Short answer: yes — that's actually one of the most common ways AI is used in trading. But there's an important distinction between what these systems do and what they are.
What's real and well-established:
- AI models can ingest many indicators at once — moving averages, RSI, MACD, Bollinger Bands, volume profiles, funding rates, order-book imbalance, even sentiment from news/social feeds — and weight them far more efficiently than a human staring at multiple chart panels.
- This is the core of machine-learning / quantitative models. Instead of hard-coding "buy when RSI < 30 and MACD crosses up," the model is trained on historical data to find the combination of features that best predicts short-term direction. That's a genuine advantage: it can surface nonlinear relationships (e.g. "RSI overbought is bullish in strong uptrends") that a fixed rulebook would miss.
- Hedge funds and market makers have done this for decades. On the retail side, signal bots and "AI advisors" that blend indicators are common.
The honest caveats (this is where it gets less magical):
- The core problem isn't combining indicators — it's signal. All of these indicators are derived from the same price history. Combining 20 of them doesn't create new information; it mostly reweights the same information. In efficient-ish markets, and especially in crypto where retail order flow dominates, past patterns don't reliably predict the future.
- Overfitting is the classic failure mode. A model can be tuned so well to past data that it's useless on new data. Many "97% win rate backtest" bots are exactly this — they fit the noise, not the signal.
- They can't predict regime shifts. A model trained on a bull market will behave badly in a bear market or a volatility shock. Correlations (including to macro events) change.
- Indicator-based AI is still technical analysis. It tells you what has happened in the pattern data — not the future. No model combination turns past prices into a guarantee.
How you can use it practically: as a decision support tool, not an oracle — a way to filter entries, manage position size, or surface conditions you'd otherwise miss, combined with your own risk management (position sizing, stops). Treat any claimed backtest with heavy skepticism unless you've seen out-of-sample and live results.
Since this is general background knowledge, I can't point to a verified live number or model for this — but if you want something concrete to look at, I can pull up today's trending coins, a specific coin's current price/indicators context, or the top perp DEXs if you want a place where you could test such a strategy yourself.