Grid spacing determines how far apart your limit orders sit. Arithmetic spacing places them at equal dollar intervals. Geometric spacing places them at equal percentage intervals. For most setups on a narrow range the difference is small. On wide ranges, high-priced assets, or setups where you care about consistent return per round trip rather than consistent dollar income, the choice matters.
Arithmetic spacing
With arithmetic spacing, every grid level is the same dollar distance apart. A range of $90,000–$110,000 with 20 grids produces levels at $90,000, $91,000, $92,000 … $110,000 — each $1,000 apart regardless of where in the range price sits.
Each round trip earns the same gross dollar amount: the grid spacing multiplied by the order size. A round trip at $91,000–$92,000 earns the same dollar profit as one at $109,000–$110,000. This makes arithmetic grids easy to reason about — income per fill is constant and predictable.
The trade-off is that the percentage return per round trip varies by price level. A $1,000 move from $90,000 to $91,000 is a 1.11% move. The same $1,000 move from $109,000 to $110,000 is a 0.92% move. At the lower end of the range you earn a higher percentage return per dollar of capital deployed; at the upper end you earn less. This asymmetry is usually small within a 10–20% range but grows with range width.
Geometric spacing
With geometric spacing, every grid level is the same percentage distance apart. A range of $90,000–$110,000 with 20 grids uses a constant multiplier between levels — each level is approximately 1.05% above the previous one. The dollar gaps between levels are smaller near the bottom of the range and wider near the top.
Each round trip earns the same percentage return on the order notional, regardless of where in the range the fill occurs. A round trip between adjacent levels at $90,000 earns the same percentage as one at $109,000. This consistency matters if you are sizing orders as a fraction of capital and want uniform return per fill across the whole range.
A worked comparison
Same setup, same range, same grid count — the only difference is spacing type.
Range: $80,000 – $120,000 (50% wide) Grid count: 10 Order size: 0.01 BTC per level Arithmetic spacing: $4,000 between levels Level pair Dollar gap % move Gross per RT $80,000 – $84,000 $4,000 5.00% $40.00 $100,000 – $104,000 $4,000 4.00% $40.00 $116,000 – $120,000 $4,000 3.45% $40.00 Geometric spacing: ~4.14% between levels (constant %) Level pair Dollar gap % move Gross per RT $80,000 – $83,310 $3,310 4.14% $33.10 $99,690 – $103,820 $4,130 4.14% $41.30 $115,290 – $120,060 $4,770 4.14% $47.70
With arithmetic spacing, every fill earns $40. With geometric spacing, fills near the bottom earn less ($33.10) and fills near the top earn more ($47.70) — but each fill represents the same 4.14% move. If price spends more time near the bottom of the range (common for a long grid in a recovering market), arithmetic spacing produces more total income in that region. If price oscillates evenly across the range, geometric spacing produces more balanced returns.
When the difference is negligible
For a range width of 10–20%, arithmetic and geometric spacing produce almost identical level placement. The divergence between them grows quadratically with range width. On a 10% range the difference between level positions is under 0.5% — effectively noise. On a 50% range (as in the example above) the difference is meaningful and worth a deliberate choice.
| Range width | Max level divergence | Recommendation |
|---|---|---|
| ≤ 20% | < 1% | Either — use arithmetic for simplicity |
| 20% – 40% | 1% – 4% | Consider geometric if consistent % return matters |
| > 40% | > 4% | Geometric for wide ranges on volatile assets |
Which to use
Use arithmetic when your range is narrow (under 20%), when you want predictable dollar income per fill, or when you are comparing results against a fixed dollar benchmark. It is also easier to audit — you can check any level placement by mental arithmetic.
Use geometric when your range is wide (over 30%), when you are running on a high-priced asset where dollar gaps at the top of the range become large, or when you want the bot to earn consistent percentage returns across all price levels. Geometric spacing is the more theoretically correct choice for assets with log-normal price distributions — which is most crypto perpetuals.
Set a wide range — $70,000 to $130,000 — with 20 grids and toggle between arithmetic and geometric spacing. The order book visualisation shows how the level distribution shifts and the profit per round trip changes at each end of the range.
Launch the simulator →