15 AI Trading Bot Prompts You Can Copy-Paste (No Coding, 2026)

Last updated: May 2026. All prompts tested against the live AI Trading Agent generator. Strategy descriptions are educational — do not deploy with capital you cannot afford to lose.

The hard part of running a trading bot used to be writing the code — translating an idea like 'buy when RSI is oversold but only in an uptrend' into Python, exchange APIs, indicator math, and persistent state across restarts. AI Trading Agent removes that step entirely. You describe the strategy in two to four sentences of English, an LLM classifies the request as safe and on-topic, a second LLM generates a working Python script that runs in a hardened sandbox, and the bot is live on your account in about sixty seconds.

The friction point most users hit isn't the technology — it's writing a good prompt. Too vague ("buy low, sell high") and the AI guesses at parameters; too detailed ("on the 1-minute chart, buy when the 9-EMA crosses the 21-EMA but only if Williams %R is below -80 and the MACD histogram is positive for at least 3 candles unless the daily ATR is above 2.5%") and the strategy is over-fit before it ever sees a real candle. This post is a copy-paste library of fifteen prompts that hit the right level — specific enough to generate a working bot, simple enough to actually edge in live markets.

Each prompt below has been run through the live AI Trading Agent generator on fomoed and produces a deploy-ready bot. The strategies span beginner (single-indicator entries, no exotic state), intermediate (multi-condition entries with trend filters), and advanced (Smart Money Concepts, multi-timeframe confluence, funding-rate plays). Pick one, paste it into the wizard, run it on paper trading first, and only graduate to real capital after you've seen at least a few weeks of live behaviour.

How AI Trading Agent generates a bot from your description

The pipeline behind AI Trading Agent is a two-stage process followed by safety checks. The first stage reads your prompt and labels it: on-topic / off-topic / unsafe. Off-topic prompts ("write me a love poem") and unsafe ones ("send all my funds to wallet X") are rejected before any code is generated, which keeps costs low and the attack surface tiny. The second stage uses a heavier code generator to translate the on-topic prompt into a Python script that composes against fomoed's AI trading agent helper library — pre-tested functions for indicators, order placement, position state, and SMC primitives.

The generated bot is then reviewed for safety before it can run, and only runs inside a secure isolated environment that has no access to your funds, your account, or anything outside the trading engine. The bot can compute indicators, read price feeds, decide entries and exits, and submit orders — and nothing else. If anything goes wrong, the bot stops and your account doesn't move.

The full architecture is covered in the pillar post How to Build a Trading Bot with AI in Plain English. If you want to write SMC-specific bots, the companion guide Smart Money Concepts with AI covers the seven SMC primitives the AI can call directly. Both posts go deeper on the safety model and the helper library; this post focuses on the prompts themselves.

How to use these prompts in 60 seconds

1. Sign up free at fomoed (email or wallet — no KYC, no card).
2. Click New Bot, choose AI Trading Agent as the strategy.
3. Paste any prompt from below into the description field.
4. Pick a pair (BTC, ETH, SOL, or anything supported on Hyperliquid / Binance / Bybit / OKX / Decibel).
5. Set paper-trading mode for the first run.
6. Click Generate. The bot script lands in the review step. Click Deploy.

The wizard shows you the generated Python code before you commit. You can edit it inline if you want to tweak a parameter, or regenerate with a slightly different prompt if the AI's interpretation differs from yours. Once deployed, the bot runs 24/7 — you'll see live decisions in the bot detail view.

5 beginner prompts — single-indicator entries

Buy $50 of BTC every Monday at 9:00 UTC. No exit — just accumulate. If price drops more than 5% in a week, double the next buy.

On the 1-hour chart, buy when RSI(14) drops below 30. Exit at +3% or when RSI crosses back above 65. 1.5% stop loss.

On the 4-hour chart, go long when EMA(20) crosses above EMA(50). Exit when EMA(20) crosses back below, or +5% target hit. 2% stop loss.

Place a 10-step grid between $60,000 and $70,000 on BTC. Each step is 1% apart. Take profit on each grid cell when price moves up 0.5% from the entry of that cell.

On the 15-minute chart, go long when price breaks above the previous 24-hour high AND volume on the breakout bar is at least 2x the 20-bar average. Exit at +4% or 1% stop loss.

5 intermediate prompts — multi-condition entries

On the 1-hour chart, buy when RSI(14) drops below 35 AND price is above the 200 EMA. Exit at +3% or when RSI crosses above 70. 1.5% stop loss.

On the 4-hour chart, go long when the MACD line crosses above the signal line AND RSI(14) is between 40 and 60. Exit when MACD crosses below the signal line or at +5% profit. 2% stop loss.

On the 1-hour chart, when Bollinger Bands (20, 2) narrow to less than half their 20-period average width, watch for a breakout. Go long on a close above the upper band; go short on a close below the lower band. 1.5% stop loss, 4% take profit.

On the 5-minute chart, buy when price drops at least 1.5% below VWAP AND RSI(14) is under 40. Sell when price returns to VWAP or hits +1% profit. 0.8% stop loss.

On the 1-hour chart, go long when ROC(14) is greater than 5% AND price is above the 50 EMA. Exit when ROC drops below 0 or at +5% profit, whichever comes first. Trail the stop 2% behind the highest price reached.

5 advanced prompts — SMC, multi-timeframe, funding

On the 1-hour chart, go long when a bullish change of character (CHoCH) is detected AND price retraces into the 0.618 to 0.79 Fibonacci OTE zone of the displacement leg. Stop loss at the swing low; take profit at the previous swing high.

Go long on the 15-minute chart when the 1-hour EMA(20) is above the 1-hour EMA(50) AND the 15-minute RSI(14) crosses above 50 from below. Exit when RSI crosses above 75 or +4% profit, whichever first. 1.5% stop loss.

On Hyperliquid BTC perpetual, open a long position when the 8-hour funding rate is below -0.03% (shorts paying longs). Close the position when funding turns positive OR after 24 hours, whichever first. 3% stop loss as a safety net.

Only trade BTC between 13:00 and 17:00 UTC (US session). Open a long on a 1-hour bullish engulfing candle with volume above the 20-bar average. Close all open positions by 18:00 UTC regardless of P&L. 1% stop loss, 3% take profit.

On the 5-minute chart, buy when ALL THREE signals fire within 5 candles: RSI(14) crosses up through 30, MACD histogram turns positive, AND price reclaims the 50 EMA. Exit at +1.5% profit or 0.7% stop loss.

How to tweak a prompt that doesn't generate well

About one in twenty prompts produces a script that does something subtly different from what you intended — usually because the prompt is ambiguous on a single dimension. The fix is almost always to add one specific sentence. The most common adjustments:

Specify the timeframe. "Buy when RSI is oversold" is ambiguous — the AI picks a default (usually 1-hour) which may not match your intuition. Add "on the 15-minute chart" or "on the 4-hour" to lock it in.

Spell out exit conditions. If you only describe the entry, the AI defaults to a generic exit (often "close after 24 hours" or "exit at the next reverse signal"). Always include: take-profit % or condition, stop-loss %, and a max holding period if you care.

Use named indicators with parameters. "Use RSI" → "RSI(14)". "Use EMA" → "EMA(50)". Without the parameter, the AI has to guess, and guesses can differ across runs.

Be explicit about position size. The default is 100% of the bot's allocated capital per trade. If you want to scale in or risk less per signal, say so: "use 25% of available margin per entry, max 4 concurrent positions".

Separate the conditions with explicit logical operators. "Buy when RSI is below 30 and price is above the 200 EMA" is unambiguous. "Buy when oversold or trending up" reads naturally but the AI may interpret 'or' liberally — use AND and OR in caps when both conditions matter.

Common AI Trading Agent mistakes (and how to avoid them)

Conflicting rules. "Buy when RSI is oversold AND when price is breaking out" — these two conditions almost never fire together. The AI generates a bot that takes maybe one trade a year. Pick one entry trigger and use the other as a filter.

Asking for impossible features. The AI can't predict the future, can't read news headlines, can't access social media. Prompts like "buy when Bitcoin is about to pump" or "sell when Elon tweets" get politely rewritten into something the bot can actually execute — but the result rarely matches the intent. Stick to price action, indicators, time-of-day, and exchange-native data (funding rates, open interest, volume).

Skipping paper trading. The platform's safety checks make sure the bot can't do anything dangerous — but none of that catches a bad strategy. A bot that's safe to run is not the same as a bot that's profitable. Paper trade for at least two weeks before risking real capital, and double the paper-trade window if your strategy depends on a specific market regime.

Over-optimizing on backtest results. If a prompt produces a bot that backtests at 78% win rate over 12 months, the prompt is probably over-fit. Real-world live performance routinely runs at 60–70% of backtest performance because the backtest sees no slippage, no order rejections, and no regime changes. Discount any backtest by at least a third before deciding.

Frequently asked questions

Can I edit the bot after it's generated?

Yes. After the AI generates the Python script, you see it in the wizard review step. Edit any line, change parameters, regenerate from a tweaked prompt — the bot doesn't deploy until you click Deploy. Once deployed, you can still pause, reconfigure, and redeploy.

What if the AI rejects my prompt?

The AI rejects two categories: off-topic (anything not a trading strategy) and unsafe (instructions that would withdraw funds, share private keys, or exploit the platform). Rejected prompts show you the reason and a suggested rewrite. Most rejections are fixable in one edit.

Do these prompts work on every exchange?

Yes — every prompt in this list is exchange-agnostic and will run on Hyperliquid, Binance, Bybit, OKX, Decibel, AsterDex, GRVT, Extended, and StandX. The AI generates exchange-aware order code automatically based on which exchange you pick in step 1 of the wizard.

Can the AI hallucinate an indicator that doesn't exist?

No. The code generator is constrained to the fomoed helper library, which exposes a fixed set of indicators (RSI, EMA, SMA, MACD, ATR, ROC, Bollinger Bands, VWAP, Williams %R, Stochastic, OBV, ADX, plus the seven SMC primitives). If your prompt requests something outside that set, the AI either maps it to the closest equivalent or declines and explains why.

How long until the bot is actually running?

The full prompt-to-bot pipeline typically completes in 6–12 seconds. Review and deploy is one click. The bot is live on your account in about 60 seconds end-to-end, including signing up.

Do I need to know Python to read the generated bot?

No, but it helps. The script is heavily commented and uses descriptive variable names; a careful reader without Python experience can follow what each section does. If you want to skip the code entirely, just deploy and watch the bot's decision log in the bot detail view — every entry, exit, and skipped signal is explained in plain English.

What if a prompt generates an unsafe-looking strategy?

Strategy danger and code safety are different layers. Code safety is enforced — no script can withdraw funds, leak credentials, or execute outside the sandbox, regardless of the prompt. Strategy danger (over-leveraged entries, bad risk-reward, fragile in regime changes) is on you to evaluate. Paper trade first; the platform shows you every trade decision before you commit real capital.

Related Resources

• How to Build a Trading Bot with AI in Plain English (No Code)
• Smart Money Concepts with AI: Build a CHoCH, BoS, and Fib Bot
• Paper Trading: Test Your Strategy Risk-Free
• Best Crypto Bot Strategy 2026: A Comparison Guide
• AI Trading Bots: How They Work and What to Expect
• fomoed AI Trading Agent
• fomoed Custom Strategy Engine