Natural Language Trading: Describe a Strategy in Plain English
Natural language trading lets you write a trading strategy in plain English and have AI convert it into executable code, backtest it on historical data, and generate live signals — no programming required. On platforms like Hey-Traders, you type an idea, review the metrics, and can enable automated execution on connected venues once you grant trading permissions.
What is natural language trading?
Natural language trading is a method where you describe a strategy in everyday language — entries, exits, position size, and risk rules — and an AI system translates that description into functioning trading logic. It removes the coding barrier that historically separated retail traders from systematic, rules-based execution.
Traditional algorithmic trading requires Python, an exchange API, and infrastructure to run it. Natural language trading collapses that into a single step: you write the rule, the AI writes the code. This is closely related to broader AI trading and no-code trading bot approaches, but the defining feature here is the plain-English input.
How does natural language trading work?
It works in four stages: you describe the strategy, the AI parses and converts it into executable code, the code is backtested against historical data with professional metrics, and — if you choose — it runs live and executes orders on connected venues. Each stage keeps you in control.
Here is the typical flow on Hey-Traders:
- Describe — Write your rules in plain English (for example, entry conditions, stop-loss, take-profit, and sizing).
- Convert — The AI interprets the intent and generates executable strategy code, mapping your words to concrete order logic and parameters.
- Backtest — The strategy runs against historical price data, returning metrics such as Sharpe ratio, maximum drawdown (MDD), win rate, and an equity curve. See how to backtest a trading strategy for what those tests can and cannot tell you.
- Deploy — Once you are satisfied, the system produces live signals and can place orders automatically on supported venues after you enable trading permissions.
Because the whole loop is transparent, you can refine wording, re-run the backtest, and iterate quickly before risking any capital.
What does a natural language trading prompt look like?
A good prompt states the instrument, the entry trigger, the exit rules, and risk controls in clear, unambiguous terms. The more specific your conditions and numbers, the more precisely the AI can translate them into order logic.
Example prompts you might type:
- “Buy Bitcoin when the 50-period moving average crosses above the 200-period moving average. Set a 5% stop-loss and take profit at 15%.”
- “On ETH perpetual futures, open a long when RSI drops below 30, and close when RSI rises above 60. Use a trailing stop with a 3% delta.”
- “Run a grid strategy on SOL between $120 and $160 with 20 levels, reduce-only on exits.”
- “On this Polymarket market, take profit when my position gains 20% and stop out at a 10% loss.”
Each of these maps to concrete order types — market, limit, stop, trailing stop, or grid — that the platform can execute. You can review the full set in the order types documentation.
How is it different from traditional algo trading?
The core difference is the input layer: natural language trading replaces code with plain English, while keeping the same rigorous backtesting and execution underneath. It lowers the barrier to entry without lowering the analytical standard.
| Aspect | Traditional algo trading | Natural language trading |
|---|---|---|
| Input | Python / API code | Plain English description |
| Skill required | Programming + market knowledge | Market knowledge |
| Time to first backtest | Hours to days | Minutes |
| Iteration speed | Slow (edit and redeploy code) | Fast (reword and re-run) |
| Metrics | Sharpe, MDD, win rate | Sharpe, MDD, win rate |
| Execution | Self-built infrastructure | Connected venues via permissions |
If you are new to the underlying concepts, our primer on what is algo trading is a useful companion read.
What order types and venues does it support?
Natural language descriptions can be translated into a wide range of order types and executed on supported crypto exchanges and prediction markets, always subject to your permissions and venue availability. This gives your plain-English rules real precision.
Hey-Traders supports order types including market, limit, stop market, stop limit, take profit market, take profit limit, trailing stop (activation price plus trailing delta), OCO, grid, TWAP, VWAP, and conditional orders. Connected and supported venues can include major crypto exchanges and prediction markets such as Binance, Bybit, OKX, Hyperliquid, and Polymarket — availability depends on venue support and the permissions you grant.
One important nuance: a trigger price is a threshold, not a guaranteed fill. When a stop or take-profit triggers a market order, it executes against available liquidity, so the actual fill price can differ from the trigger — especially in fast-moving markets. A Reduce Only flag ensures an order can only close or reduce an existing position, never open a new one. For prediction-market specifics, see our guides on Polymarket stop-loss and Polymarket take-profit.
How do you read the backtest results?
Focus on the combination of metrics rather than any single number: the equity curve shows the overall trajectory, the Sharpe ratio measures risk-adjusted return, maximum drawdown shows worst-case pain, and win rate shows consistency. Together they describe whether a strategy is robust or fragile.
- Equity curve — the account balance over the test period; look for steady growth versus sharp cliffs.
- Sharpe ratio — return relative to volatility; higher generally indicates better risk-adjusted performance.
- Maximum drawdown (MDD) — the largest peak-to-trough decline; a proxy for how much stress a strategy could put you through.
- Win rate — the share of profitable trades, best read alongside average win versus average loss.
Our deep dive on Sharpe ratio, max drawdown, and win rate explains how to weigh them together. Remember: backtested results describe the past and do not guarantee future performance.
A brief, honest risk note
Trading involves a real risk of loss. Leverage — common on perpetual futures — amplifies both gains and losses, and a strong backtest is not a promise of live results because markets change. Natural language trading makes strategies easier to build and test, but it does not remove market risk. Start small, size positions responsibly, and treat every backtest as a hypothesis rather than a certainty.
Frequently Asked Questions
Do I need to know how to code to use natural language trading?
No. You describe your strategy in plain English, and the AI generates the executable code, runs the backtest, and can handle execution. Market knowledge helps more than programming skill.
Does natural language trading place trades automatically?
It can, but only after you explicitly enable trading permissions on a connected venue. Until then, Hey-Traders can generate signals and backtests without executing any orders on your behalf.
Are backtested results a guarantee of future profit?
No. Backtests describe how a strategy would have performed on historical data. Market conditions change, and past performance does not guarantee future results. Treat backtests as one input, not a promise.
What markets and order types can I use?
You can express strategies using order types such as market, limit, stop, trailing stop, OCO, grid, TWAP, VWAP, and conditional orders, executed on supported venues like major crypto exchanges and prediction markets — subject to venue support and your permissions.
Is a trigger price the same as my fill price?
No. A trigger price is a threshold that activates an order. When it fires a market order, the fill executes against available liquidity and can differ from the trigger, particularly during volatile conditions.
Ready to turn a plain-English idea into a tested, executable strategy? Explore Hey-Traders to describe your first strategy, review the backtest metrics, and decide for yourself — at your own pace and within your own risk tolerance.