No better example to prove it. Real numbers, California Daily 3, and the most recent.
Start from July 2, 2020
End July 16, 2020
Total 30 draws (2 a day)
21 draws were singles (all digits different)
9 draws were doubles (2 digits the same)
Screen below shows the stats.

Backtrack is for the same time frame.
8 growing trend digits from 20 draws.
As usual clusters can clearly be identified.

All doubles were the winners (9 of 9). Singles scored only 57% (12 of 21)
In total there were 21 wins in 30 draws (which was also one of the best overall scores).
Screen with stats below.

All the simulations from the past point to the same: if you want to make money in Pick 3 - hunt doubles.
This is the best proof I've seen so for. And it's real, no test data. Just entered the draws today, then ran few simulations and almost immediately stumbled on these results.

ROIs would probably be better as in California prizes are variable, if I recollect correctly.

I will enter the data for Florida & New York as well and see how these are doing. Maybe another shocker? If there are others, maybe they are not shockers any more but a norm??? We will see, and soon.