[Roger Hunter]

(04-20-2014, 03:55 PM)Amit Wrote:

(03-27-2014, 11:12 PM)Roger Hunter Wrote: [quote='Island Chris' pid='205' dateline='1395959256']
Roger, your odds cannot be correct.
see:
http://earthquake.usgs.gov/earthquakes/e...qstats.php

There are 134 earthquakes M6-6.9 per year, or 150 quakes M6 or higher per year. So, almost one every other day. So, maybe 45% change of a quake any one day, and much higher chance of a quake within any 3 day period. So your odds of 0.000012 for 5 out of 6 must be off (by orders of magnitude?).

That assumes they are uniformly distributed and they are not.

Quote:OK, presumably many quakes are aftershocks, so if that is taken into account, odds are lower than if they were all randomly distributed. But your odds still do not seem possible by orders of magnitude.

What I do is divide the 1973-2013 time into 3 day windows, sort the 6.0-6.9 quakes into their respective windows and count how many windows contain quakes. Divide the number of quake windows by the total number of windows and it comes out to 0.101.

The question then is what are the odds on 5 consecutive hits if the chances are 0.101 and the answer is 0.0000105 for a single tail test.

Quote:
Also, if you were correct, comparing this to lottery does not make logical sense: Amit is correct that he would be a single lottery player.

Ok, I'll give you that. I was looking for a suitable analogy and that's all I could come up with.

In my next post I looked at the odds for the actual short time period in question and found they are much higher due to increased seismicity so the odds on 5 hits in a row were well within the range of possibility.

Roger

Quote:Roger
How many hits in March and April Roger?
Amit

Amit;

For the period 2014/03/01 to 2014/04/20 there are 17 three day windows and 14 contained quakes of mag 6+. That's 14/17=0.823 chances for a hit.

In that same period you made 8 predictions and got 6 hits. That's 6/8=0.75 success rate so right away you can see you're below chance.

Roger