A Binance Futures bot doesn’t need more signals first. It needs better state recovery.
I found this in the logs last week: one signal fired, the order was accepted, a partial fill came back after 380ms, then the websocket reconnected before the local position cache updated.
The bot thought the account was still flat.
That’s where duplicate orders start.
My basic Binance Futures rule now:
1. Sync current position before any new order
2. Track open orders separately from filled orders
3. Add a cooldown after every fill
4. Rebuild account state after every reconnect
5. Log rejected signals, not only executed trades
I learned this the hard way on a small test. Lost 0.3% before I noticed. Annoying detail, expensive lesson.
Most builders obsess over entry accuracy. I’d rather fix the part where the bot forgets what it already did.
If your bot trades Binance Futures, do you log reconnects and rejected signals, or only filled trades?