BASEBALL

Bill James has written about the Log 5 method (The Log 5 method Wikipedia page gives a reference) for computing the chance of a team with a winning percentage of p playing a team with a winning percentage of q.  The formula gives the chance of a team winning as p(1-q)/(p(1-q)+q(1-p)).  I used the MLB standings before play on August 24 to calculate the chance the Tigers or Indians would win each remaining game.  Summing these gives the expected number of wins each team will get for the remainder of the season.  Detroit is expected to win 20 or 21 more for a total of 95 or 96 wins.  Cleveland is expected to win 19 more games and finish with 88 wins.  This agrees well with the Baseball Prospectus results from randomly playing out the rest of the season 10,000 times.  (On average the Tigers win 96 and the Indians win  87 using that method.)

Of course this has flaws.  You should perhaps use the expected number of wins a team should have using one version of Pythagorean wins instead of actual wins.  Day to day the pitching match up matters more than the teams records.  The Tigers for instance will face Harvey today in New York and the Mets probably win more than 46% of the time behind Harvey.  It doesn’t account for a major injury or trade that could change these win percentages.  However, this should reassure Tigers fans who are still worried even though Detroit has a 6 game lead.  The lead should grow despite Cleveland’s slightly easier schedule.