Volume-Based Candlestick Aggregation (Medium) — Practice with Code Visualizer

In quantitative finance and algorithmic trading, traditional time-based candlestick charts (such as 1-minute, hourly, or daily candles) have a significant limitation: they treat all time periods equally, regardless of actual market activity. During low-volume periods, these candles may contain minimal price action, while during high-volume bursts, they may compress critical information into a single bar.

Volume-Based Candlestick Aggregation (also known as Volume Bars) offers an alternative approach used extensively in financial machine learning and high-frequency trading systems. Instead of forming bars based on fixed time intervals, volume bars aggregate trades until a cumulative volume threshold is reached, at which point a new candlestick is formed.

This methodology provides several key advantages:

Normalizes market activity: Each bar represents roughly the same amount of trading activity
Improves statistical properties: Volume bars tend to exhibit more consistent statistical behavior
Better signal extraction: Price movements are weighted by actual market participation
Enhanced machine learning features: Creates more uniformly distributed training data

OHLCV Format: Each candlestick bar is represented in the standard OHLCV format:

O (Open): The first trade price in the bar
H (High): The maximum trade price in the bar
L (Low): The minimum trade price in the bar
C (Close): The last trade price in the bar
V (Volume): The total cumulative volume in the bar

Bar Formation Rules:

Begin accumulating trades into the current bar
Track the open, high, low, close, and cumulative volume as each trade is processed
When cumulative volume reaches or exceeds the threshold, finalize the current bar and begin a new one
If trades remain that do not reach the threshold, include them as an incomplete bar at the end
Round all output values to 4 decimal places

Your Task: Write a function that takes an array of trade prices, their corresponding volumes, and a volume threshold, then returns a list of volume-based OHLCV bars.

Bar 1 Formation: • Trade 1: price=100.0, volume=50 → cumulative=50 (threshold not reached) • Trade 2: price=101.0, volume=30 → cumulative=80 (threshold not reached) • Trade 3: price=99.0, volume=40 → cumulative=120 ≥ 100 (threshold reached!)

Bar 1 closes with: Open=100.0, High=101.0, Low=99.0, Close=99.0, Volume=120.0

Bar 2 Formation: • Trade 4: price=102.0, volume=60 → cumulative=60 (threshold not reached) • Trade 5: price=98.0, volume=20 → cumulative=80 (threshold not reached) • Trade 6: price=100.0, volume=50 → cumulative=130 ≥ 100 (threshold reached!)

Bar 2 closes with: Open=102.0, High=102.0, Low=98.0, Close=100.0, Volume=130.0

With uniform volumes of 25 each and a threshold of 50:

Bar 1: Trades at 10.0 (vol=25) and 12.0 (vol=25) combine for exactly 50 volume. Result: Open=10.0, High=12.0, Low=10.0, Close=12.0, Volume=50.0

Bar 2: Trades at 11.0 (vol=25) and 13.0 (vol=25) combine for exactly 50 volume. Result: Open=11.0, High=13.0, Low=11.0, Close=13.0, Volume=50.0

When volume exactly equals the threshold, the bar completes on that trade.

The total volume across all trades is only 45.0, which is less than the threshold of 100.0. Since the threshold is never reached, all trades form a single incomplete bar:

• Open: 50.0 (first trade price) • High: 52.0 (maximum price) • Low: 48.0 (minimum price) • Close: 48.0 (last trade price) • Volume: 45.0 (total accumulated volume)

Incomplete bars at the end of the data stream are included in the output.