Copy-Trading My Friend for Two Months Cost Me $23. We Agreed to Stop Anyway.

Thirty-eight minutes. That is the gap between Marcus's fill and mine on the awfully first trade, March 5, 2026, his text landing at 11:52 at night saying filled ETH 2,041 long while my own order sat unopened on the couch cuz I was three episodes deep into a show I cannot remember the name of.

The terms lived in a Notes app both of us could see: $200 of my money, ring-fenced, copying his Kraken trades for two months, every fill texted within the hour, nothing bigger, nothing on margin. He keeps $2,000 of his own riding in the same moves. If it went bad, the money was tuition and the friendship was off the table. His wife knew. Mine knew. That part mattered more than the $200.

Here is what neither of us priced in: the copy would never be the trade. His entries ran 30 to 40 minutes ahead of mine almost every time, and one trade he took and cut never reached my phone at all. I will get to that one, because it is the reason this ended in agreement instead of argument.

the first text

The first text looked like a gift. ETH long at 2,041, his size $1,500, my scaled version $150. By the time my order filled at 2,068, the move he was riding had already spent its first leg, but ETH ran to 2,190 over the next two days and I closed where he closed, up $62 on my size while he made $744 on his. Great start. Wrong lesson. The gap between his fill and mine had already handed the market $27 of my edge, and I filed that under rounding because the trade paid.

$200, ring-fenced

Ring-fenced is doin' real work in that Notes app. The $200 came outta my fun budget, not the joint account, and the rule was written down before trade one: if the balance hits zero, done, no deposits, no revenge trades. Marcus has been trading since 2017 through a real job, real drawdowns, and a 2022 he describes as expensive education. I was not copying a guru. I was copying a guy whose losses I had watched him admit at a barbecue, which is the entire reason the experiment got past my own skeptic.

seventeen fills, scaled to a tenth

Across 61 days I copied 17 trades, each scaled to approximately a tenth of his size cuz my $200 next to his $2,000 is arithmetic, not ego. Fourteen of the 17 worked the way the first one did: his entry, a lag, then a version of the move with the juice already leaking. Two trades went bad on both of us at nearly the same proportional cost. Seventeen trades, two months. The 17th is the one that broke the premise, and it never showed up in my messages.

the trade he never texted

On April 9 he took a 22-minute scalp on a SOL range while I was in a dentist's chair, made $74 on his size, and closed it clean before lunch. He never texted it, because the deal said fills get sent within the hour and a 22-minute trade did not feel like a fill worth the thumb motion. His month ended with it. Mine never knew it existed. A copy misses the trades a trader decides not to mention, and he decided that one was not worth mentioning.

the delay was structural

The 30-to-40-minute lag was not laziness, it was the shape of his life. Marcus loads trucks at a warehouse from 6 AM to 2:30 PM with a phone he is not allowed to touch, trades over lunch at 11:45 in the parking lot, and texts from the loading dock when the belt stops. My fills came when his texts came. A real copy-trading rig would have bridged the gap in seconds. What we ran was two humans and a group chat of two, and the market charged us for every one of those minutes.

the tally

Final math, run twice cuz the first run made me angry. Down $23. He was up $188. Entry lag cost me about $19 across 17 fills at an average of 34 minutes late. Fees took another $12 at Coinbase Advanced's 0.6 percent taker rate over 34 total fills. The missing SOL scalp was worth $7 at my size, and his picks on the whole contributed a net positive $8 once you add the pieces. The experiment failed on mechanics, not on his judgment, which is the only reason it was easy to stay calm about.

the Cinco de Mayo verdict

We called it over tacos on May 5, same booth at Los Primos as always. Thirty-eight minutes, first trade. Slower every trade after. He said being watched made him self-conscious, checking a text twice before sending like a teenager with a crush. I said a copy of him minus 34 minutes is not him, it is a slower version of the market, and slower loses by construction. We kept the taco Tuesdays and killed the texts. The $177 stays in the account as a trophy. That is the entire postmortem.