The question comes up constantly: should I use tick data or OHLCV candles? Most strategies do not need tick data. But some absolutely do — and using candles instead produces results that are fundamentally wrong.
When OHLCV Is Sufficient
- Trend-following strategies based on moving averages, RSI, MACD
- Breakout strategies using support and resistance levels
- Mean-reversion on daily or hourly timeframes
- Grid trading bots and portfolio rebalancing
When You Need Tick Data
- Order flow strategies: buy vs sell volume imbalance within each period
- Scalping and HFT simulation: execution at specific price levels within a candle
- Market making backtests: spread capture depends on the sequence of trades
- Slippage modeling: accurate fill prices for large orders
Storage and Performance Tradeoffs
A one-year BTC OHLCV dataset at 1m resolution is roughly 18 MB in Parquet (our full 6.5-year BTC 1m file is 117 MB). The equivalent tick dataset for the same period is around 3.6GB — over 100x larger. For most backtesting workflows, OHLCV loads in milliseconds; tick data requires streaming or chunked processing.
Summary
Use OHLCV if your strategy fires signals at candle boundaries and does not depend on intra-candle dynamics. Use tick data if you are studying order flow, simulating market making, or building any system where the sequence and timing of individual trades matters.