Someone will show you a Bitsgap COMBO demo printing 40% a month and ask whether they should put $5,000 in. The honest answer needs you to understand one thing about demo modes, and it isn't that they're dishonest. It's that they're solving an easier problem than the market is.
This review works through the COMBO bot's mechanics, Bitsgap's published pricing, and where demo and backtest numbers structurally diverge from live results. It isn't a funded test, and the numbers below are worked examples rather than outcomes.
What the COMBO bot actually does
Bitsgap calls it a COMBO because it stacks two strategies: a grid layer catching oscillations, and a trend-following layer that adjusts the grid's midpoint as price moves. That targets the single biggest weakness of a plain grid bot, which dies when price runs out of range and simply stops trading. The COMBO shifts its position as a trend develops instead of going dormant.
You can run it on spot or futures. Most people choose futures, because leverage turns the grid's small per-trade returns into something that looks worth the subscription.
Speaking of which: the COMBO on futures needs Bitsgap's Advanced plan at $55/month billed annually, or $69/month paying monthly. The Basic plan at $23/month doesn't include futures bots at all, so if the COMBO is why you're here, $55/month is your floor. There's a 7-day free trial on the Pro plan with no card required, and the Bitsgap pricing page lays the tiers out clearly.
Why the demo number won't be your number
Here's the question worth asking before trusting any demo result: what does the simulator assume about your orders that reality doesn't?
Demo mode doesn't simulate slippage, execution latency, or funding. It assumes every limit order fills at exactly the limit price, the moment price touches it. In live trading, orders sitting at range boundaries are exactly the ones most likely to be partially filled or skipped entirely, because those boundaries are where price is moving fastest when it reaches them. A grid strategy's income depends on fill count, so a model that grants every fill is not modelling the strategy, it's modelling the best case.
This isn't a Bitsgap-specific flaw, every demo and backtest engine has some version of it. But the COMBO's design generates a high volume of small wins, and a high-volume small-win strategy is precisely the kind whose results are most distorted by an idealized fill model. The errors compound across hundreds of trades instead of averaging out.
Demo mode is for learning the controls, not for predicting returns. Use it to understand behavior, then set expectations well below what it showed you.
The backtester has the same property with an extra wrinkle: you can test up to 180 days of history on Advanced, which is genuinely useful for a different question. Not "how much will I make," but "does this asset actually range, or have I picked something that trends?" That's the first filter before running any grid-style bot, and the backtester answers it well even though it can't answer the profit question honestly.
Where it struggles
Funding is the hidden cost on any futures bot, and Bitsgap doesn't put it in front of you. Funding on a major pair commonly runs around 0.02% every 8 hours, which is roughly 2% a year of drag on your notional, and it spikes well above that in bullish stretches. On days when funding runs 0.06% to 0.08% per interval, a grid can easily pay more in funding than it earns in spread. The platform reports it, but you have to open the fee breakdown to find it, and beginners won't.
The COMBO also doesn't auto-extend its range on a breakout. You stop the bot, reset the range, restart. Some competitors automate that. (The manual step is honestly fine if you check positions daily, and only a problem if you wanted something passive, which a leveraged grid should never be anyway.)
Then there's the subscription against your account size, which is the calculation that actually decides whether this tool makes sense. At $55/month on $300 of capital, you're carrying 220% annualized overhead before a single trade. On $1,500 that's 44%. On $10,000 it's 6.6%. Below roughly $5,000, the subscription is a bigger drag on your returns than any setting you could tune, and no amount of grid optimization fixes a fee structure that large relative to capital.
What the platform gets right
The execution interface is genuinely well-designed, setup runs about twelve minutes including the help docs, and the AI Assistant for parameter suggestions works as a starting point provided you don't treat its output as optimal. Bitsgap connects to 17 exchanges, and the IP whitelisting step is clearly documented, which is more than several competitors manage. The mobile app is basic but functional.
None of that is nothing. A bot platform's job is to be reliable plumbing, and on plumbing Bitsgap does well. The reservations in this review are about the gap between what its demo promises and what its fee structure permits, not about build quality.
Who should actually use the COMBO bot
If you already trade on Bitsgap and want to add leverage to a grid strategy on an account where $55/month is a rounding error, the COMBO is a decent tool with a real answer to the range-exit problem that plain grids ignore.
If you're new to bots, start with a spot grid on a major pair, learn what ranging actually looks like and what happens when price breaks out, and come back to futures later. And don't run the COMBO on a volatile alt at 10x because a backtest looked good. ETH at 5x is already unpleasant on an 8% down day, and the demo that convinced you won't have modelled the fills you'd actually get on that day.
For a comparison against a platform with deeper DCA features and lower entry cost, see our 3Commas vs Bitsgap breakdown. If you're unsure which bot type fits, the bot match quiz gives a specific recommendation in about five minutes.
What this post can't tell you is how the COMBO's trend layer behaves through a real regime change, or how far live fills actually diverge from demo on a given pair. That needs a funded account over a meaningful window. What I can tell you is that the divergence exists by construction, and that anyone quoting you a demo figure is quoting the ceiling.
