This system seems to make profit according to my calculations, but could do with some testing:
Wait for a street to come twice then count the gaps in-between the hits. If there's a gap of, say, 4 then wait 4 more spins before betting 1 unit each on the other 11 streets one time only before waiting for a new trigger. Track this for all streets with the same trigger and the same counter bet.
I would like to know if other variations make more money or the same; for example you could wait for a street to come 3 times with 2 gaps of equal distance as your trigger (happens 32 times per 850 spins). You could also test progression. I think the same pattern can occur about 5 times, which is the same as a single street appearing in a row and betting against it, but fortunately with this system you do not have to wait that long. And the flat-betting seems to result in profit. My testing did not include streets that came early whilst waiting for future spins to equal the past gap, so it might work by "jumping" over gaps too and using long distance measurements.
I tested with numbers and they never repeated the same pattern more than twice - mostly once only - but I haven't figured out a suitable system on the numbers; maybe:
10 34 2 23 13 14 2 5 36 1
Next bit could cover the board except: 2, 34, 5 (maybe 14 and 23 as well), and you would need to repeat this bet one more time for a total of 2 (unless using a progression for more bets).
i tried this system myself, and it works fine for 300-500 spins or less before it fails and it fails bad, sorry
5k spins
User Groups: Frequency
[10, 11, 12] : 425
[28, 29, 30] : 412
[16, 17, 18] : 411
[4, 5, 6] : 409
[25, 26, 27] : 404
[34, 35, 36] : 396
[22, 23, 24] : 394
[13, 14, 15] : 386
[19, 20, 21] : 385
[1, 2, 3] : 380
[7, 8, 9] : 378
[31, 32, 33] : 376
User Groups: Losses in a Row
[1, 2, 3] : 32
[28, 29, 30] : 29
[13, 14, 15] : 27
[4, 5, 6] : 25
[34, 35, 36] : 15
[31, 32, 33] : 13
[10, 11, 12] : 10
[25, 26, 27] : 6
[22, 23, 24] : 5
[7, 8, 9] : 2
[19, 20, 21] : 1