You cut the winner at the first sign of green.
You were up 22% eleven minutes in and you took it. You kept the chart open, because you always keep the chart open, and forty minutes later the same token was up 380% and someone was posting a screenshot of the position you used to have.
It felt like discipline. It was not. Discipline is naming a level before you enter and taking it when it prints. What happened is that the position stopped being a bet and started being money, and the urge to protect money beat a plan you never wrote down.
You have never done this in the other direction. You have never closed at down 22% eleven minutes in because the number moved. Red gets patience. Green gets executed on sight. That gap is the mistake, not the sell itself.
Why green pulls the trigger
Unrealised profit does not read like a position. It reads like cash you are currently in danger of losing. The second the number turns green, holding stops being the default and becomes an active decision to risk something you already have. Selling ends that. What you bought was not the 22%, it was permission to stop watching.
The reflex is not baseless. You have watched a token round-trip from 4x to under your entry in ninety seconds. That is why it survives: every token that dies after you sell confirms the rule, and every token that runs gets filed as bad luck instead of the same decision with a different outcome.
What small wins do to a record
The problem is not the size of the win. It is the size of the win next to the size of the loss. Winners that average 20% against losers that average 50% need five wins in seven to finish flat. Shrink the winners and leave the losers where they are, and the hit rate you need climbs again.
Cutting every winner at first green puts a ceiling on your best outcome and leaves the worst one open. You cap the upside yourself. The downside is capped by whatever the token decides to do. Fees push the same way: priority fees and slippage on a thin pair take a real bite out of a 12% win and are noise against a 4x.
You are running two different rules
Compare how long you hold green positions against how long you hold red ones. If the losers sit three times longer, you are not running one strategy. You have a fast rule for winners and no rule for losers.
The cause is simple. A green position offers a painless exit available at any moment. A red one offers a single exit that costs you the admission you were wrong. So you take the free exit whenever it appears and refuse the expensive one until the chart takes it for you. Both halves come from the same reflex, which is why the selling half does not stay fixed on its own.
Telling an early exit from a good one
An early exit and a correct one look identical when the token dumps afterwards. The difference is what you were reacting to, not what happened next. Dossier grades the exit against what was visible on the chart before your sell. The same test works when you run it on yourself.
- You named a level, a market cap or a multiple before entry, and you sold there. That is a plan being executed, whatever the chart did after.
- Something changed: volume died into your candle, the bid thinned, the move broke structure. That is information.
- You hit an exposure limit you set earlier. Risk management is allowed to look early.
- Nothing changed except the P&L going green, and you sold seconds after it crossed a round number. That is a flinch.
- You would still be holding if the same candle printed with you 20% down. That is the tell.
What actually changes it
Deciding to hold longer does not work, because you make that decision flat and break it in the position. The decision has to sit where you are holding nothing: before entry, written down, in terms of the token rather than your P&L. A rule expressed as a percentage of your own money is a feeling with a number attached. A rule expressed as market cap, volume or a level is a condition that either happens or does not.
The other half is the record. Tokens you sold early stop existing the moment you close the tab. The 380% never enters your version of events. What survives is the ones that died after you got out, and that memory will keep telling you the rule works until something outside your head is keeping score.
- Hold duration set against where your exit sits on the reconstructed chart. Nine minutes held, sold into a candle with volume still building, is a different trade from nine minutes held into a stalled tape.
- Price and market cap at entry and exit, then the same chart carried 60 minutes past your sell. That is how the size of what you left is measured, in market cap rather than a percentage you can argue with.
- Whether anything actually deteriorated before the sell: volume through the candles running up to your exit, and whether the bid was thinning or still filling when you clicked. An exit into strength reads differently from an exit into a broken chart.
- The gap between your average hold time on green exits and on red exits, taken from entry and exit timestamps across your closed trades. The asymmetry shows up in the timestamps alone.
- Position size in SOL and where the trade sits in the sequence of your closed swaps. A cut that lands straight after a loss, or on a position sized too large to sit through, is a different trade from the same cut made cold.