Grid trading bots are among the most heavily marketed tools in crypto: beginner-friendly, "low-risk," and — with AI automation — supposedly a source of "consistent profits." KuCoin's Spot Grid bot is the best-known example, with millions of bots running. We looked at the claims — not to dunk on anyone, but because the mechanics say something different from the marketing in several places. That gap is worth knowing before you start your first bot.

What a grid bot actually does

A grid places buy and sell orders at fixed intervals across a price range that you — or the AI — define. When price touches a level on the way down, it buys; on the way up, it sells. Profit comes from oscillation within the range: many small buy-low-sell-high round trips.

The key sentence no marketing copy puts in bold: a grid earns on sideways oscillation, not on trends.

The edges the marketing stays quiet about

In a long spot grid you're maximally one-sided at the boundaries:

• At the upper bound you've sold off step by step — you're almost entirely in cash, with no inventory. If price keeps rising, you don't participate. Your capital sits idle.
• At the lower bound you've spent all your cash — you're holding the full coin position, the "bag." If price keeps falling, you carry the entire loss, unrealized.

That's not a defect — it's the nature of the strategy. But it contradicts "low-risk": at the lower edge you're 100% in a falling asset.

"The bot trades almost all the time" — what really happens

KuCoin's "AI Plus" promises to auto-adjust the range so the bot never sits idle and delivers "consistent profits." Mechanically, that only works by trading against its own prior fills:

• If price breaks up, you're in cash. To keep trading, the bot has to buy inventory back — at higher prices than it just sold. Re-entering higher after exiting lower.
• If price breaks down, you're fully in the asset. To keep trading, the bot has to sell part of the bag — at lower prices than it bought. Exiting lower to buy even lower.

Both are trend-chasing. The "always active" feature doesn't create profit — it inverts the clean mean-reversion logic and, in a trend, eats exactly the spread the grid was meant to capture.

The fee trap

Tighter spacing means more round trips and therefore more fees. On stable pairs (USDT/USDC) the moves are so small that fees often eat the tiny spread. Any honest grid evaluation has to net fees — otherwise you're measuring gross trades, not net results.

"The AI sets your parameters"

The auto parameters are derived from historical price action — that is, backward-looking. They optimize for the regime that just happened. A grid fitted "optimally" to the last quiet sideways phase is exactly wrong in the next trend. Handy as a starting point, but no edge — and no guarantee.

"Consistent profits"? The same docs contradict it

The most telling part: the same guide that promises "consistent profits" elsewhere advises closing the bot in a clear downtrend to cut losses. Both can't be true at once. "Consistent profits" is a marketing promise, not a property of the strategy. A grid is a regime-dependent tool: strong in sideways markets, weak in trends.

How to actually evaluate a grid bot

Not by a pretty APR leaderboard — that shows one bot's lucky day, not the expectation. Instead:

• test across multiple regimes — sideways and up and down trends, not just the phase the settings came from
• out-of-sample: derive parameters on one window, trade them on the next
• always compare against buy-and-hold of the same asset
• net fees and report time at the edges (idle in cash, or bag-holding)

This is exactly why Backtesting Arena exists: not "trust the AI bot," but "here's the data across multiple regimes." A grid backtester that shows precisely that — performance per regime, drawdown, vs B&H — is on our roadmap.

The honest bottom line

A grid bot is not an ATM and not "low-risk." It's a good tool for sideways markets with a defined range — and a poor one for trends. "Consistent profits" and "the AI handles it" are promises the mechanics don't keep. Understand that, and a grid can be useful. Believe it, and the next trend charges tuition.

Not financial advice. All backtests are historical simulations — no guarantee of future results.