Giordano Salvatori, founder of STA · 31/10/2026 · 3 min read
A trading signal is a trade proposal: which instrument, in which direction, at what price to enter, where to put the stop and where to aim the target. In AI signals it is produced by a model, or a group of models, that reads prices, volumes, derivatives and news faster and more methodically than a person can. It is not a forecast of the future, it is an analysis with numbers written beforehand. For the picture on where crypto signals come from, read reliable crypto trading signals.
The honest answer is that it depends, and that nobody can guarantee you a result. A model can read a lot of data, but markets change and a well built idea can still close at a loss. Anyone who shows you only the signals that went well is doing marketing, not statistics. To form a view you need the whole history, with the losses, and you need to understand how much a loss weighs against a gain.
For the technical side of AI applied to markets, what it does well and where it gets things wrong, see the guide to trading with artificial intelligence.
A good signal gives you at least four numbers, and they must all be read together.
Before entering, also decide how much to risk. The basic rule, explained in the guide to risk management in trading, is to start from a small percentage of your capital on each trade.
Five simple checks you can do in ten minutes. Is the history public and complete, losses included? Are the signals dated so they cannot be changed afterwards? Does every signal have a written stop? How much does the service cost, and when do you pay? Do your funds stay in your account or do you hand them to someone? A service that does not answer these questions does not deserve your trust, whatever it promises.
With a demo account, which uses real prices and fake money. Follow the signals for a few weeks and note what happens, including the days when you skip a trade out of fear or haste. We explain how to use it well in the guide to the demo trading account. The general picture, from the basics to the costs, is in the guide to mindful online trading.
STA's cards are signals of this kind, for crypto and US stocks on OKX. They are built by ten artificial intelligence agents, with a debate between a Bull and a Bear, who analyse 175 markets on OKX Europe, including 44 US stocks and ETFs, twice a day. Every card has its own stop and the funds stay in your OKX account. The history, losses included, is public on GitHub, with hashes timestamped every day with OpenTimestamps. This is general information, not personal advice.
The DEMO is free forever, with real prices and the same cards as REAL. An account is required. During the launch phase, which began on 16 October 2026, STA's fees are temporarily disabled, even in REAL; the ordinary price list, suspended while the launch lasts, is 18% of the net positive result of each day, with no offsetting of losses.
Every card as soon as it closes, with the result and what STA learned from it. It's free and you don't need to sign up.
Follow the channelA DEMO account with virtual capital and real prices, the same cards as REAL users.
Try DEMO for free →This article is for information only and is not investment advice. Trading crypto-assets involves a high risk, including the total loss of the capital used. Numerical examples are for illustration.
Funding rate: what it is, who pays whom in perpetual futures, how it is calculated with an example and what it says about crowded longs and shorts.
Trailing stop: what it is, how to set the distance and when to use it instead of a fixed stop loss, with an example and the limits to know.
Trading psychology: the 5 emotional mistakes that cost the most (FOMO, loss aversion, overtrading) and concrete rules to keep them in check.
