Understanding Bar Types

Learn what bar types are, when to use each, and how to choose the right one for your strategy or ML model.

BitPredict provides over 15 proprietary bar types. Unlike traditional time-based charts that close every fixed interval regardless of market activity, most BitPredict bar types close based on market events — making them inherently more information-dense and better suited for algorithmic strategies.

Navigate to the Market Data Explorer

Go to Data → Market Data Explorer in the left sidebar. Select any symbol (BTC/USDT recommended) and make sure you're on the Time bar type first so you have a baseline to compare against.

Understand Time Bars

Time bars are the standard: each bar closes after a fixed time period (1h, 4h, 1D, etc.). They're familiar and easy to interpret but have a major flaw: a 1h bar at 3am on a Sunday contains far less information than a 1h bar during peak US trading hours, yet both appear identical in size on the chart.

Switch to Dollar Bars

Dollar bars close when a fixed dollar value of the asset has traded. This means bars during high-activity periods are shorter in clock time, and bars during low-activity periods are longer. The result: each bar contains roughly equal market information. Dollar bars are excellent for momentum strategies because the signal density is uniform.

Try Volume Bars

Similar to Dollar bars but close on fixed traded volume rather than dollar value. Good for assets where price is relatively stable but trading activity varies significantly.

Explore Tick Bars

Close after a fixed number of price ticks (individual trades). Best for ultra-high-frequency strategies. Very noisy at small tick counts but extremely granular.

Check the Data Quality Page

Go to Data → Data Quality. Each bar type has a ML Suitability score (0–100) with a 7-criterion pass/fail checklist. Before using a bar type for strategy development or ML training, check that its score is above 70 and that it passes the autocorrelation and variance ratio tests.

Use the right bar type for your strategy

Trend-following strategies: Dollar or Volume bars
Mean-reversion: Time bars at higher timeframes
High-frequency: Tick or Imbalance bars
ML model training: check ML Suitability score first

Dollar bars are generally the best starting point for any strategy that will be trained on or evaluated with ML models. They produce the most stationary return distributions and score highest on the ML Suitability checklist.

Don't use Tick or Imbalance bars with a short history for backtesting — their higher granularity means fewer total bars over a given time period, which can result in statistically unreliable backtest results.

What's Next

DataBeginner

Reading Market RegimesUnderstand BitPredict's four-state AI regime system and how to use regime data to make better strategy decisions.

6 min read

BacktestingBeginner

Run Your First BacktestConfigure and run a backtest from scratch, read the results, and understand what the key metrics are telling you.

5 min read