A grid strategy places multiple price levels for potential buys and sells within a defined interval. As orders fill, the account accumulates inventory and may place opposing orders at other levels. The grid spacing, quantity, starting inventory, rebalancing, costs and behavior outside the interval are all part of the method. A diagram of evenly spaced lines does not reveal the resulting account exposure.
The grid is a rule for managing orders and inventory, not a forecast that price will stay inside the band. A spot grid uses cash and the asset; a futures grid can create long or short contract exposure and may add leverage, funding and liquidation rules. The CFTC’s virtual-currency risk advisory explains that leverage magnifies risk and adverse moves can force margin additions or closeouts; contract and venue rules determine the actual mechanics. Neutral, long-biased and short-biased configurations begin with different exposures. These choices change the strategy being tested, so name them before comparing results.
Choose the spacing rule before you test
Arithmetic spacing keeps a constant price distance between levels. Geometric spacing keeps a constant ratio, so the percentage step is similar across the range. In either design, distinguish completed buy/sell pairs from the value and unrealized gain or loss of remaining inventory. This article uses those as separate ledger categories for its fictional example; venues can report them differently.
For an index-only example, take a $90–$110 interval split into four gaps. Arithmetic levels are $90, $95, $100, $105 and $110. Geometric levels are approximately $90, $94.63, $99.50, $104.62 and $110. They allocate the same endpoints differently. Neither spacing scheme knows whether the next move will reverse, and the illustrated values are not calibrated to a token or a volatility estimate.
Follow a fictional two-level example
Suppose a fictional spot grid starts with no asset and $200 of dollar-value quote balance. It places one-unit buys at 98 and 96. If both fill, the account holds two units bought for $194 in total and has $6 left in cash. The average purchase price is $97 per unit.
Now suppose price later trades up through the sell levels at 100 and 98, closing one unit at each. The gross difference is $2 on each pair, or $4 total. Fees reduce that result. Hyperliquid’s fee page currently lists a base-rate tier-0 spot maker fee of 0.04% and taker fee of 0.07%; its actual tier depends on rolling volume and other account-specific discounts. If all four hypothetical fills count as maker fills with no discount or rebate, the fee arithmetic is (98 + 96 + 100 + 98) × 0.0004 = $0.1568, leaving $3.8432 before spread and other costs. If all four are charged at the listed taker rate, fees are $0.2744, leaving $3.7256. The fee schedule was checked 26 September 2026; these are venue-specific assumptions, not a quote for every account. A resting limit order can still fail to fill, and an order’s maker/taker status follows the venue’s execution rules.
The same initial buys look different if price keeps falling. At 92, with no sell fills, the two units are worth $184 and the $6 cash remains: marked equity is $190, a $10 unrealized loss before fees against the $200 starting balance. A record of two possible $2 completed grid cycles would not show this open inventory risk.
If price instead falls through the grid, buys can fill without a later sale. Inventory and exposure grow while the price moves against the initial range assumption. If price rises through the top, sells may execute and the strategy can become underexposed relative to a continued move. A grid must define whether it stops, recenters, hedges, holds inventory or replenishes orders outside its original bounds. Each behavior is a distinct risk decision.
Model actual order mechanics
Resting limits can remain unfilled, fill partially or execute before a displayed chart touch reaches the queue. Fees depend on venue and maker/taker treatment. Spread and slippage can consume the distance between neighboring grid levels. On a leveraged perpetual, funding, collateral, maintenance requirements and liquidation risk apply in addition to grid inventory. A fixed spacing may be narrower than costs during some conditions and wider than expected movement in others.
Record the venue and instrument, tick and quantity rules, price boundaries, spacing, order size, total notional cap, reserve cash, fee tier and cost assumptions. Simulate how many levels could fill during a one-way move and what happens if an order is rejected or canceled. Mark open inventory to the same stated price used for the rest of the account, and include all realized fees. Do not count each completed buy-sell cycle as a separate profitable trade while leaving the remaining inventory unvalued.
For a one-unit spot round trip, let the buy fill be B, the sell fill be S, and the two proportional fee rates be fB and fS. Before other costs, net cash change is S − B − (B × fB) − (S × fS). If the fills are $95 and $100 and each assumed fee is 0.10%, gross difference is $5.00, fees total $0.195, and the result is $4.805 before spread, slippage or any other charge. This formula assumes both fills occur and ignores tax, conversion and inventory opportunity cost. Real order status, fill quantity and fee asset can change the ledger.
The price increment must therefore clear both sides’ costs with room for execution uncertainty. A tight grid can generate more order events while leaving less gross distance per completed pair; a wider grid can leave capital idle and may trade rarely. These are hypotheses to test on the selected instrument and period, not general rules that one density is best.
Predefine the response to a range break
Before testing, write down what happens when price closes beyond the upper or lower boundary: stop placing orders and keep the residual inventory, close it under a stated execution assumption, recenter using a rule fixed in advance, or hedge under a separate instrument model. A stop instruction does not guarantee its trigger or fill price during a gap or fast market. Recentered grids also create a new rule version; do not redraw the range after seeing the same test’s outcome and then present it as a fresh result.
Check what the research tool represents
SuperQuant’s current examples do not implement an exchange grid-order manager. Its rule-based backtests cannot demonstrate queued limit fills, automatic re-centering or exchange inventory management. The included hourly histories are synthetic. A custom research model must state these assumptions rather than imply the product ran a grid against live markets.
Read the range trading guide, fee calculation and position sizing guide. A grid can accumulate directional exposure and lose money; this mechanics article is educational, not a recommendation or promise of returns.