I had to do it.
Randomizing the timer, we wrote a pseudo-random number generator to generate 5 integers between 1 and 35, sort them, toss out any sets with dupes, and write it to a junk file. In all we ran 15295 arbitrary sets. To make it a little more precise, rather than just plotting how many 1s fell in the first position, and 2s in the second, we further broke the numbers into front and back digits, producing 10 positions as such.....
01 12 23 34 35
Blue numbers are denoted in the first column of the chart as the odd numbers 1,3,5,7,9 (going down).
Red numbers are denoted in the first column of the chart as even numbers 2,4,6,8,10 (going down).
We did have some concerns looking at the random numbers produced that things would not go well. And once the chart was complete, looking over the list, the sets seemed askew to the extremes with some sets being too low and others seemingly very high. So we marked the highest occurrences in the front digits, those being the biggest concern. (In row 10 columns 6 and 7 are highlighted for a different reason)
This is all well and good, except we have nothing to compare it too. So we ran the matrix......
Amazingly the pseudo-random number generator closely matched the matrix! Not sure how much closer we can expect to get. Naturally the real question is how do these compare to the game.......
Well, it's close except those red numbers seem a bit askew. But after some thought we realized when compiling the Cash5 game file it included winning draws back to 1995 when the Texas Cash5 was a 5/39 and then a 5/37. With that understood we can see why we have less 2s falling in the 9th position. And the larger matrix would mean more 3s falling in the 7 and 9 position. The gaudy double digit percentages in the 6 and 7 column are the result of the larger matrix as well. But as we see the junk file and the matrix produce comparable results in those same columns.
Now, at some point that pseudo-random number generator will begin to repeat itself. I don't know at what point that is. (thought JadeLottery did a post on this once but can't find it). Of course by using the timer for the seed we don't start out with the same seed each time, which means we are more diverse in the sets produced.
None of this is to say that pseudo-random number generator is capable of producing the result you wish, or one that suits your needs. It is basically nothing more than the satisfaction of a curiosity.
G