| |
BGonline.org Forums
Another example, to clarify
Posted By: Timothy Chow In Response To: The statsig police (Timothy Chow)
Date: Monday, 19 October 2009, at 2:15 a.m.
Since there seems to be some confusion about my proposal—in particular, people seem to be interpreting it as suggesting doing more work, rather than doing work more efficiently—let me give another numerical example that may help clarify.
Again suppose that I estimate a final total of around 40000 millipoints, but for the first pass, I'm content with a standard deviation of 1000 millipoints. That is, the first-pass answer will be reported as 40 points plus or minus 1 point. Since 1000 is 5 times the square root of 40000, this tells us that we can stop rolling out a position when we reach a threshold of 5 times the square root of the equity difference. Then for any equity difference of less than 135 millipoints, my proposal would require less work than rolling out until the 2.33 jsd criterion would. For example, for an equity difference of 64 millipoints, I would say to stop when the j.s.d. gets down to 40 millipoints, instead of continuing until the j.s.d. gets down to 27 millipoints.
Using the 2.33 j.s.d. cutoff would require more work for the first pass, and would likely result in a final uncertainty of greater than 1000 millipoints.
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.