Sunday, January 22, 2006

Something very hard to prove one way or another

I read on a forum that in the Conference Finals in the NFL, the team that won SU has covered 58 out of the last 60 games...with 2 games as ties. What do you make of that? Here are some initial thoughts:

- wow, that's a real high winning rate. Even if the middle (fave winning, but dog covering the pointspread) should occur about 10% of the time, there is a very small chance that none has covered in the last 60 games. That's 30 years worth of conference finals.

- This is Super Bowl XL (40th Super Bowl) coming up. So there has been 78 total Conference Final games. How come that stat only goes to 60 games? My guess is the first 18 Conference Final games would make the record look less impressive. So whoever compiled that stat stopped right there. If the record had only been 30-0 in the last 30 games, but say 20-10 in the previous 30, I'm sure the person who had the data and reported it would have used 30-0 in their last 30 as the important "oh my gosh" stat instead.

- The last Super Bowls were middles. NE won both games but didn't cover either. Is that relevant? I sure think so.

- What about other playoff games? What makes the conference Finals special compared to the Divisional round? I can't logically think of anything....but I do know that whoever reported that 58-0-2 stat didn't report other stats with the other that to me is another sign of data-mining.

My opinion on this stat is that it is another case of an amazing stat that can easily fool many people. When a team is a 3.5 point favorite, they should win but not cover about 13% of the time. That is about the percentage over all 3.5 pointspread games in NFL history. I expect that number to stay very close to the real number for both the SEA/CAR and DEN/PIT game. But is there any way to prove that? No there isn't. Because the argument on both sides is all looking at historical numbers. There is no is simply a judgement call based on historical information. Sure, I think I can make a better judgement call based on historical information than most, and I think whoever looks at this historical info and says the middle is much less than 13% (say they argue it is only 5%) is wrong...but I can't prove they are wrong. They can't prove I'm wrong either. The great thing about sports betting though, is that there is no reason to try to prove to each other that you are right. Just BET IT! The money won/lost will be the reward or punishment for your opinion.

No comments: