10-22-2009, 01:09 AM
Pretty sure we had 3 at least, because I was making a mental note of each and wanted to say something on vent, but I've kept this 'debate' out of raid. 3/20 = 15%, which isn't bad, considering we can even not notice the times where anub chases someone before coming up, then he goes under and chases that same person again (which would be a 'valid' back to back in this logic).
At this point I really don't want to think about that creating an algorithm that does back to backs with multiple chases, separated into subsets because that would really fuck with the numbers more. More than likely it would be a probability OR statement like (3,2) OR (4,2) (F(3,2) + F(4,2) * (1-F(3,2)) by basic statistics) if we wanted to match it to our current strat (which comes out to roughly 18.5%, surprising).
Again, real performance never matches 'probability' because, after all, it's a chance.
Until I hear from you, I await for your case breaker. Just saying "LOL IT DOESN'T LOOK REASONABLE" will not float. You'll have to try for something above the 7s, as I used the 3s, 4s, 5s, 6s, and 7s as my test cases to verify the algorithm works. Right now my big cases are still running so I'll post them when they become available. Like I said I'm brute forcing the actual probability so it's taking longer since it has to determine if each possible combination has the right or wrong set up.
Finally some alpha examples, so if you do go with a program you can verify you're getting the alpha I expect.
(7,3) - Alpha = 2
(8,4) - Alpha = 3
(8,5) - Alpha = 3
(7,4) - Alpha = 3
At this point I really don't want to think about that creating an algorithm that does back to backs with multiple chases, separated into subsets because that would really fuck with the numbers more. More than likely it would be a probability OR statement like (3,2) OR (4,2) (F(3,2) + F(4,2) * (1-F(3,2)) by basic statistics) if we wanted to match it to our current strat (which comes out to roughly 18.5%, surprising).
Again, real performance never matches 'probability' because, after all, it's a chance.
Until I hear from you, I await for your case breaker. Just saying "LOL IT DOESN'T LOOK REASONABLE" will not float. You'll have to try for something above the 7s, as I used the 3s, 4s, 5s, 6s, and 7s as my test cases to verify the algorithm works. Right now my big cases are still running so I'll post them when they become available. Like I said I'm brute forcing the actual probability so it's taking longer since it has to determine if each possible combination has the right or wrong set up.
Finally some alpha examples, so if you do go with a program you can verify you're getting the alpha I expect.
(7,3) - Alpha = 2
(8,4) - Alpha = 3
(8,5) - Alpha = 3
(7,4) - Alpha = 3
![[Image: 2unznl0.gif]](http://i42.tinypic.com/2unznl0.gif)