What a crypto bot actually does
A crypto bot is not a robot with opinions. It is a set of rules running on a server, connected to your exchange account, placing orders while you sleep. You choose the strategy, the pair, and the limits; the bot does the repetitive clicking. That is the whole magic trick, and it is useful — as long as you understand what the rules do when the market stops cooperating.
There are three strategies you will meet on almost every platform, and they are worth understanding before you compare any product, because the product is just packaging for one of these ideas.
Grid: buy low, sell high, on repeat
A grid bot draws a ladder of prices across a range you choose — say, every $100 between $60,000 and $70,000 on Bitcoin — and places a small buy order at each rung. When the price falls to a rung, it buys. When it rises back to the next rung, it sells. Rinse, repeat, forever, as long as the price stays inside your ladder.
Grids love sideways markets. They are miserable in trends. If the price breaks out above your ladder, your bot sits fully sold and watches the rally leave without it. If the price falls through the bottom rung, your bot ends up holding a full position at a loss, waiting for a recovery that may take months. Every grid bot is, at heart, a bet that the price will keep coming back.
DCA: buy the dips on a schedule
DCA (dollar-cost averaging) bots buy a fixed amount at regular intervals, or buy more each time the price drops a set percentage. The idea is simple: instead of guessing the bottom, you average your entry price across many buys. It is the least dramatic strategy on the shelf, which is a compliment. It also has a quiet catch: in a long downtrend, a DCA bot keeps buying all the way down, so your average price keeps improving while your total loss keeps growing. DCA controls *when* you buy, not *whether* buying was a good idea.
Combo: grid plus DCA, with extra moving parts
Combo bots run a grid on top of a DCA base — they accumulate a position and then grid-trade around it. They sound like the best of both worlds, and in a choppy market they often are. But every extra moving part is another assumption to tune, and another way for the whole thing to behave differently from what you pictured. Complexity is not a feature. It is a cost you pay in understanding.
The fees that eat them
Here is the part the marketing pages whisper. A grid bot that trades fifty times a day on a 0.1% fee is not paying "just 0.1%" — it is paying it fifty times, on every round trip, out of your profit. Bots that look brilliant in a backtest and mediocre live are often not broken; they are just fee-blind. Before you switch anything on, work out the fee per trade on your exchange, multiply by the bot's expected trade count, and subtract it from the advertised picture. If a platform does not let you estimate that easily, that is information too.
Fees also come in a second flavour: the platform's own subscription. The shelf shows the honest spread of what that costs.
What the shelf actually offers
On the Crypto Bot Platforms shelf, the entry points run from free to roughly $30 a month, and the shape of each product tells you who it is for:
- Pionex is an exchange with 16 built-in bots and no platform subscription — its revenue is a 0.05% trading fee. The bots are free; the trades are not, which makes the fee maths above doubly important.
- Bitsgap runs grid, DCA and COMBO bots across 15+ exchanges, listed from $23/mo on annual billing. One dashboard, many exchanges — useful if you do not want your bots scattered across five logins.
- Gainium is a cloud platform for grid, DCA and combo bots with a free-forever plan and paid tiers listed from $15. A free tier is the right way to learn whether you even like babysitting a bot.
- 3Commas is the long-running subscription platform, with an entry tier listed at $20 in our inventory. Its strength is breadth: many bot types, many exchanges, and a large community writing about it.
- Cryptohopper offers a free Pioneer tier with paid plans listed from around $24/mo on annual billing, and leans on templates and copy-trading for beginners.
- Gunbot is the opposite shape: a self-hosted bot with a one-time licence listed from $59. You run it on your own machine, which means you own the uptime problem — and the privacy upside.
- Hummingbot is free and open-source, aimed squarely at developers who want to build market-making strategies rather than click a template.
None of these is "the best". The right question is narrower: which strategy do you actually understand, on which exchange, at what fee, with how much babysitting?
The honest checklist
Before any bot touches real money, answer these in writing:
- What market does this strategy need? Grid needs range. DCA needs eventual recovery. If you cannot name the market your bot needs, you do not have a strategy yet.
- What happens outside the range? Every grid has a bottom rung. Know yours, and know what you will do when price crosses it — before it happens, not during.
- What do fees do to the maths? Multiply fee by expected trades. If the answer surprises you, the bot was never as good as it looked.
- What is the worst week? Not the average week. The worst one. Size the bot so the worst week is annoying, not fatal.
- Can you turn it off? A bot you cannot stop cleanly, with open orders you cannot see, is not automation. It is a liability.
Stackbot's house rule: a bot is a tool for executing a decision you already understand — never a substitute for making one.
Start boring, on purpose
The unglamorous path works: pick one strategy, one pair, a small amount, and a free or cheap tier. Run it for a month. Watch what it does on a bad day, not a good one. Read the fee statement at the end of the month. Then — and only then — decide whether to scale it, change it, or switch it off. The traders who last are not the ones with the cleverest bot. They are the ones who still understand it in month six.