Quote: Originally posted by RJOh on April 27, 2004
I have seen that theory many times about playing combinations with the highest number less than 32 and playing birthdays, but I've seen many such combinations drawn with no winners at all or only one winner so I don't count it in any strategy that I use.
RJOh
This has become much less of an issue with the super lotteries, which drain money from existing independent state lotteries, make their jackpots seem trivial (even when they are actually impressive), saturate "lotto fever," and thus reduce the size of the field coverage. It is also true that the field is so large in powerball and megamillions that double coverage is very rare, as I point out after each drawing on the "Probability of a rollover thread." Note that the values I post there, based on poisson distributions are in fact, minimal rollover probabilities. As I have pointed out in the past, the use of wheeling of restrictive number sets and the (still existing, I'm sure) tendency to play numbers associated with dates actually increases the probability of a rollover, albeit it slightly.
At this point therefore, the purpose of my particular stream is certainly diluted, but it's pretty clear that a stream is ideal if it is randomized as mine moreorless is, being based on very old California 6/54 data (with arbitrary start and finish dates). It is known very well that even in super lotteries, the numbers 1-2-3-4-5 Megaball 6 are usually multiply covered. (Recently a pick-5 jackpot was very low - I think it was in PA - for exactly this reason, because of a huge number of winners.) Therefore I continue to use my old stream, even knowing that it only now gives me a very, very slight boost in expectation value compared with what used to do in the days before powerball and megamillions. One could use any stream one wishes, randomized or otherwise, Doug's or someone elses, and still have the same probability of winning as someone with another stream.
As you say however, at the end of the day it just comes down to luck. Even if I have a set of numbers that is 100% unique 100% of the time, if the numbers are not drawn, I don't win. You might play the lottery in the most foolish way, but as long as you have a ticket, you still can win and undo the relative value of all my analyses.
By the way, the tendency to play date related numbers is not merely theoretical, ie, based on supposition. As part of a statistical study, the powerball released, in 1997, the distribution of numbers played in player selected tickets. The effect was quite pronounced and clear, at least back then. The number 7 was over represented by a quantity of over 19 standard deviations! The next highest represented numbers were 9, for September, a popular birth month owing to its coming 9 months after Christmas, and 6 for June, a common wedding anniversary month. This compares with a value of roughly 3 standard deviations for a randomized set. I cannot say, however, whether it has changed in any way in the last 7 years.