The settings that keep the most

We tracked 28,600 Telegram calls, priced every ten seconds, and tested exit rules on the real price paths — entering ten seconds after the call, exiting at the next tick after the rule fires, and subtracting the fees, tips and rent we measured on chain.

ChannelBankroll SOL Per trade% Entry Bot

About the entry setting. Our first price reading lands within a fraction of a second of the call, so “at the call price” is the theoretical best — nobody can buy there, because landing a Solana transaction takes one to three seconds. “Ten seconds later” is slower than a good bot. The truth for an automated buyer sits between the two and we cannot measure it: our prices are sampled every ten seconds, so we default to the slower, more conservative one.

This opens the bot through our referral link so we get a share of the fee you were going to pay anyway. It does not place a trade and it costs you nothing extra — you still choose what to buy and when.

The median trade we actually measure on these bots is about 0.08 SOL. Try that size and watch what happens — that is the number most people are really trading.

Every bot at that size

Same calls, same rule, same moment. The only thing changing is who executes it.

Net per trade as position size changes

These lines used to cross zero at a size where trading paid for itself. On the corrected numbers none of them do: the gross return is smaller than the fee at every size, so the curve tells you which bot loses you least, not which one pays.

Every exit rule, ranked

Ranked by what each rule nets at the bot and size you picked. We split the calls in half by date and ranked all of these on each half separately: the order barely moved, and the fastest exits led both times. That is why we tell you to exit fast but do not hand you one tuned setting — the difference between the top few is inside the noise, and the best take-profit level we could find on the first half of the data lost -3.28% on the second. Pick from the top of this table, not from a number someone optimised for you.

What this looks like over days

Trades you would actually take per day

One thousand simulated runs, drawing each trade from the real distribution of outcomes above rather than from an average. The good case is almost entirely rare large winners: a run that never catches one lands near the median, not near the mean. Fixed position size, no compounding.

What the numbers imply

Not advice and not an optimum — just what follows arithmetically from the measurements above, for the bot and size you picked.

Why we do not give you an "optimal" setting. We searched take-profit and stop-loss combinations on the first half of the calls and tested the winner on the second half. The best in-sample rule, take +20% and stop -20%, returned -0.72% in-sample and -3.28% out-of-sample. The 30-second exit returned -0.80% in-sample and -0.82% out. Simple time exits survive; tuned targets do not. Any dashboard showing you a finely optimised win rate has almost certainly not run that test.

What this does not tell you

Our window is ten minutes, so nothing here evaluates holding longer — some calls keep running for hours and this study cannot see it. Returns are trimmed at the top and bottom percent, because a single call that ran 122x would otherwise carry the whole average: the untrimmed mean is positive for several rules purely on that tail. We assume you take every call from the channels we track, at one size, with no judgement applied. And past calls are not future ones.

Fees and fixed costs come from The League. Call paths come from the tracker. We take referral rebates from the bots we link to and name whichever is cheapest at your size, including when it is not the one that pays us most.