Freqtrade Setup Guide 2026: The Free Bot That Actually Works
Freqtrade is the most-used open-source crypto trading bot — free, self-hosted, Python-based, and battle-tested since 2018. Zero subscription, full control. The catch: you run it, you configure it, you own the results.
What You Need Before Starting
- A VPS — 2 vCPU/4GB RAM minimum (~$6-10/mo), Linux Ubuntu recommended. Never run it on your laptop: VPS setup guide
- Exchange account — Binance, Kraken, Bybit, OKX, KuCoin via CCXT
- Python 3.10+ and basic terminal comfort
- API keys — trade permission only, withdrawal OFF: API security
Step-by-Step Setup
Start Automating Your Crypto Profits Today
Our top all-rounder pick — DCA, grid, signal bots and a strategy marketplace. Free trial available.
Start Free Trial
git clone https://github.com/freqtrade/freqtrade.git
cd freqtrade
docker compose pull
docker compose run --rm freqtrade create-userdir --userdir user_data
docker compose run --rm freqtrade new-config --config user_data/config.json
2. Configure — edit config.json: exchange, API keys, whitelist pairs (BTC/USDT, ETH/USDT...), stake currency, max_open_trades, and stake_amount.
3. Pick a strategy — strategies live in user_data/strategies as Python classes. Start with a bundled sample (SampleStrategy) or community strategies like BinHV45. Define entries via indicators (RSI, BB, MACD) and exits via ROI table + stoploss.
4. Backtest — mandatory before any live run:
docker compose run --rm freqtrade backtesting --strategy MyStrategy --timerange 20250101-
5. Dry run — dry_run: true trades with fake money for 2+ weeks. Then go live with minimum stake.
> No-code route instead? Freqtrade demands Python comfort. For a point-and-click bot running in 10 minutes, Cryptohopper's free trial skips the terminal entirely.
FreqAI — The ML Layer
Freqtrade's machine-learning module trains prediction models on your pair data — feature engineering, adaptive retraining, confidence-based entries. Powerful but a real time sink; expect days of tuning, not hours.
Freqtrade vs Cloud Bots
| Freqtrade | Cloud bots (3Commas/Cryptohopper) | |
|---|---|---|
| Cost | Free (+ VPS) | $20-100/mo |
| Setup | Hours-days, technical | 10 minutes |
| Strategies | Unlimited, Python | Visual/marketplace |
| Backtesting | Excellent, free | Tier-limited |
| Support | Community Discord | Official support |
Full landscape: open source vs cloud · self-hosted vs built-in hosting.
Honest Limits
- No free lunch — bundled strategies are demos, not profit machines; you'll write or tune your own
- Ops burden — updates, monitoring, logs, VPS uptime are on you
- Steep curve — non-programmers hit walls fast (Telegram control helps:
/profit,/status)
FAQ
Is Freqtrade profitable?The tool works — profitability is your strategy's job. Backtest honestly, dry-run, start tiny. Most retail strategies fail; Freqtrade executes faithfully either way.
Is Freqtrade safe?Open-source code you can audit, keys stay on your VPS. Safer than cloud custody — if your VPS is secured.
Freqtrade vs Hummingbot?Freqtrade = strategy trading bot. Hummingbot = market-making/arbitrage toolkit. Different jobs.
Bottom Line
Freqtrade is the best free path to real automated trading — if you can handle Python and a VPS. Budget days for setup, weeks for strategy validation. Prefer clicks over code? Run a cloud bot free and trade the same strategies in a tenth of the time.