How crypto grid trading works
If you've watched crypto for a while and wondered whether a bot could quietly harvest the spread while you sleep, grid trading is the most rule-based way to try. This post walks through the mechanics with an ETH/USDT worked example, two realistic P&L scenarios (the ranging market where grid bots thrive and the trending market where they don't), and the fee math beginners skip.
What a grid bot actually does
A grid bot places a ladder of staggered limit orders inside a chosen price range. When price fills a buy at a lower level and later crosses a sell at a higher level, the bot captures the spread between them. Stack enough lines and every small oscillation in price turns into a realized trade.
The CashflowLoop default spec is the ETH/USDT 15-grid: 15 evenly spaced limit orders across an auto-set price band of roughly 95% to 105% of the current spot price, with equal capital allocated per grid line. Every grid line that fills buys one level lower and sells one level higher — so each completed round-trip captures that level's spread, before fees.
The strategy is fully deterministic. There is no chart-reading, no sentiment signal, no FOMO exit. The bot's only job is to keep the ladder filled and to harvest the spread when price oscillates back across a filled line.
A worked ETH/USDT example
Say ETH trades at $3,000 on Binance and you allocate $1,500 across all 15 lines. That's $100 per line, or roughly 0.033 ETH per line at the spot price. Three of those lines, focused around the spot price:
| Price | Side | Qty (ETH) | Capital | Spread per round-trip |
|---|---|---|---|---|
| $2,900 | buy | ~0.0345 | $100 | ~4.6% to next sell |
| $2,933 | buy | ~0.0341 | $100 | ~3.4% to next sell |
| $2,967 | buy | ~0.0337 | $100 | ~2.2% to next sell |
| $3,033 | sell | ~0.0330 | $100 inventory | ~2.2% from prior buy |
| $3,067 | sell | ~0.0326 | $100 inventory | ~3.4% from prior buy |
| $3,100 | sell | ~0.0323 | $100 inventory | ~4.6% from prior buy |
A round-trip on the middle pair — buy at $2,933, sell at $3,067 — is a 3.4% gross move. After a 0.1% taker fee on each leg, the net capture per round-trip is about 3.2%. That is the number the strategy is actually earning, and the number that has to beat fees + slippage + idle inventory.
When grid bots make money: ranging markets
Grid bots earn the spread in ranging / sideways markets — places where price oscillates inside your band and repeatedly fills buys and sells. The below is an illustrative six-day net P&L path for a hypothetical 15-grid ETH/USDT run with $1,500 deployed during a choppy range. Any number here is illustrative, not promised.
Day 1 +$2.40 +0.16% net (2 round-trips) Day 2 +$4.10 +0.27% net (3 round-trips) Day 3 -$0.80 -0.05% net (1 stale limit cancel fee) Day 4 +$5.60 +0.37% net (5 round-trips, tight range) Day 5 +$3.20 +0.21% net (3 round-trips) Day 6 +$1.50 +0.10% net (2 round-trips) ----- Total +$16.00 +1.07% net over 6 days
The shape that matters: many small wins, very few losing days, and almost no day where the bot is "down big". That is what you are paying $20/mo for — not a home-run alpha, but a steady harvest of the spread that range-bound markets leave on the table.
When grid bots lose money: trending markets
Grid bots typically underperform in strong trends. Picture the same bot running through a runaway ETH bull move:
- The lowest three buy limits fill on day 1 as price rips up $3,000 → $3,200. Inventory is now 0.1 ETH sitting at an average $2,950.
- The sell limits at $3,033 / $3,067 / $3,100 keep filling — but they are selling ETH you just bought, and the proceeds get redeployed into higher buy limits further up the ladder.
- By day 5 ETH is at $3,400. The bot is fully long from the early buys and its ladder has shifted entirely into the upper half of the band. The next time price dips, those filled buy limits re-fill at unfavorable prices.
- The realized P&L is smaller than if you had just held ETH. The unrealized inventory is up big, but it is also exposed to any reversal.
Trades are still executing — the fills look normal in the dashboard — but the strategy is fighting the trend, not capturing the spread. Any specific return figure is illustrative, not a promise. Your results depend on the pair, volatility, spread, fees, and how often price revisits your range. Trading bots involve risk. Past performance does not guarantee future results. Start in paper mode, watch the fills, and only commit capital you can afford to leave working.
Fees, slippage, and the part beginners skip
Binance Spot taker is 0.1% per leg. On the 3.4% round-trip above, that's 0.2% in round-trip fees — leaving 3.2% net. Shrink the band and the fees take a much bigger bite:
| Grid spacing | Gross round-trip | Round-trip fees | Net capture |
|---|---|---|---|
| 1.0% | 1.0% | 0.2% | 0.8% |
| 0.5% | 0.5% | 0.2% | 0.3% |
| 0.3% | 0.3% | 0.2% | 0.1% |
| 0.2% | 0.2% | 0.2% | 0.0% (break-even) |
The line that matters: at ~0.2% spacing the round-trip does not pay for its own fees. Going tighter than that without using BNB fee discounts or a maker-rebate path is burning money on every fill. Net is what matters, not gross.
Risk disclosure
Reading this before subscribing? The grid bot overview is a faster skim.
Already comfortable with the mechanics? Is crypto grid trading profitable? walks through a worked ETH/USDT 2,800–3,200 P&L example, the fee math, and the breakout case where a grid bot trails a buy-and-hold.
Still choosing between grid and DCA? The crypto grid bot vs DCA comparison lays out regime fit, capital efficiency, the fee/slippage tradeoffs, and when to combine both on the same $500 ETH/USDT 2,800–3,200 spec.