Forex Money Management: Kelly and Fixed Fractional Risk
Position sizing can kill a winning strategy. I've got a screen recording from my second year of trading that I still can't watch without wincing. Direction? Right. Size? Wrong. Account died anyway.
That's the part nobody posts about. You can win most of your trades and still lose everything. Why? Because position sizing turns a good strategy into a coin flip. Every blown account I've seen — including mine — died the same way. Inconsistent size. Inconsistent stop.
Ten years trading XAUUSD. Mostly D1 structure with Fibonacci levels, mostly around the US session open. I spent years trimming my toolkit down to almost nothing. Five thousand indicators became one. But the single change that actually kept me alive wasn't an indicator. It was deciding, in advance, exactly how much of my account goes on the line every single trade.
Two rules do most of that work. Kelly's formula and fixed fractional stops. Most traders have heard both names. Almost nobody uses either correctly.
The real reason good strategies blow up
Good strategies don't blow up. Bad sizing blows them up.
Inside a dying account, the strategy was usually fine. The sizing wasn't. A trader risks a big chunk on a setup he loves, loses, then risks half of what's left on the next one because now he's scared. The account goes down through a door that a consistent fraction would have kept shut.
Two traders take the same signals in the same order. One risks the same fraction of her account on every trade. The other scales up when confident and scales down when rattled. Same signals. Same win rate. Two completely different equity curves. Who posts about the market being rigged? Usually the second trader.
So why does nearly everyone ignore sizing? Because sizing is boring. It doesn't screenshot well. Nobody's going to post a chart of their last hundred trades all risking the same small fraction. A big win on an oversized position gets attention. That gap between what looks good and what survives — that's the whole trap.
Kelly's formula gives you a ceiling
Kelly gives you a ceiling. It does not give you a target.
It comes out of betting theory and information theory, and it answers one question. If you know your edge, what fraction of your bankroll maximizes long run growth without ever wiping you out?
The formula takes two inputs. Your win probability, and the ratio between what you win when you're right versus what you lose when you're wrong. Feed those in. It returns a fraction.
The math is elegant. The problem is that it assumes you know your win rate and payoff ratio with precision. Do you? I've been at this a decade and I wouldn't bet on my exact win rate over the next hundred trades. The number moves. Regimes change. A strategy that worked in a trending gold market behaves differently in chop.
That matters because Kelly is asymmetrical. Underbet Kelly and you grow slower than you could. Overbet Kelly and growth collapses. Past a certain point, the expected outcome turns negative even with a positive edge. Bet twice what Kelly says and you can lose money on a strategy that makes money. Sit with that for a second.
This is why serious traders talk about fractional Kelly. Half Kelly. Quarter Kelly. You give up some of the theoretical growth rate and buy back a lot of survival. I'm not going to tell you that half Kelly is a magic number, because I don't believe in magic numbers. What I believe is that the gap between full Kelly and the realistic Kelly is the gap between a formula and a life.
Fixed fractional stops save you
Fixed fractional stops save you. That's what you do under Kelly's ceiling every single day. It's the least exciting rule in trading and the one that's kept more accounts alive than any indicator ever will.
The idea is simple. You decide, once, what fraction of your account you'll lose on any single trade if your stop gets hit. Not what you hope to make. What you will lose. And it stays the same whether you feel amazing or terrible.
Position size falls out of that decision. Take the amount you're willing to lose, divide it by the distance between your entry and your stop, and that's your size. The stop distance is set by structure, not by how you feel. Where does the trade become wrong? That's your stop. Then the math tells you how much to put on.
This flips the whole thing. Most traders pick a size first and then figure out where to put the stop so they don't get taken out. That's backwards, and it's why stops get moved. When you set size from structure, your stop has a reason to exist. When your stop has a reason to exist, you stop widening it at the worst possible moment.
On gold I usually place stops behind the shelf that just rejected price, the retracement level that held every prior pullback. I don't guess a distance. I look at where the structure would break my idea, and that becomes the number the size math uses. The Fibonacci level does the work. I just respect it.
The part most people skip
Order beats numbers.
Here's the piece that changes how you trade, so if you read nothing else, read this. The order of operations matters more than the specific numbers you land on.
Structure picks the stop. The stop and your fixed risk fraction pick the size. Kelly sits on top as a sanity check, telling you whether your edge is even strong enough to size up at all. If Kelly's number comes back tiny, that's the market telling you your edge is thin. Trade small or don't trade.
That sequence — structure, then stop, then size, then Kelly as the overlay — is the whole game. Get it in the wrong order and you're just gambling with a chart in front of you.
And if you want to know whether your edge is real, track it. Write down every trade, its win or loss, and the ratio of your average winner to your average loser. After enough trades you have your own p and your own b. That's the raw material Kelly needs, and guessing them is how people fool themselves.
Anyway, back to why this matters so much in practice.
What this looks like when it goes wrong
A good month can become a spiral.
I watched a friend do the opposite of everything above. He had a genuinely good month. Because he had a good month, he decided his edge was bigger than it was, so he sized up. Then he took a loss twice his usual size. Then he sized up again to get it back. That's not a strategy. That's a spiral, and it moves fast.
The cruel part is the recovery math. Losses and gains don't scale the same way. Lose a chunk of your account and the gain you need to get back is bigger than the loss that created it. The deeper the hole, the steeper the climb. Blow off half your account. What's the recovery? Not a normal year. A near double. That single fact is why risk management is the whole job. Not entries. Risk.
And no, I don't have a clean recovery story. The honest version is that I learned this the expensive way, and the tuition was real money. I watch my sizing like a hawk now. Not because I'm disciplined by nature. Because I remember.
I also still get it wrong. There are weeks where I size up a little because the structure looks clean and the US session confirms my read, and I have to remind myself that confirmation is not permission to overbet. I'm not fully cured of wanting to press. I just have a rule that catches it.
How to actually apply both rules
Start with loss, not profit.
If you want to start tomorrow, here's what I'd do.
Pick a fraction of your account you're willing to lose per trade, and make it small enough that five losses in a row doesn't change how you sleep. Set your stop from structure first, at the level that would tell you the trade is wrong.
Then size from that stop and your fraction. Every single time. No exceptions. No "this setup is special." Use Kelly as a check, not a command. Treat its answer as the absolute maximum you'd ever risk, then cut it down to something you can actually sit through.
The math isn't going to feel exciting. It's going to feel like wearing a seatbelt. That's the point. Nobody notices a seatbelt until the day it saves them.