What Are LLMs Good At?
Large language models are good at reading text, extracting points, and classifying information.
After earnings news, they can quickly answer:
- Did revenue beat expectations?
- Did management cut guidance?
- What risk words appeared?
- What are analysts focused on?
Useful, but not yet a trading signal.
Understanding News Does Not Mean Predicting Price
Markets react to the gap between result and expectation.
A company may grow profit by 30%, but if the market expected 50%, the stock can still fall.
If an LLM only reads the article and does not know expectations, positioning, liquidity, or options pricing, it may mistake “good news” for “buy signal.”
Missing Steps From News to Signal
| Step | What Is Needed |
|---|---|
| News summary | What happened |
| Expectation comparison | Was it above or below expectations |
| Price reaction | Is it already priced in |
| Cost calculation | Spread, slippage, fees |
| Risk rule | When to exit if wrong |
Beginner Mistake
Do not confuse “AI explains well” with “AI predicts well.”
Explanation and prediction are different. An LLM can explain yesterday clearly. Tomorrow’s price depends on market structure.
Check Before Using
- Did the model see price and volume?
- Did it compare against consensus or history?
- Is the output summary, sentiment, or a trade plan?
- Is maximum loss defined?
- Is signal failure tracked?
Quiz
Q1. Does news understanding equal direct trading ability?
A. Yes B. No
Q2. Good news can make a stock fall because:
A. It missed expectations B. The headline was short
Q3. A news signal needs:
A. Expectations, price, costs, risk control B. More adjectives
Answer Key
Q1: B Q2: A Q3: A
Further reading: IBM — Large Language Models · Investopedia — Earnings Surprise
For education only. News interpretation is not price prediction.
Get the pre-trade checklist.
We are turning these guides into a searchable checklist for checking terms, rules and risk before you trade.