Grid or DCA: lessons from 120 ETC backtests

A grid and a DCA bot can both buy in stages, but they do not close their positions in the same way. We reviewed an ETC/USDT grid experiment to explain the difference. The case uses 120 distinct configurations evaluated in three periods. The downloadable data lets you inspect the results and their limits.

How grid and DCA exits differ

A grid places buy and sell levels to capture movement inside a range. A DCA strategy adds buys and typically manages an exit around the combined average entry. If the market keeps falling, both can accumulate exposure. The strategy name does not remove that risk.

A simple example without fees: buying 20 USDT at 100 and another 20 at 90 produces an average entry of about 94.74. The average falls, but committed capital rises from 20 to 40 USDT. Explore that effect in the no-signup DCA calculator.

What was tested

ETC/USDT on Binance Futures, hourly candles and 200 USDT of initial capital. The preserved run dates from October 14, 2026. It includes long and neutral grids at 1×, 3× and 5× leverage, multiple level counts and spacings, with no global stop loss or take profit. Combinations where level count times spacing reached 100% were excluded.

Simulator assumptions: 0.04% commission, 0.03% slippage and 0.01% funding every eight hours. These fixed assumptions do not reproduce variable fees or every detail of exchange execution.

The verifiable results

• Full 2020–2026 period: 94 of 120 configurations liquidated; 26 finished with a positive return.
• 2025 separately: 47 liquidated and none finished positive.
• 2026 separately: 34 liquidated and 35 finished positive.

The highest full-period return was +174.50% with a 78.43% maximum drawdown: a 3× long grid with 10 levels spaced 5% apart. This is a simulated historical observation, not an expected return or recommended configuration.

The file calls its third period “validation 2026.” That label alone does not establish independent validation. Choosing the best rows after seeing that year still introduces selection bias. Meaningful validation requires choosing the configuration before inspecting the reserved period.

What a DCA comparison would require

This experiment does not include a complete, equivalent DCA run. It cannot establish that one algorithm is better than the other. A fair comparison needs the same asset, periods, costs, capital and risk, with selection completed before validation. Comparing only the top rows of different searches can be misleading.

How to use this information

Inspect total exposure, liquidation risk, maximum drawdown and behavior outside the range. A large return can hide a difficult path. Start with virtual funds and reserve a period you did not use for parameter selection.

Download 360 results: 120 configurations across three periods (CSV) · Read the summary and file hash.

Limitations include one asset, a fixed initial grid and simplified execution. This is backtest output, not live account performance or evidence of future profitability. This article does not recommend trading.

Explore Free and the calculator, or read how the laboratory works.