Quote:
Originally Posted by BrianTheMick2
9am-10am EST gets same results for US stock and bond futures.
If you look at 24 iid random walks, you can make 276 (=24 choose 2) comparisons, or even 23 comparisons if we're just comparing adjacent hour periods. That allows room for finding statistical anomalies. Even 1 test for equal means at 95% confidence is expected to fail here.
I'm not dismissing the findings entirely, but I don't think it's as interesting as most people look at it initially, and certainly not a tradeable finding. That would give pairs trading a whole new meaning!