger.comBlogger3125tag:blogger.com,1999:blog-18284688.post-19585184320107548722009-07-07T14:38:41.023-04:002009-07-07T14:38:41.023-04:00"First, it makes total sense for the underlyi..."First, it makes total sense for the underlying construct (say, a player's ability to hit a baseball) to change over time, including over very short intervals."<br /><br />This is often noted by the anti-streak crowd as something you and I and the rest of the amateurs see as self-evident but may not be true among *professional* athletes who have spent years and years perfecting consistency in their approaches/motions. If a robot were tasked to hit a baseball I imagine it would not be streaky, and a robotic style is arguably the goal of most hitters/throwers/golfers/shooters/etc.dzusahttp://www.blogger.com/profile/12809056952371737262noreply@blogger.comtag:blogger.com,1999:blog-18284688.post-71936249271437600292009-07-06T08:33:37.860-04:002009-07-06T08:33:37.860-04:00Non-independent outcomes?
Obviously this is a case...Non-independent outcomes?<br />Obviously this is a case for Francesca!<br /><br />The first thing she will do is require you to focus....Chttp://www.blogger.com/profile/13261809871004788294noreply@blogger.comtag:blogger.com,1999:blog-18284688.post-81986324239394906262009-07-05T20:53:48.156-04:002009-07-05T20:53:48.156-04:00It looks like it's "McCotter". As s...It looks like it's "McCotter". As somebody who knows a little something about statistics and next to nothing about sports, I always thought streaks were (at least sometimes) real. First, it makes total sense for the underlying construct (say, a player's ability to hit a baseball) to change over time, including over very short intervals. Why wouldn't it be affected by things like mood, fatigue, adrenaline, etc.?<br /><br />Second, it's far too easy for statisticians to confuse a model for reality. To say "a model assuming independence of outcomes fits the data" isn't the same as "the outcomes are independent."<br /><br />Third, although a stochastic model might work well for modeling a complex process, arguably there is no randomness in many sports processes. Take (switching sports) free throws. All parameters - the weight and size of the ball, the distance to the basket, the angle and amount of force needed to propel the ball - are (intuitively) knowable. Delivering the proper amount and angle of force is difficult, but it isn't random. You could argue that ability to do so _is_ random; but the claim that the probability of doing so is constant is a much stronger, harder-to-justify claim.Anonymousnoreply@blogger.com