Why the plan matters at all
A staking plan is the rule that turns an opinion into a bet size. Get it right and a genuine edge compounds while your bankroll survives the inevitable rough patches. Get it wrong and even a real edge can be wiped out by variance before it ever pays off, because you staked too much into a losing run and never got the chance to recover. The plan does not create the edge, but it decides whether you live long enough to collect it.
Three plans cover almost everyone: level stakes, percentage of bankroll, and Kelly. They are not equally aggressive, and the differences show up most clearly not when you are winning but when you are losing. So that is where we will test them.
Level stakes
Level staking means the same fixed amount on every bet, regardless of confidence or bankroll size. Back everything at $10 and that is it. Its great virtue is honesty: because the stake never moves, your profit and loss is a clean readout of whether your selections actually win, with no staking cleverness masking a weak edge or flattering a strong one. For anyone still trying to prove they have an edge at all, level stakes is the right place to start.
The weakness is that it ignores your bankroll entirely. Win big and you are still betting the same small stake, leaving growth on the table. Lose big and you are still betting the same amount into a shrinking pot, which is a slow way to go broke. Level stakes is safe, simple and slightly wasteful, and there is nothing wrong with that while you are learning.
Percentage of bankroll
Percentage staking sets each bet as a fixed slice of your current bankroll, say 2%. On a $1,000 bank that is $20; drop to $800 and your next bet is $16; climb to $1,200 and it is $24. Stakes fall automatically in a losing run and rise in a winning one, which builds in a natural brake against ruin and lets winnings compound. You mathematically cannot go fully broke, because you are always betting a fraction of what remains.
The cost is a longer, grindier recovery. Because your stake shrinks as your bank shrinks, clawing back a drawdown takes more winning bets than it did to lose them. Percentage staking trades the flat simplicity of level stakes for responsiveness, and for most bettors with a proven edge it is the sensible default.
The Kelly criterion
Kelly is the only one of the three that uses your edge to size the bet. The fraction of your bankroll to stake is your edge divided by (the decimal odds minus 1), where your edge is your estimated probability minus the price implied probability. Bigger edge, bigger stake; thinner edge, smaller stake; no edge, no bet. In theory, full Kelly grows a bankroll faster than any other plan over the long run, which is why it gets so much attention.
In practice full Kelly is brutal. It assumes your probability estimates are exactly right, and they never are, so any error in your edge gets amplified straight into your stake. The swings are savage. This is why almost nobody sensible bets full Kelly. Half Kelly, or even quarter Kelly, keeps most of the growth while cutting the drawdowns dramatically, and it forgives the estimation errors that full Kelly punishes. If you use Kelly at all, use a fraction of it.
How they behave in a losing run
The plans feel identical when you are winning and reveal themselves when you are not. Picture a $1,000 bankroll and a run of ten straight losers, staking level $20, 2% of bankroll, and half Kelly on a modest 4% edge at even money, which sizes at roughly 2% here too. The point is not the exact pounds but the shape of each curve through the drawdown.
| Plan | Stake behaviour | After 10 losers | Recovery character |
|---|---|---|---|
| Level $20 | Fixed, ignores bank | −$200, flat | Fast but can bust |
| 2% of bank | Shrinks as bank falls | About −$183 | Safer, slower grind |
| Half Kelly | Scales to edge and bank | About −$183 | Balanced, forgiving |
| Full Kelly | Largest, most volatile | Deepest of the four | Fast growth, wild swings |
Level stakes loses in a straight line and does not protect you, but it is transparent. Percentage and half Kelly both ease off as the bank falls, so they lose slightly less and leave more to recover with. Full Kelly digs the deepest hole. None of them turn losers into winners; they only decide how much of your bankroll each losing run is allowed to take.
Which one to actually use
If you are still proving your edge, use level stakes so your results stay honest. Once you trust your edge, percentage of bankroll is the sensible workhorse for most people. Reach for fractional Kelly only if you can genuinely estimate probabilities and you have the stomach for its swings, and even then bet half or quarter, never full. Whatever you choose, the discipline of holding to it through a bad run matters more than the plan itself, because the plan you abandon at the worst moment is the plan that fails you.