Nice article and photos!
Great interview AND great drawing!
Very cool that you had Quirin as your prof! Just kind of thinking about what you've said here, it occurs to me that statistical significance does not equal a positive ROI. Last, your sources are good, for sure. I will say this: it's probably not right to say that his data don't reflect anything. they reflect his own performance. And yeah, there are things that he could do differently. But, as I look at this, it's an interesting study of one bettor's betting patterns, and yes, they are a bit messy...but I would say that they have value as a heuristic, both for Dark Horse, and maybe for others. My take...yours are intresting observatons too!
So, while I agree that more is better, it is not simply a matter of x cases conferring the possibility of significance and below x not. It is more an issue of how much power to detect significance a particular test might have, given the sample size, and alpha level, etc. 10,000 seems high...and I think it varies by test, among other things. That said, these are great questions to kick around...and The Dark Horse's data, while not reflecting the sine qua non of research methodology, provides grist for the discussion mill, eh?
I would argue that while the splitting might yield more data...sometimes the lumping is the best solution. In my estimation, the best statistical arbiter of any of this is a validation sample.
Hi Travel Vic...good observations for some of this...I would point out No research is perfect. The Quirin data are all a mishmash of different tracks too, and for those data that are not, he really does not say where the data are valid, and where they are not. Just my memory...I don't have the book with me.
Favorite '80s hit? Ha! It's BCD. Blog Crossover Disorder! Nice writeup Chris!
I appreciate the likes, you all. Not feeling well these days, but I'll bounce back sometime soon. Watching the Oaks and Derby are tonic for the soul.
Copyright © 2010 -