The best staking plan would be one that has a larger bet on the animals that are the Larger Overlays.

The way you find out whether you are capable of finding Overlays, is to pretend money on perhaps 1000 horses that you assumed were Overlays.

Pretend a Unit on each horse and if the return is higher than unvestment, it means you can do it , and have overcome the 15% deductions the TAB takes.

Pretend these same horses , but this time bet to Prices, (S.P.) , and see if there is still a Profit on turnover.
I don't know whether there will be or not, but I assume there may be.

But this time, bet to your own calculated Prices , and this time, your Profit on Turnover will increase heaps compared to the level stakes.

But the very best way to increase POT is to ignore bets like where you figure an animal as being an $2 Chance, but is available at $2.05
This animal would cause a substantial investment for very little return in the long run.
These animals may very well salute often, but just as often, they will go down.

People who reckon Mathematics can prove staking plans that don't show a Profit on level stakes are doomed are misguided.
If Mathematics were such an infallible method, then these so called Mathemeticians would be winning all the loot due to applying Mathematics to who will win the particular race.

There are staking systems where an increase in bet size is due to the Odds of the next horse bet winning.
For Example if you followed say, and I'm not a horse person, Bart Cummings, and icreased your bet size , surely every time he loses, he is 1 race closer to a win.
Its not like roulette where there is no reason Black must come up after a Red.
With Trainers, or Jockeys, they MUST win or they don't eat, so the next race they will try harder and that makes it more likely following a loss.