Short answer: A grid bot profits from price oscillating inside a range you define, and loses ground when price trends out of it. A DCA bot buys in steps as price falls and sells on a recovery, and struggles in a sustained downtrend. Pick based on the market regime, and test the settings in demo mode before using real funds.
What a grid bot actually does
A grid bot places buy and sell orders at fixed price steps inside a range you choose. Each time price drops one step and recovers, a buy-sell pair completes and captures roughly the step size minus fees. The profit comes from oscillation, not from predicting direction.
That also means the bot has no opinion about trend. It simply keeps cycling as long as price stays inside the range, and every extra grid level multiplies the number of fills and therefore the fees paid.
Where grids break
When price drops below the bottom of the range, the bot is left holding the coins it bought at the higher levels, so it carries an unrealized loss until price returns. When price rises above the top, every sell has executed and the bot sits idle while the market keeps going.
A grid that is too narrow gets run over quickly. A grid that is too wide spreads capital across many levels and earns little per cycle. Choosing the range is the real decision, and it is a market-regime call, not a setting you can optimize in isolation.
What a DCA bot does differently
A DCA bot opens a position and adds safety orders as price falls, which lowers the average entry. It then closes the whole position when price recovers a target percentage above that average. It works well when price mean-reverts after a dip.
Its weakness is a sustained downtrend: each safety order commits more capital, and once they are exhausted the bot is fully invested into a falling market. Order size, step spacing and the number of safety orders define how much drawdown it can absorb.
Match the bot to the regime, then verify
Before running either bot live, decide whether the market you are trading is ranging or trending and write that assumption down. Then compare the bot’s trade history against the assumption: if a grid keeps accumulating inventory on the way down, the regime was not a range.
Bots automate execution, not judgement. Treat any settings you did not stress-test as a hypothesis, not a strategy.