Grid Trading Basics
Place buy and sell orders at preset intervals around a base price. Profit from price oscillation.
Setup
- Choose price range
- Set grid spacing
- Place orders at each level
Best Conditions
- Sideways market
- High volatility within range
- Liquid trading pairs
Risks
- Breakout losses
- Fees can eat profits
- Capital intensive
Building a Working Grid Setup
A grid divides a chosen price range into equal steps, with a buy order placed just below each step and a sell order placed just above it. When price falls to a step, the buy fills and immediately places a sell at the next level up; when price rises, the sell fills and re-places the buy below. The profit per oscillation is roughly the grid spacing, and the number of filled oscillations in a session determines the total income.
Choosing the Range and Spacing
Range selection is the entire game. If the range is too narrow, a breakout empties the grid into a long position; if too wide, the grid sits idle for weeks. Start with the 20-day high and low as the outer bounds, and set the grid width so that at least 2-3 oscillations are expected per week. For BTC/ETH perpetuals on Indian exchanges, a 2-3% grid on a 15-20% range is a common starting point; for altcoins use wider spacing because their ranges are sloppier.
- Range: use the Bollinger-band width of the last 20 sessions for an adaptive box.
- Levels: 10-20 grid lines is a good balance between fill frequency and idle capital.
- Direction bias: an upward-tilted grid (more buys below, fewer sells above) works in bull markets.
Capital Efficiency and Order Arithmetic
Each grid line needs reserved cash and reserve coins. If you commit 10 grid lines, split the cash so every buy level plus its assigned sell is covered; leaving a single level unfunded creates a stuck position when price retraces past the level. Add a trailing bailout: if price closes outside the range for two consecutive sessions, cancel all orders and flatten, accepting a loss on the trailing leg rather than carrying a naked breakout position through a crash.
Fee and Funding Effects
Every oscillation pays maker and taker fees on both sides. On a 2% grid, a 0.1% total round-trip fee consumes 5% of the gross profit per oscillation. Funding rates on perpetual swaps add a further cost that is invisible during range-bound trading but becomes material when the market trends. Measure net yield against the opportunity cost: a grid returning 1.5% monthly while BTC is trending is a poor use of capital.
When Grids Underperform
Grids lose in exactly two regimes: violent breakouts and dead calm. In a breakout the entire inventory tilts to coins falling in value; in calm the grid sits unfilled while cash earns nothing. The fix is a regime filter: only run the grid when the realized volatility sits inside the 30-70 percentile band of the last 90 days, and switch it off when daily range collapses below the grid spacing.
Re-balancing and Monitoring Rituals
Review the grid once daily at a fixed time: top up any level that is more than one spacing from current price, remove levels where the exchange's borrow rate exceeded 0.5% annualised, and log the number of fills. A journal that records fills, fees and the daily high-low tells you within two weeks whether the grid spacing matches the coin's actual oscillation behaviour.
Arithmetic vs Geometric Grids on BTC
Two grid geometries price very differently. An arithmetic grid places every level an equal rupee distance apart, which is simple to read and fine in a narrow range but under-sells the grid's classic bet on volatility scaling; a geometric grid spaces levels by a fixed percentage so the ladder tightens as price falls into deeper drawdowns, buying more heavily where the snap-back is historically strong. On a BTC/USDT range of 90,000 to 110,000, a geometric grid with 0.5 percent spacing places far more buy levels in the upper-middle territory than a flat 500-unit ladder would. Most serious volatility-neutral grids choose geometric spacing precisely because it mirrors how crypto price much scales with its own level.
Grid Density in Terms of Volatility, Not Fixed Percent
Tune the spacing from the instrument's realised volatility rather than a guess. If BTC's daily range averages 2.5 to 3 percent, a grid spaced 1 percent apart inside a 15 percent band creates a dense series of quick fills; a 3 percent spacing produces a sparser, slower book with larger per-level wins. The ratio that matters is spacing divided by the average daily move: between 0.3 and 0.7 the grid trades pleasantly, above 1 it starves, below 0.25 it churns on noise and pays the venue for nothing. Compute the number on a rolling 30-day window quarterly and re-tune the ladder to stay inside the region.
Capital Reserve for the Worst Day
Every grid has a drawdown profile it will drown in one day. Model the grid's notional at the deepest historic level inside its band - spot at the bottom of the ladder, all buy orders filled, the entire reserve deployed - and size the grid so that state is survivable rather than theoretical. Reserve a percentage of capital outside the grid entirely as a manually held cash floor, because the grid that buys the bottom forever has no opinion, only a deeper ladder. The symmetric truth: the rescue capital that buys after the grid's band breaks is what separates a grid that handles a crash from one that pretends the crash never happened.
Profit Maths With Fees and Funding
Walk the math before coding. A 100-unit geometric grid over a 15 percent range with 0.5 percent spacing and a 0.1 percent fee per side nets roughly 0.3 percent per completed round trip plus the fee drag from frequent small fills on lower legs. Subtracting swap funding if the venue charges it on leveraged crypto, and the modest spread on INR-adjacent pairs, leaves a narrow but real edge that compounds across many cycles - and a narrow edge that dies the moment either fee rises. The grid's honesty check is whether you can compute per-level net profit in your sleep; if that number is foggy, the grid is a gamble wearing engineering clothes.
Monitoring and Rebalancing Rituals
Automate the checks that matter: the number of open buy orders remaining, the distance to the grid's floor, and the realised profit per completed cycle this month. Rebalance only at range boundaries, refresh the grid levels when the band is re-established after a breakout, and mark in the journal how many cycles each 1 percent of the band produced. The grid fails when it is treated as a hobby and succeeds when it is treated as a hypothesis with a monitoring dashboard.
- Prefer geometric spacing calibrated to realised volatility.
- Size the deepest-fill state as survivable, not theoretical.
- Run the per-level net profit math before funding.
- Keep a manual cash reserve outside the ladder.
- Rebalance only at band boundaries; log every cycle.
Adaptive Re-Anchoring and the Kill Conditions
A static grid inherits the dead range: when the volatility regime shifts, re-anchor the ladder to the new range and write the kill conditions that decide when the grid is abandoned rather than extended - a volume-weighted average price that breaks the floor by more than a stated band, or funding that turns sharply against the pair's direction. The capture that matters is the case where the grid's math says the range is dead: compute the level at which the grid's mid-deployed capital equals the notional at the floor and pause there, because the grid that buys forever is the grid whose range was never modelled as deadly. Write the re-anchor and pause triggers in the config file, and let the bot obey the file while the trader owns the decision to rewrite it.
