A coin suits a grid bot when three things are true at once: its price swings back and forth inside a range, the orders you rest on it actually get filled, and the gap between two neighbouring grids still leaves something after you pay a fee on the buy and another on the sell. The help center backs the first one in so many words: "Binance Spot Grid Trading performs the best in volatile markets when prices fluctuate within a specific range." The third comes with a formula. At the default 0.1% fee, adjacent grids need to sit roughly 0.2% apart before a grid stops costing you money (fee checked 2026-10; the rate shown at the time you trade is what counts). The second has no official number at all, only consequences: thin markets can leave you holding coins when you stop the bot, and the worst cases get delisted.
No coin gets recommended here. What follows is each filter in turn, with the arithmetic laid out so you can run it on the pair you had in mind. Whether the next few weeks will chop or trend is a market call, which lives in why grid bots lose money and which markets suit them. This piece only asks which pair. Picking a coin for DCA is a different question again, about whether you would hold it for years; that one is answered in which coins to DCA into.
Volatility: does it move, and does it come back?
A grid earns when price crosses a level and then crosses back. A coin that barely moves leaves your grids untouched. A coin that moves hard in one direction is worse: going down, the bot keeps buying all the way; going up, it sells out of your position one level at a time. The first filter wants movement, and specifically the back-and-forth kind.
There is an official way to measure movement. Log in and click [Trade] - [Trading Bots]; the Trading Analytics tab can sort pairs by volatility. The help center spells out where that figure comes from: "The volatility % is calculated based on the symbol's hourly closing prices over the past 30 days," and the final step is "Calculate the annualized volatility." (The 30-day window is as published when we checked in 2026-10.)
Two things trip people up here. It is an annualized number, so a big percentage on the list does not mean the coin rose or fell that much this month. And a standard deviation measures how violently price moves, not which way. A coin that drops a little every single day can land near a coin that bounces up and down every day, and gridding the first one is catching a falling knife on a schedule. The volatility list can rule out pairs that barely move. Whether price comes back is something you still have to read off the chart yourself.
Some high-volatility coins should be crossed off outright: anything carrying the Monitoring Tag. The delisting guidelines say the tag appears "on the Binance Spot and Binance Margin trading pages, as well as on the Markets Overview page," with additions and removals "completed in the first week of every month" (as of our 2026-10 check). The Chinese version of the same FAQ states the purpose more bluntly: to help users spot tokens whose volatility and risk are clearly higher than other listed tokens. A grid needs volatility, but the kind the exchange itself has flagged as unusually risky tends to fail the second filter as well.
Liquidity: will anyone take the other side?
Underneath, a grid is just a row of limit orders, and every fill needs somebody on the other side. The bot does not get a private lane; its orders queue in the spot market alongside everyone else's.
The most concrete consequence of a thin market shows up when you stop the bot. The Chinese-language spot grid explainer notes that the option to sell all base coins on termination is on by default for new strategies, and adds under that option that if you hold a large amount of the base coin or market liquidity is poor, the system will protect your assets and not sell your base coin. In plain terms: even with that option on, the day you shut the grid down, the part you expected back in USDT may stay in your account as coins. Nobody notices that sentence while the bot runs smoothly. It matters on the day you want out.
One step further out is delisting. Among the stated reasons for removing a spot pair is "poor liquidity and trading volume." A pair that might be pulled is a bad home for a bot you plan to leave running for weeks.
How deep is deep enough? There is no published threshold, we did not find one we could cite, and we are not going to invent one. What you can do is open the pair's trading page before committing and look: are there orders sitting on both sides, and does the trade history tick over every few seconds? If you find yourself waiting a long while for a single print, set that pair aside.
Fees: can each grid step pay for two trades?
This filter you can work out to two decimal places. The help center's geometric grid formula:
Price ratio r = (upper limit / lower limit) ^ (1 / number of grids)
Profit/Grid = (1 - c) * r - 1 - c
Here c is the trading fee, which the help center sets at "0.1%" for the calculation. Its own worked example uses an upper limit of 450, a lower limit of 400 and 5 grids, which comes out at 2.18% per grid. The fee is subtracted twice because every grid is a round trip, one buy and one sell.
Set the profit to zero and solve backwards: r has to beat 1.001 / 0.999, which is about 1.002. In everyday terms, neighbouring grids need roughly 0.2% between them just to break even, and at a 0.3% gap you keep only about 0.1% after both fees. Plug in 0.075% for paying fees in BNB and the line falls to about 0.15% (fee details in how Binance trading fees work; checked 2026-10, the live page governs).
Two hypothetical coins make it concrete. Coin A trades in a tight 98 to 102 band, 4% wide in total. Coin B ranges from 80 to 120. Both are geometric grids at the default 0.1% fee. These numbers exist only to show the formula working; they are not the market data of any real pair.
| Hypothetical | Grids | Gap between grids | Profit per grid after two fees |
|---|---|---|---|
| A: 98 to 102 | 10 | about 0.40% | about 0.20% |
| A: 98 to 102 | 20 | about 0.20% | about 0.00% |
| A: 98 to 102 | 30 | about 0.13% | about -0.07% |
| B: 80 to 120 | 10 | about 4.14% | about 3.93% |
| B: 80 to 120 | 20 | about 2.05% | about 1.85% |
| B: 80 to 120 | 30 | about 1.36% | about 1.16% |
At 20 grids each, A nets roughly nothing per grid while B keeps around 1.85%. Working back from the 0.2% line, A's band tops out at about 20 grids; B could take something like 200. A can still be gridded, just with fewer levels: at 10 grids, for example, each one waits for a 0.4% move before it fills. Inside a band only 4% wide, how many round trips that produces in a day depends on how busily the coin oscillates, which lands you back at the first filter.
The classic mistake on a narrow coin is piling on grids by feel until the per-grid profit sinks to zero without anyone noticing. Put the range, the grid count and your own account's fee into the grid profit simulator first, or use the grid spacing calculator to see at which grid count the range hits the line. Arithmetic grids compute profit per grid differently; how the two spacing methods compare is in arithmetic vs geometric grid spacing.
Reading the trending list, the volatility list and marketplace returns
The trading bot section hands you ready-made lists, and plenty of people pick straight off them. Each one measures something different:
- Trending list. In the help center's words, it shows "the top 10 trending asset pairs by the number of grid strategies running." So it tells you how many people have a grid on that pair, not how those grids are doing. A crowd at least means the pair is not obscure, which is weak support for the liquidity filter; it says nothing about whether the pair is ranging right now.
- Volatility list. The annualized figure from 30 days of hourly closes, again a top 10. Use it as described under the first filter: cut the pairs that barely move, and do not copy the rest wholesale.
- Returns in the Bot Marketplace. The help center notes the data "is refreshed hourly and will only display strategies that are currently active." Bots that were stopped, or shut down after losing, are not in there. What you see is the survivors.
Our own habit is to let all three narrow the field and nothing more. Glance at the trending list for coins you already follow, use the volatility list to strike the sleepy ones, and treat a marketplace return as one example of a range setting that happened to work in the stretch just gone, not as next month's forecast.
What the filters leave for you to decide
The filters only answer whether a pair can be gridded. Where to draw the range and how many grids to use belongs to how to set grid parameters; whether your money stretches across that many grids is in how much a grid bot needs to start.
The other decision is your exit for when things go wrong. The same Chinese-language explainer notes that the bot stops placing orders once the market price falls below the lower limit, that during a crash or a spike it may keep executing as normal, and it recommends getting into the habit of setting a stop-loss price. A well-chosen coin still cannot stop the market from turning, so work out your stop with where to set a grid bot's stop-loss.
Questions
Is the most volatile coin the best one to grid?
Not necessarily. The volatility % shown in the trading bot section is an annualized figure built from the past 30 days of hourly closing prices. It measures how hard price moves, not which way, so a coin sliding down every day can score high too. Volatility only gets a pair through the first filter; it still has to come back inside your range, fill your orders and leave enough per grid to pay two fees.
How far apart do grids need to be to avoid losing money?
With the help center's geometric formula, Profit/Grid = (1 - c) * r - 1 - c, and the default fee c of 0.1%, the price ratio r between adjacent grids has to exceed about 1.002, roughly a 0.2% gap, before each grid makes anything. Plug in 0.075% for paying fees with BNB and the line drops to about 0.15%. Fees as shown on Binance at the time you trade, checked 2026-10.
Not investment advice. Crypto prices are volatile, a grid bot does not guarantee profit, and positions can end up stuck below your range (or liquidated, on futures). Assess your own situation. Every range, grid count and per-grid profit above is a hypothetical example used to explain the formula; none of it is real market data or a platform promise. Rules and formulas were checked against official material on 2026-10-06: the trading bot guide (English page updated 2026-01-30), the spot grid parameters page (updated 2026-01-07), the delisting FAQ (updated 2026-06-10), plus the Chinese-language spot grid explainer where noted. The 0.1% default fee, the 30-day volatility window and the monthly Monitoring Tag review are as shown on Binance at the time; the live pages govern. Sources: trading bot guide, spot grid parameters, delisting guidelines and FAQ.