The other day, Corey Lidle when asked about Barry Bonds' home run milestones said, "I don't think it's legitimate". He added, "It's sad. I'm not a player-hater. I like to see players get paid as much as they can. But without friggin' cheating."
Of course, Lidle won't be facing the Giants in the current series (in which the Phils have already taken two games). So it's a really gutsy move. Besides, sadly, Corey Lidle's stats are "legitimate".
When I pointed this out, my friend Chris wrote back:
I noticed Lidle's walked only three hitters this year.
Let me ask you if you have any thoughts on this: how many games do you think two teams would have to play against one another, head-to-head, before you could determine with a high degree of certainty which was one was better?
Actually,Lidle doubled his season total in his last start and now has 6. He's also striking out about a man an inning. He's making his case to stay in the rotation after Cole Hamels gets promoted sometime next month, which is bad. If Lidle stinks up the place, he gets replaced. If he doesn't, they move Madson back to the bullpen (cutting Leiber, since he's supposedly #1, would be too imaginative for this team).
Re. the number of games to prove superiority, well, it depends. First, let's assume that the outcome is 50-50. That is, that either team has an equal chance of winning. I know it ignores homefield advantage but it simplifies things a bit. It also depends on the confidence interval we're talking about.
At 80% confidence, you'd need about 10 games (9.615 to be exact).
At 90% confidence, you'd need about 68 games (67.240).
At 99%, it's exactly 16,641.
So given that, there have only been two postseason series that have declared a winner with at least 80% confidence, and both were in the 1880s, the 1887 & '88 NL-AA Championship series.