Tuesday, August 19, 2014

Where's Waldo?

There is a popular proposal to rename the Waldo Tunnel, just north of the Golden Gate Bridge, after Robin Williams, who lived nearby in Tiburon.  At first, makes some sense.  Many people, myself included, loved Robin Williams, why not name something after him?

Here's a view as you leave the tunnel southbound:



Anybody sensing a "BUT. BUT..." coming???

Robin Williams was a famous and wonderful actor who died of an apparent suicide.  The tunnel is close to the Golden Gate Bridge, which is the second most used suicide site in the world.  Any chance that the close proximity of the two will "inspire" even more suicides?

Sorry, this is a really bad idea.

Monday, August 4, 2014

Fantasy Baseball: What factors control your wins?

In a weekly head to head points fantasy baseball league, such as ours, you win by scoring more points that your opponent each week.

  • some is "skill" - picking good players, spotting matchups, and racking up points for your team.
  • some is luck - hoping your opponent has a bad week.
Here are some charts from our league after the first 17 weeks.  For the most part, every team has played every other team twice, so things should be evening out.

The first chart, "Points For vs. Wins", shows the correlation between "points for" (points you earned for your team) and wins.  The slope is 0.0036, which means that every 278 points is worth an additional win.  With a fairly weak R2 "goodness of fit" of 0.4075

There second chart, "Points Against vs. Wins", compares opponent's points vs. wins.  The R2 "goodness of fit" correlation is somewhat weaker, at 0.2991.  However, the slope is roughly twice as great, -0.0071, which means that every 141 points an opponent earns is likely to cost you a win.

In conclusion, 
  1. Your own score explains roughly 40% of your wins. 
  2. Opponent's scores explains roughly 30% of your wins.
  3. However, your opponent's points seem to hurt more against you than your points help.
At least, in our league.  Obviously, it would take a big study to make conclusions over a larger number of leagues.  Anybody looking for a math project?

See charts below: