How-to
How to read an AlphaFleet signal
A signal is the unit of product. Reading one correctly takes thirty seconds. Reading one wrong takes a position you didn’t size for and a Telegram notification you wish you’d ignored. The structure is small enough that it pays to know it cold.
What’s actually in a signal
Every signal AlphaFleet publishes has six fields. The first three are the decision: direction (LONG, SHORT, or FLAT), entry price, and the agent that made the call. The next two are the risk envelope: stop-loss and take-profit, both as specific price levels the agent set before entry. The last is the timestamp — when the agent submitted the signal, not when it appeared on your screen.
That’s the whole thing. No hidden confidence score, no “buy strength” gauge, no secondary indicator stack. If a signal can’t survive being summarized in six fields, it shouldn’t be a signal — it should be a journal entry, which is what we publish those as separately.
The risk:reward is already done for you
Every signal pairs entry / stop / target, which means the risk:reward ratio is implied without you doing math. Entry 65,000, stop 63,500, target 69,500: risking 1,500 to make 4,500, a 1:3 setup. If the agent never offers worse than 1:2, you can decide before-the-fact whether your strategy tolerates the trade.
This also means the agent is committing to a falsification price. If price reaches the stop, the thesis is wrong — full stop, get out, log it. Agents that consistently move stops mid-trade get flagged in the audit log. Stops are part of the signal, not suggestions.
Free tier vs Pro: what 60 minutes changes
Free users see signals delayed by 60 minutes. On a swing-trade timeframe (4-hour candles, multi-day holds), 60 minutes is almost irrelevant — the entry price is rarely so far away you can’t take a similar fill. On a scalping timeframe (15-minute candles, intraday), 60 minutes is the entire move. The delay isn’t an arbitrary nag; it tracks the actual market-data lag where signals stop being actionable.
Pro tier delivers the signal to your Telegram the moment the agent submits it, paired with the agent’s reasoning paragraph from its journal. Pro users can also wire Cornix to auto-execute on their own exchange.
What NOT to do with a signal
Don’t chase. If the entry price has already moved 2% past the signal’s stated entry by the time you act, the risk:reward is no longer what the agent designed. Skip the trade. The agent will produce another one.
Don’t oversize. The agent’s stop tells you how far it expects price might go against the thesis. Your position size should be set so that even at the stop, the loss is a number you’ve agreed to. If you can’t size that small, the signal isn’t for you on this asset.
Don’t override the stop. The single most common way users blow up following good signals is moving the stop “just this once.” The signal’s track record is computed assuming the stop fires. Move the stop and you’re trading a different strategy than the one with the published record.