Why the winner keeps changing
These three strategies aren't competing on skill — they're each structurally suited to a different market shape. Buy and hold wins when there's a sustained trend to hold through, because every dollar is exposed from day one. DCA wins when the period has a real drawdown followed by recovery, because averaging in mechanically buys more coins on the cheap days without requiring you to call the bottom. A grid bot wins when price oscillates inside a range without a clear direction, because it's structurally designed to sell high and buy low on repeat inside that band — and it structurally loses in a strong trend, because it keeps taking profit into a rising market and can run out of buy-side capital in a falling one. Run last year on BTC, then run the last 90 days — the ranking rarely stays the same twice.
What this report does and doesn't capture
The grid bot simulation here uses daily high/low, not order-book fills — it assumes every price level the candle's wick touched during the day would have been filled, which is usually a reasonable approximation for a wide grid but can overstate fills for a very tight grid on a volatile day. It also ignores exchange trading fees, which on a busy grid bot doing dozens of round trips can meaningfully eat into the realized edge — a 0.1% taker fee on 50 round trips is roughly 10% of capital in fees alone, easily larger than the grid's edge in a quiet period. The DCA and buy-and-hold legs ignore fees too, for a fair three-way comparison, and none of the three account for slippage, funding costs on leveraged versions, or taxes. Treat this as a shape-of-the-market comparison, not a promise of what any specific bot or exchange fee schedule would have returned. For the fee-aware version of the grid math alone, see the Grid Spacing Calculator, which nets out the fee floor per fill.
FAQ
Which strategy backtests best — buy and hold, DCA, or a grid bot? It depends entirely on the shape of the period you test. In a strong sustained uptrend, buy and hold usually wins because every later DCA buy or grid re-entry happens at a worse average price. In a choppy, range-bound market a grid bot usually wins because it repeatedly sells the same range for small realized profits while buy and hold just sits flat. In a sharp crash-then-recover (V-shape), DCA usually wins because it mechanically buys more coins during the cheap days without you needing to time the bottom. Run the same lookback window on BTC, ETH and SOL below — the ranking changes.
How is the grid bot backtest calculated? The grid spans the actual realized high/low of the lookback period, split into N geometric (constant % step) levels. Capital is split evenly across the N cells. Each day, a cell buys if the daily low touches its buy line and it isn't already holding, and sells if the daily high touches the line one step above and it is holding — realizing the profit back to cash. Any cell still holding at the end is marked to market at the final close. This uses daily high/low, not tick data, so it's a reasonable approximation, not an exact replay of what an exchange grid bot would have filled intraday.
Where does the historical price data come from? Live daily candles fetched directly from Binance's public klines API in your browser when you click Run backtest — the same OHLC data the exchange shows on its own charts. Nothing is precomputed or cached server-side, and no inputs or results are sent anywhere; the backtest math runs entirely on your device.