Well, there are some problems involved and maybe mostly if playing off-line with the states as they don't pay much.
For example if you was playing straight and you would filter 1 digit out for all 3 positions (The same digit), that would leave 900 straight numbers and maybe 9 out of 10 draws you might get your money back online $900, but 1 time out of every 10 draws you might fail, if so them every 10 draws in the long run you would or might lose $900, that might be about $90 per each draw played in the long run, why filter and play then?
When you analyze stats you might see the the "Filters Patterns" "Fail" every so often, they show failure rates, the stats are not quite even might maybe in the very long run try to equalize themselves.
First of all, the last few stats-patterns have priority over older stats-patterns.
All stats have their own uses, short-term, medium-term and long term, it depends mostly on prediction technique(s), but short term and medium term have precedence over long term-stats and mostly short-term stats have precedence over the other 2, but as I said, it mostly depends on the prediction technique(s) used.
You can expect the failure rates to be at least as bad as they should be, like there are 10 patterns so each should fail at least 10% of the time, if there are 5 patterns then maybe each might in the long run fail 20% of the time or once out of every 5 patterns-runs.
I am just more or less guessing here.
The more filters-patterns used the greater the risk of failure as the rate of failures might "Compound" in direct relation to the number of filters-patterns used, their kinds, numbers, Etc.
Filtering is both an art and a science and it might also take some intuicion.
You can beat "Random", but it might be harder to also make a profit.
Trying to find relation(s) between different kinds of patterns and also finding and using "Filters Tricks" might cut down some on filters failures, also at times some filters-patterns seem to "Show" more "Stability" and fewer failures, not all filters-patterns are equal, some might appear to be more "Dependable" at times or often or more often than other filters-patterns.
A better understanding of "Changes" might come little by little thru time after long practice-use-trials and meditation.
One must inderstand that even if 1 thing might not have anything to do with another, statistics can be used to try to relate them with in some way(s) and that some or many things are not 100% unpredictable nor 100% predictable, the more understanding that a person has maybe often the less unpredictable that something might be.
A "Regular" person will think that there is a good chance that the sun might come out sometime tomorrow, but a meteorologist might know much better and even at what time it will come out or if it will not.
-
The study of stats is not a waste of time, stats can be of use.
But, What to study, How and How to interpret them?