COMMENT "BLEW"
The risk setup that stops a $20k-to-$0 blowup (Freqtrade)
A bot made $5k into ~$20k, then lost it all on one trade with no stop. The fix is not a better entry, it is risk control in code. Freqtrade gives you a stop on every trade, per-trade risk caps, and auto-shutdown protections so no single trade can wipe the account.
01Quick steps
- 1Freqtrade (freqtrade.io), free open-source bot, build it in 3 parts.
- 2Scanner: turn on a pairlist so the bot pulls active coins automatically.
- 3AI brain: use the built-in FreqAI module to score each trade before it buys.
- 4Hard rules: a stop on every trade, risk 1 to 2% per trade (stake_amount + max_open_trades), and StoplossGuard + MaxDrawdown protections on.
03Full tutorial
The bot was never the problem, the missing risk control was. It 4x'd, then one oversized trade with no stop gave it all back. In code, three things prevent that: a stop on every trade, a hard cap on risk per trade, and an auto-shutdown after a losing streak. Freqtrade has all three built in.
Configure a Freqtrade pairlist (e.g. VolumePairList) so the bot auto-selects active, liquid coins instead of you hand-picking. This keeps it trading names it can actually enter and exit.
Enable the built-in FreqAI module to train a model on your features and score each candidate trade before entry. It is the decision layer: only take trades above a confidence threshold. This is optional but it filters low-quality entries.
Set a stoploss in the strategy (e.g. -8% or an ATR-based stop) so every single position has a hard exit. No position without a stop, ever. This alone prevents the single-trade wipeout.
Set stake_amount and max_open_trades so each trade risks only 1 to 2% of the account and total exposure is bounded. Position sizing is what makes a losing streak survivable instead of fatal.
Turn on Freqtrade protections: StoplossGuard (pause trading after N stop-outs in a window) and MaxDrawdown (halt when drawdown exceeds a limit). Now a cold streak stops the bot automatically instead of letting it revenge-trade to zero.
Backtest and run in dry-run with these rules on, then check the worst drawdown. If the account survives a bad stretch on paper, the risk layer works. Only then go live, small.
Risk control caps the downside, it does not guarantee profit. Most people lose money trading. Only trade what you can afford to lose. Not financial advice.
A hard stop and a drawdown cap are the difference between $20k and $0. Most people lose money trading. Not financial advice.