tag:blogger.com,1999:blog-2893512317902577458.post8809118713174475183..comments2024-01-06T02:22:33.000-05:00Comments on Camden Depot: Just Some Graphs: Run Scoring and EarningJon Shepherdhttp://www.blogger.com/profile/03521809778977098687noreply@blogger.comBlogger1125tag:blogger.com,1999:blog-2893512317902577458.post-60350258941759896352012-08-20T11:26:07.806-04:002012-08-20T11:26:07.806-04:00I think this problem requires a paradigm shift.
...I think this problem requires a paradigm shift. <br /><br />Run Differential shouldn't be linear. Basically, if I score 11 runs instead of 12 runs, then my chances of winning will drop by 3%. If I score 3 runs instead of 4 runs, then my chances drop about 13%. And likewise, if I allow 10 runs instead of 11, I should still expect to lose. If I allow 2 runs instead of 3 runs, my chances of winning have increased significantly. <br /><br />Run differential is a quick and dirty way to determine whether a team is overachieving. If you want depth, then what I propose to do is determine the Os winning percentage for runs scored and runs allowed and then compare that to the league average(I had to use the 2011 average because getting the 2012 league average is time consuming and difficult).<br /><br />I'm not sure if the above is clear, so basically you would find each game where the Os scored two runs(and then four, five, zero, one, etc) and determine their win-loss record(8-10) in those games. Then you'd find each game where the Os allowed two runs and determine their win-loss record in those games (15-2). Then you'd compare those numbers to the league average. That would allow you to determine whether the Os have been lucky or average.<br /><br />I did this and discovered that your hypothesis is on the right track. When this team scores one run, it wins at the 2011 league average. When they score six or more runs, then win at the 2011 league average. When they score between two and five runs, they win many more games than league average. Given the distribution of how many runs they've scored in each game, I'd expect 61 wins instead of 66.<br /><br />Likewise, when this team allows more than 6 runs they win at league average (2-25). When this team allows four or more runs, they win more often than league average by a considerable but reasonable amount. They'd be expected to win 15 games out of 41 and instead they've won 17. This team has excelled when allowing between one and three runs. In those 47 games, the team has won 41 and should be expected to win 35.5 ~ 36. Given the distribution of how many runs they've allowed in each game, I'd expect 58.5 wins instead of 66. <br /><br />Just to use an glaring example, we're 8-10 when scoring two runs and we're 15-2 when allowing two runs. Instead of winning nearly two thirds of those games, you'd expect us to win nearly a half. <br /><br />Unfortunately, there's no way to post my dataset as a post, so you'd have to gather the data yourself to determine the accuracy of my work. But if I'm correct, I think I've shown that the reason why this team has won so many games is because they dominate whenever they allow three runs or fewer. If this is the case, I don't really have a different hypothesis for why this is the case. You'd expect us to be on the wrong end of a 1-0 game so far given the amount of opportunities. After all, we've been shut out more often than we've shut another team out and yet we've won one of those games but haven't lost one yet. <br /><br />I don't really have an answer for this. But I think this at least uses data to make the question more specific.Matt Pnoreply@blogger.com