I seem to recall reading here recently a suggestion to use a moving average of the sum of Pick-3 numbers as a base for the next draw.
Moving averages of various kinds are often used to determine trends of gradually changing data, not for random fluctuations from one extreme to another. But why not try the idea if a short test indicates it might work for lottery numbers. Such a test should be based on a few hundred draws.
The evening draws in the CT Play-3 game from June 03 were made available to me together with their sums. A total of 406 draws were tested. In 406 draws one would expect only 334 different numbers and the balance, 72, would be duplications, bearing in mind there must be allowance made ffor standard deviation. No effort was made to count hte duplicates. All data were analysed with Excel. The daily moving average for 3 and 4 days were examined, with the idea of comparing the average with the sum for the next draw. Only whole numbers were used in Excel.
I got these results: 28 matches=7% for 3-day average and 33 matches =8% for 4- day averages. Most of these matches were based on the sums 10-17. Playing boxed numbers in this range would require betting on 13-15 boxed numbers each day. Tis does not look like a money making idea.
If some body has performed a more extensive test in another state, please report.
Bertil