Grid trading bot: how grids work and when they fail

A grid trading bot splits a price range into evenly spaced levels and places a buy order below the current price and a sell order above it at each level. Every time price drops a step, the bot buys. Every time it rises a step, it sells. In a market that swings back and forth, those small round trips add up.

The weakness is just as mechanical. When price stops ranging and trends, a grid keeps buying into the fall or selling into the rally. It ends up holding its biggest position at the worst moment. This guide explains how grid bots work, where they do well, and a worked example with numbers. It also covers why a trend can take out a loss limit in a single session, and why most prop firms, Velotrade included, do not allow grid trading.

Quick answer: A grid trading bot places buy and sell orders at fixed price intervals inside a chosen range, profiting from each small move back and forth. Grids tend to work in sideways markets and lose in strong trends, because every level fills against the move. Grid trading is prohibited on Velotrade evaluations and funded accounts.

---

Grid trading bot: key facts

What is a grid trading bot?

Grid trading is older than crypto. Investopedia describes it as placing orders above and below a set price at regular intervals, creating a grid of orders, and notes it is most associated with forex. Exchanges such as Binance now offer grid bots as built in tools. Binance describes its spot grid as placing orders at preset intervals within a configured price range to profit from small price changes.

There are two versions:

• Range grid (against the trend): buy orders below the set price, sell orders above it. The bot buys dips and sells rallies. This is the version most crypto grid bots use.
• Trend grid (with the trend): buy orders above the set price, sell orders below it. The position grows as price runs in one direction. Investopedia points out that this version struggles when price chops back and forth.

This guide focuses on range grids, because that is what almost every "grid trading bot" product means.

How does a grid trading bot work?

You choose five settings:

1. Range: the lower and upper price limits.
2. Number of grids: how many levels sit inside the range.
3. Spacing: arithmetic (equal price steps) or geometric (equal percentage steps).
4. Order size: how much the bot buys or sells at each level.
5. Stop: where the whole grid is closed if price leaves the range.

Binance's parameter guide gives the maths. In an arithmetic grid, the price difference per level is (upper limit minus lower limit) divided by the number of grids, so profit per grid is largest at the bottom of the range. In a geometric grid, each level is a fixed ratio above the last, so profit per grid is the same everywhere.

Once running, the bot pairs each buy with a sell one level higher. When a buy fills, it places the matching sell. When that sell fills, it places a new buy one level below. The loop continues while price stays inside the range.

When does grid trading work, and when does it fail?

Grid bots make money from oscillation. The more often price crosses the levels in both directions, the more round trips complete.

The downtrend case is the dangerous one. A range grid has no view on direction. It is designed to keep buying as price falls. Binance's own risk note warns that bots keep executing orders at the set intervals even during a rapid collapse or a strong rise. On a leveraged futures grid, that inventory also carries liquidation risk; check the maths with the crypto futures calculator.

This is why grid results look smooth for weeks and then give back a large share in a day. It is the same trade off as mean reversion trading without a stop: many small wins, then one large loss.

Worked example: sizing an ETH grid on an exchange account

Take a 10,000 USDT spot account on an exchange that charges 0.1% per side, the fee used in Binance's own grid example. The trader sets two personal limits: no more than 500 USDT lost in a day (5%), and no more than 1,000 USDT lost overall (10%).

The grid on ETH, with ETH at an illustrative 4,000 USDT:

• Range: 3,600 to 4,000, arithmetic, 100 USDT steps
• Buy levels: 3,900, 3,800, 3,700 and 3,600, each paired with a sell one level higher
• Grid stop: 3,500
• Order size: to be decided by the worst case

Step 1: work out the worst case

The worst case is that all four buys fill and then the stop is hit. With an order size of q ETH, the loss from entry to 3,500 is:

(400 + 300 + 200 + 100) x q = 1,000 x q USDT, plus fees.

Step 2: size to the daily cap

At q = 0.5 ETH, the move costs 500 USDT. Fees add about 7.50 USDT on the buys and 7 USDT on the stop exit, for about 514.50 USDT. That breaks the 500 USDT daily cap.

At q = 0.45 ETH, the move costs 450 USDT, and fees add about 13 USDT, for a worst case near 463 USDT. That fits under the daily cap with a small buffer. The capital needed to fill all four buys is about 6,750 USDT, so the spot account can fund it.

Step 3: compare it with the profit per grid

One round trip, buying 0.45 ETH at 3,900 and selling at 4,000, earns 45 USDT gross. Fees on about 3,555 USDT of turnover are about 3.56 USDT, leaving roughly 41.45 USDT.

One stop out at about 463 USDT therefore wipes out more than eleven completed round trips. That ratio is the real risk of a grid bot.

Step 4: what happens without a stop

Remove the stop and let a trend run:

• ETH falls to 3,200: loss of (700 + 600 + 500 + 400) x 0.45 = 990 USDT, close to the 1,000 USDT overall cap
• ETH falls to 3,000: loss of (900 + 800 + 700 + 600) x 0.45 = 1,350 USDT, through the overall cap

Each 100 USDT drop below the last level adds another 180 USDT of loss, because the bot now holds its full 1.8 ETH. The loss per step is four times larger than the profit per step was.

Why can a grid blow through a daily loss limit in a trend?

Daily loss limits are usually measured against equity, which includes floating losses on open positions. A grid in a trend builds a full position within hours, and the floating loss on that inventory counts straight away.

Three features make this worse:

• The loss is back loaded. The first fills look harmless. The damage comes from the last levels, when the position is largest.
• Speed. A grid can go from flat to fully loaded in a single sharp move, faster than a human can step in.
• Hope is built in. Without a stop, the bot holds its full position and waits for a return to the range that may never come.

That is why a grid sized only by "how much do I make per grid" so often breaches a daily limit on its first real trend day.

Can you run a grid trading bot on a prop firm account?

Not on Velotrade. The Velotrade trading rules list grid trading as a prohibited strategy type, described as a strategy that "places orders at fixed intervals without analytical basis; exposes trader to large losses in trending markets". The ban applies to evaluations and funded accounts, and automating a grid through a bot or EA does not change that. The same rules prohibit arbitrage, tick scalping and high frequency trading, and treat continuously averaging down a losing position as risk abusive.

Other parts of the rule set show why grids clash with prop limits. The maximum drawdown is static: a fixed dollar floor set from the starting balance, 10% on CLASSIC 2-Step, 7% on CLASSIC 1-Step and 3% on PRO 1-Step. The daily loss limit is 5%, 4% or 3% by plan, reset at 00:30 UTC from the higher of balance or equity and monitored against live equity. A trend that fills every grid level hits exactly those limits.

What to run instead on a prop evaluation

Velotrade does allow bots, EAs and algorithmic strategies on every account through the DXtrade REST and WebSocket API, at no extra fee. The difference is that each trade needs an analytical entry and a defined exit:

• Trend following: trade in the direction of the move, with a stop that limits each loss to a fixed amount.
• Mean reversion: fade stretched moves at a measured extreme, with one entry and a fixed stop instead of a ladder of orders.
• Breakout systems: enter when price leaves a range with confirmation, rather than placing orders at every level inside it.

Build the logic with how to build a trading bot, test it with backtesting trading strategies, connect it with the DXtrade API guide and the API access page, and size each trade with the position size calculator. Browse more strategy guides in the algo trading library, then compare plans on the challenges page.

Velotrade provides education and simulated trading only. It is not a broker and is not regulated as a financial institution. The grid example is illustrative, uses assumed prices and fees, and is not investment advice.

Sources

• Investopedia, "Grid Trading Definition": range and trend versions of grid trading and their drawbacks.
• Binance, "What Is Spot Grid Trading and How Does It Work?": how exchange grid bots place orders, and the warning that bots keep executing in a rapid collapse.
• Binance, "Binance Spot Grid Trading Parameters": arithmetic and geometric grid formulas and profit per grid.
• Bailey, D. H., Borwein, J., López de Prado, M. and Zhu, Q. J., "The Probability of Backtest Overfitting": why optimised grid settings often fail out of sample.
• CFTC, "Customer Advisory: AI Won't Turn Trading Bots into Money Machines": warning signs of trading bot fraud.
• Investor.gov (SEC), "Artificial Intelligence (AI) and Investment Fraud: Investor Alert": how automated trading claims are used in fraud.