It appears that the better systems will average just north of 20% S/R win.

Why not just select certain races, provincial races, maidens, city group races .... whatever and bet only when the market is favourable. By "favourable" I refer to looking at backing a horse that is paying the closest to 4/1, but must be no less than 4/1. Now that method alone will not win you money, but just say you will be betting in 5 races on the day. If you decided that when (IF) you backed a winner that you would bet all the winnings, plus the stake on one of the other races. So for a $20 bet all up, to cover your 5 bets for the day, your outlay would be $200. Five horses 1, 2, 3, 4, 5. $20 on #1 all up #2. $20 on #1 all up #3 and so on. $20 on one bet, all up one other bet (@ say 4/1 and 5/1) would give you $600, a profit of $400 for the day.

So, what are the odds of selecting 2 winners from 5 bets @ better than 4/1?

If you can do it at a better rate than one meeting in three you are in front. If you happened to fluke 3 winners in one day that would be a profit of $1600 on the above scenario.