OKX Grid Trading Settings
Range, grid count and the fee threshold
A grid bot buys low and sells high inside a range you set. It earns in a market that chops sideways and loses in a one-way decline, where you end up holding the full position. Before you open one, be able to answer two questions: is the market really range-bound, and are your upper and lower bounds reasonable? If the price leaves the range, the bot stops earning — that is how it works, not a fault.
How a grid bot works
The plain-language explanation:
- On BTC/USDT you set a range, e.g. upper $70,000, lower $60,000;
- The range is divided into N grid levels, e.g. 50;
- Each level is one price step (e.g. $200 / level);
- At each level the bot places a buy and a sell order;
- Price falls and triggers a buy → uses USDT to buy → price rises one level and triggers the matching sell → earns one level of spread;
- As price oscillates back and forth, the bot repeatedly captures the spread.
A grid bot doesn't predict direction — what it earns is "volatility itself". The choppier the price action, the more daily spread it earns. If price is completely flat, the bot earns nothing.
On the other side:
- Price breaks out of the upper bound: the bot sells out, sits in cash waiting for a pullback. If price keeps rising and never returns, the bot is completely idle.
- Price breaks down through the lower bound: the bot ends up full-position, stuck waiting for a rebound. If price keeps falling, the book loss grows.
Which markets to run a grid in — and which to avoid
| Market type | Example | Grid suitability |
|---|---|---|
| Range-bound chop | BTC oscillating in 60–70k | Very suitable |
| Slow uptrend | BTC weekly uptrend with 3–5% intraday swings | Suitable (re-base the range upwards as price rises) |
| Slow downtrend | BTC weekly continuing lower | Not suitable (you'll be stuck) |
| Sharp upside breakout | BTC rallying 60k → 75k in a week | The bot sells out and watches the rally |
| Sharp drop | Heavy negative catalyst, −20% in a day | Stuck full-position with a deep book loss |
How to judge whether the current market is range-bound
Two simple checks (no technical-analysis background required):
- Look at the past 30-day candles — is price oscillating inside a clear box?
- Have the recent highs and lows held, or has each low been lower than the last?
A clear box with highs and lows holding → suitable for a grid; a series of lower lows → wait and watch.
6 steps to set up your first grid
Log in → Trading Tools → Grid trading
Pick the trading pair
For beginners we recommend a BTC/USDT or ETH/USDT spot grid. Why:
- Good liquidity, low slippage;
- Plenty of historical data, makes the range easy to set;
- Less prone to collapsing than a small coin — though both can still fall a long way below your range.
Don't run a grid on a newly-listed / small-cap / meme coin. Their "chop" is often "from $1 to $0.1" chop.
Choose "spot grid bot" (not futures grid bot)
OKX offers both spot and futures grid bots. Beginners should only choose the spot grid — the futures grid carries leverage; when you get stuck, it's not just a book loss, you can be liquidated.
Set upper/lower bounds + number of grids
The most important step. Two core parameters:
- Upper / lower bounds — look at the past 30–60 days of highs and lows; set the upper bound just above the recent high and the lower bound just below the recent low, with 5–10% safety on each side.
- Number of grids — the examples below use 20 to 80. More grids = smaller spread per level but more frequent fills; fewer grids = the opposite.
Check the per-grid yield before launching. Grid orders are limit orders resting on the book, so each buy and each sell pays the Maker fee; one full grid (a buy plus the matching sell) has to clear 2 × Maker. At the regular-user spot Maker rate of 0.08% (OKX fee page, checked October 2026), that threshold is 0.16%. If the per-grid yield is below it, reduce the grid count or widen the range.
Funding amount
First run: a small amount you can afford to lose. Divide it by the grid count and check each order clears the minimum order size OKX shows for the pair.
OKX will display how much USDT will be parked at the grid levels under your parameters and how much will be held as "waiting to buy" reserve.
Launch + monitor
Click "Create grid". The bot immediately places buys and sells across your defined range.
After launch, don't refresh every hour — a grid is slow work. Check once a week for cumulative return, current holding ratio, and whether price is still inside the range.
3 core parameters — how to tune them
| Parameter | Beginner recommendation | Trade-off when adjusting up / down |
|---|---|---|
| Range width | Past 60-day high/low ± 5–10% | Wide → less likely to break out but smaller per-level yield; narrow → higher per-level yield but easier to break out |
| Grid count | 20–80 (see the examples below) | Many → high execution frequency, small per-trade size; few → larger per-trade size, less sensitive to small moves |
| Capital share | 10–20% of account | High → higher absolute return but larger risk exposure; low → the opposite |
Three example set-ups, with the arithmetic
Per-grid yield ≈ (upper − lower) ÷ number of grids ÷ price. The examples use an illustrative BTC price of $65,000 and the regular-user Maker fee of 0.08% per side (0.16% per round trip, OKX fee page checked October 2026):
| Set-up | Range | Grids | Step per grid | Per-grid yield before fees | After 0.16% fees | Breakout risk |
|---|---|---|---|---|---|---|
| Conservative | ±20% ($52,000–$78,000) | 80 | $26,000 ÷ 80 = $325 | $325 ÷ $65,000 = 0.50% | 0.34% | Low |
| Balanced | ±15% ($55,250–$74,750) | 50 | $19,500 ÷ 50 = $390 | $390 ÷ $65,000 = 0.60% | 0.44% | Moderate |
| Aggressive | ±8% ($59,800–$70,200) | 20 | $10,400 ÷ 20 = $520 | $520 ÷ $65,000 = 0.80% | 0.64% | High — the price can leave this range quickly |
The per-grid yield is measured at the illustrative price; near the lower bound the same step is a slightly larger percentage, near the upper bound a slightly smaller one. Put in the current price when you do the sum yourself.
The four numbers to read before you launch
- Per-grid yield after fees — the most important one. If it isn't above 2 × the Maker fee you pay, the grid is mostly paying fees.
- Price step — range width ÷ number of grids: how far the price has to move to complete one buy-and-sell. A very small step means fills that barely clear the fee.
- Capital per grid — total capital ÷ number of grids: the size of each order. If it's very small, check it against the minimum order size OKX shows for the pair.
- Distance to the bounds — how close the current price is to the upper and lower limits. Below the lower bound you're left holding the full position; above the upper bound you're out of the rally. Leave 5–10% of room on each side.
6 grid pitfalls
⚑ Opening with too narrow a range
"Higher per-level yield" is tempting, but a narrow range can break out within days. Once it breaks out the bot stops earning, and a slightly wider range would have kept working.
⚑ Running a grid in a one-way trending market
"Looks volatile" doesn't equal "range-bound" — it can be a counter-trend pullback inside a one-way move. Running a grid in a downtrend = walking yourself into "stuck full-position".
⚑ Expecting "X% daily yield" from launch
Grid returns are extremely market-dependent: a choppy market fills many grids, a quiet one very few, and a one-way trend can make the result negative. Any promise of "guaranteed 1% daily" is a scam.
⚑ Adjusting parameters in a panic after a breakout
Widening the range after price has already broken out is closing the gate after the horse has bolted. Pre-commit psychologically to "what to do if it breaks out" when you set the grid — if it breaks out, stop, don't double down.
⚑ Opening multiple grids on the same coin with the same capital
The capital is diluted, each grid becomes inefficient. On the same coin, one grid is enough.
⚑ Using a futures grid instead of a spot grid
A futures grid carries leverage: if price runs through the range, the position can be liquidated and the margin lost. Beginners should stay away.
When to stop / tear down a grid
Four situations call for stopping the grid:
- Price breaks out of the range and a short-term reversion looks unlikely — e.g. it holds above the upper bound and there is fresh fundamental news supporting the move;
- Cumulative 30-day net return is negative — often a sign the market isn't range-bound after all;
- You judge the market is entering a trending phase (this judgement is easy to get wrong, use with care);
- You need the capital for something else.
When you stop a grid, decide what happens to the coins it is holding. Closing them at market costs Taker fees but leaves no leftover exposure, which is the simpler choice for a beginner; keeping them means you now hold that position outside the grid.
Risk disclosure
The grid bot is not a guaranteed-return tool. Running a grid in an unsuitable market = losing money in a disciplined way. This tutorial only covers operation and mechanics; it does not forecast "is now a good time to open a grid" — that judgement has to combine your read of the current market with your own risk tolerance. Any promise of "grid guaranteed returns / minimum daily yield X%" is a scam. This site does not take responsibility for P&L you generate by following this tutorial.
FAQ
Does a grid bot definitely make money?
No. A grid makes money in a "range-bound" market and loses money in a "one-way down" or "breakout-and-leave" market. Any "guaranteed return" claim about grids is wrong.
How do I set the upper / lower bounds?
Look at the past 30–60 days of highs and lows; set the upper bound near the recent high, the lower bound near the recent low, leaving a 5–10% safety buffer on each side. If price breaks out, the bot just "hangs" — sells out and waits for a pullback on the upside, or holds full coin and waits for a rebound on the downside.
Is more grids always better?
No. More grids → smaller per-level yield → larger fee share. Whatever the count, keep the per-grid yield above 2× the Maker fee (0.16% at the regular-user rate of 0.08%).
Can I touch the capital after opening the grid?
No. Capital locked inside the grid cannot be moved directly while the bot is running — all of it is spread across buy/sell orders at each level. To free it up you have to stop the grid first.
Can I rely on suggested grid parameters?
Only as a first draft. Suggested parameters are based on past prices, and the past range gives no guarantee the price will stay in the same range. Work out the per-grid yield and the distance to the bounds yourself before launching.