Why my take profit is 0.3%, not 0.5% — and why my bot has no tight stop
A reader asked: did I backtest a 0.4–0.6% take profit, and did I try a stop instead of sitting in a lock? Yes — both.
Take profit. I ran take profits from 0.2% to 0.79% on two years of history. Intuition says: bigger take profit, bigger profit per deal. But the bigger the take profit, the longer the bot sits in a position — and the more time price has to fill the whole grid and keep falling. Above ~0.4% results only got worse, and from 0.5% with regular fees some stretches of history ended in liquidations. 0.3% is tuned for the zero maker fee on USDC pairs.
Stop. The tighter the stop inside the grid, the smaller but more frequent the losses. The further away, the rarer but heavier — growing with every filled order. A tight stop simply doesn’t let the grid work: over 30 days on $DEXE , a stop at −1% from the average fired 2,090 times versus 43 for a wide stop, and the result was −$347 versus +$544. The whole point of DCA is to let the grid work. If you want short stops, trade single orders.
So my bot has one stop, and it’s deep — below the whole grid, about −30% from entry: that’s the liquidation of isolated margin. Plus a safety net: exit if a lock lasts longer than 30 days. Idle time has a price too: while the deposit is stuck in one coin, it isn’t earning on others.
Questions in the comments are the best topics for this diary. Thanks for this one.
#DCA #TradingBot #GridTrading
Not financial advice.
A reader asked: did I backtest a 0.4–0.6% take profit, and did I try a stop instead of sitting in a lock? Yes — both.
Take profit. I ran take profits from 0.2% to 0.79% on two years of history. Intuition says: bigger take profit, bigger profit per deal. But the bigger the take profit, the longer the bot sits in a position — and the more time price has to fill the whole grid and keep falling. Above ~0.4% results only got worse, and from 0.5% with regular fees some stretches of history ended in liquidations. 0.3% is tuned for the zero maker fee on USDC pairs.
Stop. The tighter the stop inside the grid, the smaller but more frequent the losses. The further away, the rarer but heavier — growing with every filled order. A tight stop simply doesn’t let the grid work: over 30 days on $DEXE , a stop at −1% from the average fired 2,090 times versus 43 for a wide stop, and the result was −$347 versus +$544. The whole point of DCA is to let the grid work. If you want short stops, trade single orders.
So my bot has one stop, and it’s deep — below the whole grid, about −30% from entry: that’s the liquidation of isolated margin. Plus a safety net: exit if a lock lasts longer than 30 days. Idle time has a price too: while the deposit is stuck in one coin, it isn’t earning on others.
Questions in the comments are the best topics for this diary. Thanks for this one.
#DCA #TradingBot #GridTrading
Not financial advice.