Somewhere on the grid creation screen there is a choice of how to cut the range into levels: arithmetic or geometric. Plenty of people click one and move on, because it reads like an implementation detail. It is an implementation detail — and it decides how many of your levels work for nothing.
The usual answer online is one line: "use geometric for wide ranges." That is not wrong, but it does not explain why, so it cannot help you judge whether your range counts as wide. What follows unpacks it: the mathematical difference, why the fee schedule is the only thing that really decides, and a three-step test you can run on your own numbers. This logic holds on any platform that offers grid bots; Binance is simply the venue used for the examples.
1. The difference in one line each
- Arithmetic: adjacent levels are a fixed amount apart. Range 20,000 to 60,000 in 40 levels means every level is 1,000 apart.
- Geometric: adjacent levels are a fixed ratio apart. Every level is about 2.79% above the last, and multiplying up from 20,000 lands exactly on 60,000 after 40 steps.
Drawn on an ordinary price chart, an arithmetic grid is a ruler with even marks and a geometric grid spreads out as it climbs. Switch the vertical axis to a log scale and they swap: geometric becomes the even one and arithmetic bunches up near the top.
Here is the sentence the whole article rests on: price moves in percentages, fees are charged in percentages, but an arithmetic grid is cut in dollars. The units do not line up, and that mismatch is where the problem starts.
2. Same range, two spacings, very different levels
What follows is an illustrative worked example, used to show structure only. It is not a forecast and not a claim about fills you will get. Take a range of 20,000 to 60,000 — top exactly double the bottom — cut into 40 levels.
| Where in the range | Arithmetic: fixed 1,000 gap | Geometric: fixed ~2.79% |
|---|---|---|
| Near the bottom (around 20,000) | 1,000 ÷ 20,000 = 5.00% | about 2.79% |
| Middle (around 40,000) | 1,000 ÷ 40,000 = 2.50% | about 2.79% |
| Near the top (around 59,000) | 1,000 ÷ 59,000 ≈ 1.69% | about 2.79% |
Down the arithmetic column, the percentage loses about two thirds of its value from bottom to top. The same fill earns 5% near the floor and under 1.7% near the ceiling. You thought the levels were interchangeable; they are not. The geometric column is flat all the way, and that is its one and only advantage.
3. Why fees are the only real test
Whether a level makes money is never about how many dollars it spans. It is the percentage that level earns, minus the percentage cost of the two trades that complete the round trip. Fees are a percentage, so they impose the same flat hurdle at every level, while arithmetic spacing hands you a percentage that drifts. Put those two facts together and the conclusion is immediate:
An arithmetic grid on a wide range always has a stretch of levels closest to the hurdle, and it is the ones at the top. Stay with the example: if your round trip costs 0.2% in total (illustrative — use your own account's current rate), the 5% level nets about 4.80% and the 1.69% level nets about 1.49%. Both fine. Now raise the level count from 40 to 200, which makes each gap 200 wide, and it turns:
| Where in the range (200 levels, 200 apart) | Gross per level | After a 0.2% round trip |
|---|---|---|
| around 20,000 | 1.00% | about 0.80% — fine |
| around 40,000 | 0.50% | about 0.30% — thinning out |
| around 59,000 | about 0.34% | about 0.14% — barely worth filling |
Inside one strategy, the bottom levels are earning normally while the top levels are essentially working for the exchange — and not because you set something wrong. It is what arithmetic spacing does on a wide range. Switch to geometric and 200 levels come out around 0.55% each, netting roughly 0.35% after the same round trip, identical across the range. Either the whole thing is worth running or it is not, and you only have to make that judgement once.
4. The three-step test
Instead of memorising what counts as "wide", run these three steps on your own numbers:
- Compute the range ratio: upper bound ÷ lower bound. Under about 1.2 — a range spanning twenty percent — the two methods produce nearly the same percentage per level, and arithmetic is friendlier because levels land on prices you can think in. At 1.5, 2 or more, the gap between the methods gets real.
- Find your thinnest level: with arithmetic spacing it is always at the top, and a good enough estimate is the gap size divided by the upper bound. That number is the floor on what any level in this configuration can earn.
- Compare it to your round-trip fee rate: subtract your two-way fee percentage from that thinnest gross. What is left is what that level actually earns. If it is close to zero or negative, you must change one of three things — the spacing method, the level count, or the range width. Leaving all three alone is not an option.
| Your situation | Better fit | Why |
|---|---|---|
| Narrow range, bounds within about 20% of each other | Arithmetic | The percentage drift is tiny, and levels on round prices match how you already think about support and resistance |
| Wide range spanning a doubling or more | Geometric | The only spacing that keeps the profit rate equal everywhere, so no stretch of levels works for free |
| Low-priced assets with many decimals | Geometric | Cutting by amount runs into tick-size limits quickly; cutting by ratio scales cleanly |
| You specifically want orders sitting on round numbers | Arithmetic | Readability wins when the levels are aligned to prices you actually watch |
| You have set a very high level count (hundreds) | Compute the thinnest level before choosing | With enough levels either method can drop under the fee hurdle — at that point the problem is the level count, not the spacing |
5. What it does not fix
Geometric spacing gets talked about as the more professional option, so be clear about what it cannot do before you reach for it:
- It does not reduce losses. Break below the floor and a geometric grid leaves you bagged much the same as an arithmetic one. The difference is in how profit rate is distributed across levels, not in directional risk. What to do once price leaves the box is in price left your grid range.
- It does not read the market for you. Whether a grid earns comes first and foremost from whether price actually chops inside your range over the period you run it. Spacing is a third or fourth order concern.
- It changes nothing about futures liquidation. The magnitude difference between the two products is covered in spot grid vs futures grid.
- It is not a mid-run toggle. Spacing is fixed at creation, so switching means terminating and rebuilding — a full re-entry with the costs that go with it.
Put back in context: range width, level count and profit per level all constrain each other, and arithmetic versus geometric is the dial that decides how profit per level is distributed across the range. The order to set all of them in is in how to set grid parameters. And once it is running, the annualized figure the interface shows you deserves its own reading — that is where a grid bot's annualized return comes from.
6. FAQ
What is the actual difference between arithmetic and geometric grid spacing?
One splits the range by a fixed amount of money, the other by a fixed ratio. With arithmetic spacing every pair of adjacent levels is the same number of dollars apart, say 100 USDT. With geometric spacing every pair is the same multiple apart, say 1 percent. The consequence is that an arithmetic grid gets thinner in percentage terms as price rises: a 100-wide gap is 0.5 percent at 20,000 and only about 0.17 percent at 60,000. A geometric grid holds the same percentage across the whole range, at the cost of much wider dollar gaps near the top. Neither is universally better; it depends on how wide your range is.
Why are fees the thing that actually decides it?
Because fees are charged as a percentage of the trade value, not as a flat amount. Whether a level makes money is a contest between the percentage profit of that level and the percentage cost of the two trades that complete the round trip. How many dollars the level spans is irrelevant. Arithmetic spacing on a wide range makes the percentage uneven, so the levels near the bottom are fat and the ones near the top are thin, and any level thinner than your round-trip fee rate loses money every time it fills. Geometric spacing flattens the percentage, so if one level clears the fee hurdle then all of them do.
How wide does a range have to be before geometric is the right call?
There is no official threshold, but there is a test you can run yourself: divide the upper bound by the lower bound. If that ratio is close to 1, say under 1.2, the two methods produce nearly identical percentages per level and arithmetic is easier to reason about in round price numbers. If the ratio is clearly above 1, for example if the top is double the bottom, an arithmetic grid will have top levels worth a small fraction of the bottom ones and geometric becomes the sensible choice. The point is not to memorise a number but to compute the thinnest level yourself and compare it to your own round-trip fee rate.
Is a geometric grid safer?
No. Geometric spacing solves exactly one thing: it makes the percentage identical at every level, so the economics of each fill are the same. It does not reduce your losses and it does not change the risk profile of grid trading at all. Price breaking below your floor still leaves you holding a losing position, price running out the top still leaves you on the sidelines, and a futures grid still carries liquidation risk. Choosing the spacing method is a parameter-level refinement; judging whether price will actually chop inside your range is the step that decides the outcome. Do not confuse the two.
Can I change the spacing method on a bot that is already running?
Spacing is a structural setting fixed when the strategy is created, so it generally cannot be switched mid-run. Changing it in practice means terminating the current bot and creating a new one with new settings, which is a full re-entry and carries the cost of settling out and building the position again. That is why this choice is worth making carefully before you start. Which settings can be edited while running, and what termination does to your holdings, follow whatever your platform's current pages say.
Nothing here is investment advice. Crypto prices are volatile, grid trading does not guarantee a profit, and it carries the risk of being left holding a losing position and — on futures — of liquidation. Judge your own situation and size accordingly. The 20,000 to 60,000 range, the 40 and 200 level counts and the 0.2% round-trip rate are illustrative assumptions used to show structure, rounded for display; they are not forecasts and not your account's real fee rate. Actual rates, available parameter ranges and the names given to each spacing method follow whatever your creation screen shows at the time. Official product page to check against: What is spot grid trading.