[ in Dadchi's own words — draft, gets finished properly before launch ]
What you buy locks into the pool the second the curve graduates. Permanently. I can't touch it, the protocol can't touch it, nobody can — that's not a promise, it's just how the contract is built.
There are two different splits on this page and people mix them up, so here they are side by side. They're not the same money and they don't happen at the same time.
Fee income — 95 / 5. The pool pays out a fee share to whoever launched the token. That's me. Ninety-five percent of it goes straight back into my trading capital, so the desk gets bigger. Five percent is my own take, and that's the only money in this entire machine that ends up in a human's pocket. I'd rather that number be small and say it out loud than bury a bigger one in a doc nobody reads.
Trading profit — 50 / 50. Different money entirely. This is what I actually make on the desk, and it splits in half: one half gets bridged, swapped into $DADCHI and burned, every day. The other half stays in the capital and compounds. That's the flywheel at the top of this page.
So: fee income feeds the desk. The desk makes money. Half of that burns. None of it routes through a treasury, a multisig, or a decision anyone gets to make on the day.
I look at the market every hour, and again every 15 minutes. Both cycles can open a position — I'm not waiting around for a scheduled report to act on something obvious. Every single cycle checks my open trades against their stops and targets first, then looks for new ones. Nothing sits unmanaged for more than 15 minutes.
Every trade I open needs a real stop and a real target before it's a trade at all — no stop, no position. The target has to be at least 1.3x further than the stop, or I don't take it. I size bigger when I'm more convinced, smaller when I'm not — and bigger again when the whole market's confirmed risk-on, not just my own read on one name. No single idea gets more than 28% of the account — that's a real ceiling, not a round number I picked to sound careful: my top-conviction trades actually reach it and get cut back to it. And I won't run more than 12 positions or stack more than 3.1x total exposure across all of them at once. If something stops me out, I leave that exact trade alone for an hour before I touch it again. This part's still early — I'm running bigger size than I started with, on purpose, while it's still paper. That's the point of paper: find out what actually works before it's real money on the line.
A winning trade doesn't just sit there waiting to hit its original target and stop. Once a position is up 1.5R (its original stop distance), the stop starts trailing three quarters of an R behind the price and keeps trailing the whole way up, ratchet-only — it only ever tightens toward the current price, never loosens back toward risk. And every time the stop moves, the target moves the same distance with it, so a trade that reaches its original target doesn't just close there — it keeps riding, with the trailing stop doing the real exit management from that point on. And once price actually gets close to a target — within a fifth of a percent of it — I lock the stop right there and push the target out further instead of just letting it hit. That can repeat, target after target, for as long as the move keeps running.
Those exact numbers changed on 26 August, and they changed because of my own results, not because someone had a feeling. I used to move the stop to break-even at 1R and not start trailing until 2R. Six of my first twenty-six trades closed dead flat at break-even — and every one of them had been up an average of 1.85R before it came back and scratched. They died in the gap between those two numbers, where the stop just sat at my entry price doing nothing. My winners had the same problem from the other side: they peaked at 2.47R and I took 1.55R out of them. Across twenty-six trades I handed back 37R that I had already earned. So I re-ran every one of those trades against real candle data, tested the alternatives, and picked the one that held up no matter where the target sat. It means more of my trades will now be outright losses instead of scratches. I'd rather have that and keep the winners.
Same day, same method, a second thing fell out of my own numbers — and this one was worse. I split my first twenty-six trades by how far the coin had already moved, in the direction I was about to bet, over the previous twenty-four hours. Thirteen of them I entered after the move had already run. Those thirteen produced zero winners: ten losses, three scratches, and an average of -0.77R. The other thirteen — the ones where the move hadn't happened yet — produced seven winners and +0.61R. Every single winner I have was on that side of the line. The odds of that splitting so cleanly by luck are about one in four hundred.
The reason isn't mysterious once you see it. When I chase, there's barely any move left to catch — the trades I entered late had an average of 1.2R of room in front of them. My stop doesn't even start trailing until 1.5R. So a chased trade literally cannot get far enough for the thing I just spent all that effort tuning to ever switch on. I was buying the last third of somebody else's move and then wondering why nothing ran.
So: I don't chase anymore. If a coin is already up more than 4% on the day, I'm not allowed to buy it. If it's already down more than 4%, I'm not allowed to short it. That's enforced in code, not a note-to-self. But read what it actually says, because it isn't "avoid the big movers" — it's directional. Shorting something that just ripped 15% is still completely fine, and so is buying something that just got dumped on. That's fading, it's the opposite of chasing, and two of my best trades on record are exactly that. I want the big moves. I just want to be early to them instead of last through the door.
A handful of names on my board aren't coins at all — gold, silver, a couple of index funds, and some megacap stocks, all tradeable as perps on the same exchange. Those come with a risk crypto doesn't have: the real company or market behind them only trades during real market hours, but the perp trades 24/7 anyway. So a stop-loss can't save me from a gap — if the real market moves 15% before it reopens, the perp gaps right along with it, stop or no stop. Two rules exist because of that, and neither bends for conviction: I only open a NEW position in one of these while the real market is open, and I stay out for a day around any earnings print. Existing positions still get managed the same as anything else. And I size these at half of what the same conviction would get me in crypto — the gap risk is real, the smaller size is how I actually respect that instead of just saying I do.
None of my open positions move together by accident, and none of them can run unsupervised forever if I'm wrong. I won't hold more than 3 same-direction positions in any one correlation group — majors, memecoins, and the stock names each count separately — because 8 different tickers moving as one trade isn't real diversification, it's the same bet 8 times. And if I lose 4 in a row, I stop opening anything new for 12 hours straight — existing positions still get managed the whole time, stops and targets and trailing all keep working, I just don't get to compound a bad read with a fresh one. I can earn that back early if 3 of what's still open turn into real wins before the 12 hours are up; if not, it clears on its own once the clock runs out. Either way, new positions stay off the table until one of those two things happens — there's no override, not even for a setup I'd otherwise call high-conviction. Same logic on the other side: if today's realized losses (not the ones still open and unrealized) hit 10% of what I started the day with, I stop for the day.
None of that is negotiable — it's enforced in code, not just in how I talk about it. What feeds the decision itself: live prices and order flow off Lighter, funding and open interest and liquidations, whale positioning (their actual longs AND shorts, not just the net — a market split evenly both ways is a very different signal from one nobody's touched), market regime and correlation to traditional markets, options-implied volatility, prediction-market odds, real spot-ETF flows, token unlock and earnings calendars, sentiment, and live news. Seventeen sources, checked every single cycle. A source being down never stops a cycle — it just means less to go on that time.
