CoinGPT
How to Find Crypto Trades with an AI Scanner
A crypto scanner does the first pass for you: it runs the same checks on every coin and timeframe and lists where something just fired. The trick is reading it right. A score of 7 does not mean a 70% chance to win. Here's what a scanner actually checks, how to read the setup score, the filters that cut 65 setups down to 15, and a five-step routine.
The manual way, and why it doesn't scale
By hand, you open a chart and check the trend, the momentum, the volume and the candles. Then the next timeframe up. Then the next coin. On the evening of 5 November 2026, the CoinGPT Scanner was showing 107 setups across 8 timeframes. Checking that by hand takes hours, and by then most of them have moved.
How to read a crypto scanner
In the CoinGPT Scanner, each row is one setup: the coin, the price, the move in the last 24 hours, the setup score, the signals behind it, and the timeframes where it showed up. The button on the right tells you the side, long or short. Sort by score and the most evidence goes to the top.
The setup score is not a win rate
From 0 to 10, the score measures how much independent evidence agrees on one side. It is not a probability and it is not a win rate. Here's how one score is built, using a HYPE short on the 5-minute chart from 6 November 2026, scored 6.4:
| Evidence | Points |
|---|---|
| 2 of 5 signal types agree | 2.2 |
| The 15-minute chart agrees | 1.8 |
| It fired on this candle | 1.2 |
| How strong the reading is, and how many signals | 1.2 |
| Total | 6.4 (Fair) |
The bands: 5 is Fair, 6.5 is Good and 7.5 is Strong. Different kinds of evidence count for more than more of the same: trend, momentum and a candle agreeing beats three momentum readings saying the same thing.
A five-step scanner routine
- Filter. Ask for two or more signal types agreeing, and keep the minimum score at 5. On 6 November that, plus a coin filter, cut 65 setups down to 15.
- Start at the top. The highest scores have the most evidence behind them. Don't skip down the list because you like the coin.
- Read why. Open the setup and tap any signal; it explains itself. Then look at the timeframes. A setup that shows up on the bigger charts too carries more weight.
- Check the bigger trend. In the filters, switch on Higher timeframe agrees and the list only shows setups where the higher timeframe agrees.
- Plan it with a stop. Send it to the Trade Agent, which drafts the entry, stop and target. How to check those is in where to put your stop loss.
HUMA, 7.1 (Good), long. Filtered for one signal, MACD Bullish. But HUMA had already fallen 20% that day. It never got more than about 1.5% higher, and the next day it traded 6.5% lower. Oversold signals on a coin in free fall are a bounce bet, not a trend.
HYPE, 6.4 (Fair), short. Sent to the Trade Agent: entry 78.45 on the break, stop 79.15, target 77.10, with a heads up that the higher timeframe trend was against the short. The entry triggered, price dipped to 77.85, then turned, and about two hours later it hit the stop.
A handful of setups proves nothing about a win rate, which is exactly why the score isn't one. But the two that lost had something in common: lower scores, and a bigger trend pointing the other way. On HYPE, the app said so.
- Filter for two or more signal types.
- Start at the top.
- Read why.
- Check the bigger trend.
- Plan it with a stop before you take it.
Questions
This guide is for education only and is not financial advice. AI analysis can be wrong, and in a fast market price can move past a stop. Trading involves substantial risk of loss. Review every setup before you trade.
Scan every chart at once
Crypto and US stocks, every timeframe, one list. Free to start.
Open the App