Now he has to sit for a few days, and then he might need to take his talents to Canada to make it 7 in a row. Who can handicap all that? The estimates will swing from .05 to .2. Just as long as you never say never, you’ll never be wrong (but you migt place some bad bets).
Incidentally, the Michigan-Michigan State game Saturday ended with a last second touchdown off a muffed punt. An in-game-analytics tool had the MSU prob of winning at 0.2% with 10 seconds left, something that broadcasters kept repeating over and over in amazement. No one questioned how good an estimate that was. (I think that play, in that situation, results in a touchdown more than 2 in every 1000 times, but yes of course it was a low chance event.) It was a good example of lay-statistics (as well as black swans) because the 0.2% estimate was treated as an absolute fact. Somehow p-hat played a big role on sports shows that night to convey just how dramatic the game had been. Maybe we need to dial down the p < .05 threshold?
But seriously, your post holds up very well the next day. Even if he hadn’t hit the home run in the 8th, he hit a deep fly earlier that, while not a home run, certainly justified an estimate of something like 20% even though looking just at Murphy’s stats during the season wouldn’t have justified anything nearly that high.
I was following the game on ESPN Gamecast (no video, just graphical interface of what was going on). They had a field for probability of scoring 2+ to 5+ runs this inning and (reasonably) kept changing it based on if there were runner on base (and what bases) and how many out remained. I kept thinking of yesterday’s discussion and wondering how off their numbers were.
I assume they were basing their probability on all games played this season and not just games between the Cubs and Mets, at Wrigley Field, in October (when it’s colder and windier). They certainly didn’t account for my biased belief that the Mets usually score less than 3 runs or more than 7 runs. Basically, I assumed their numbers were absolutely worthless.
All this talk of baseball put me in mind of the current XKCD, which I think sums up pretty well how most of the world views the game: http://xkcd.com/1593/
They all laughed at Daryl Bem but then he got statistically significant results and we now know that he was right all along. And that’s why we’re all riding magic carpets to work every day.
Hot hand!
Very nice !
When will then be now?
Soon.
+1
(How soon?)
Now he has to sit for a few days, and then he might need to take his talents to Canada to make it 7 in a row. Who can handicap all that? The estimates will swing from .05 to .2. Just as long as you never say never, you’ll never be wrong (but you migt place some bad bets).
Incidentally, the Michigan-Michigan State game Saturday ended with a last second touchdown off a muffed punt. An in-game-analytics tool had the MSU prob of winning at 0.2% with 10 seconds left, something that broadcasters kept repeating over and over in amazement. No one questioned how good an estimate that was. (I think that play, in that situation, results in a touchdown more than 2 in every 1000 times, but yes of course it was a low chance event.) It was a good example of lay-statistics (as well as black swans) because the 0.2% estimate was treated as an absolute fact. Somehow p-hat played a big role on sports shows that night to convey just how dramatic the game had been. Maybe we need to dial down the p < .05 threshold?
Huh? They play baseball in Canada?
Yes, but they haven’t been in the playoffs for 22 years, so you might forget about Toronto.
They play baseball in Canada?
Yes, but they call it American Baseball
But seriously, your post holds up very well the next day. Even if he hadn’t hit the home run in the 8th, he hit a deep fly earlier that, while not a home run, certainly justified an estimate of something like 20% even though looking just at Murphy’s stats during the season wouldn’t have justified anything nearly that high.
I was following the game on ESPN Gamecast (no video, just graphical interface of what was going on). They had a field for probability of scoring 2+ to 5+ runs this inning and (reasonably) kept changing it based on if there were runner on base (and what bases) and how many out remained. I kept thinking of yesterday’s discussion and wondering how off their numbers were.
I assume they were basing their probability on all games played this season and not just games between the Cubs and Mets, at Wrigley Field, in October (when it’s colder and windier). They certainly didn’t account for my biased belief that the Mets usually score less than 3 runs or more than 7 runs. Basically, I assumed their numbers were absolutely worthless.
All this talk of baseball put me in mind of the current XKCD, which I think sums up pretty well how most of the world views the game: http://xkcd.com/1593/
Bem would be proud.
Jkrideau:
They all laughed at Daryl Bem but then he got statistically significant results and we now know that he was right all along. And that’s why we’re all riding magic carpets to work every day.
It’s times like this that Internet anonymity was invented for.
Yesterday was Back to the Future II Day, but unfortunately, in contrast to the film, the Cubs didn’t make it to the World Series.