Why Volume and Dollar Bars Are Hard to Apply Across Multiple Days
Summary
The document compares fixed-time sampling with event-based sampling. Time bars record prices or returns at regular intervals, while tick, volume, and dollar bars are formed when a trade occurs or a specified quantity or traded value accumulates. It asks whether the latter methods can also be used with daily, weekly, or monthly data.
The response highlights an aggregation problem: a bar defined by accumulated activity can begin on one day and finish partway through another. Its span may therefore combine very different trading conditions, such as ordinary activity and an earnings announcement or broad volatility. That makes such a bar difficult to compare with bars covering different periods. The answer also notes that multi-day summaries are often treated as moving averages rather than multi-day OHLC bars. This is a conceptual caution, not a comprehensive evaluation; it does not rule out alternative designs or provide empirical tests of their usefulness.
Key ideas
- Tick bars, volume bars, and dollar bars use trading activity rather than fixed clock intervals to set observations.
- A multi-day activity bar can cross calendar boundaries and combine unlike market conditions.
- Bars that cover different periods may not be comparable when activity varies sharply.
- Multi-day summaries are often expressed as moving averages rather than OHLC ranges.
Tags
Full text
# Volume bars, dollar bars from low-frequency data? # Volume bars, dollar bars from low-frequency data? Financial models by default use time bars of prices/returns for input data. I use time bars to refer to both intraday (high frequency) and interday (low frequency) data since the sampling occurs at fixed intervals of time for both intraday and interday time bars. Instead of time bars, the alternatives are (1) tick bars, (2) volume bars and (3) dollar bars which sample prices/returns based on the occurrence of (1) a new transaction, (2) a fixed volume threshold, and (3) pre-defined monetary amount traded, respectively. I have only seen these alternative types of bars applied to intraday data (seconds, minutes, hours). Can't the same premise of volume bars and dollar bars be applied to interday data (daily, weekly, monthly)? ## Answer by chrisaycock (score 2) https://quant.stackexchange.com/a/54841 Multiday aggregations are usually moving averages, not OHLC. For example, I might get the 10-day MA; I would not ask for the 10-day high. (Charts might take the latest sample, but that's only for visualization.) As for multiday bars that aren't defined by time, I'm not sure how to interpret something like that. The bars would span multiple days and then cut-off in the middle of the day. Consider the below example; bar #2 begins in day #1 and ends in day #2. ``` day: |-----1-----|-----2-----|-----3-----|-----4-----| bar: |---1---|---2---|---3---|---4---|---5---|---6---| ``` Trading could vary wildly between those two days, like with earnings announcements or broader market volatility. So bar #2 would not be comparable to bars #1 or #3.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.