Why OHLCV Data Cannot Directly Identify Bid and Ask Volume

Summary

Bid volume and ask volume refer to trades executed at the bid and ask prices. The note asks whether these quantities can be approximated when only open, high, low, close, and total volume are available. Its response says a model can be constructed, but the form must be specified rather than inferred directly from the available fields.

The main lesson is that constructing an estimator is only part of the task: it also needs a credible validation method. The answer offers no particular functional form, empirical results, or validated approximation. It reports that the author could not identify relevant published work, while acknowledging that search terms may have missed it. Any proposed method therefore needs a defensible account of how volume relates to price and testing against suitable data before its estimates can be trusted.

Key ideas

• OHLCV bars do not directly report how much trading occurred at bid versus ask prices.
• A model-based estimate is possible in principle, but its functional form must be developed.
• Validation is a separate and essential challenge from constructing an estimator.
• A proposed model should explain how volume affects price and then be tested empirically.

Tags

# How to approximate bid and ask volume?


# How to approximate bid and ask volume?












https://www.tradingtechnologies.com/xtrader-help/x-study/technical-indicator-definitions/volume-delta-vol/

Bid and ask volume is defined as the volume traded at bid or ask prices.

Given only OHLCV data, how to approximate the bid and ask volume?

## Answer by Dave Harris (score 1)

https://quant.stackexchange.com/a/70514

The bigger issue is the validation of the method.

It is technically possible to model such things, I am providing an example that is used as a teaching tool for modeling three-sided coins (landing on the edge is included). https://artowen.su.domains/courses/306a-1011/threesidedcoin_amstat.pdf

However, you have to build a functional form yourself. Even if you did that, you have to find a way to validate the model.

Can it be done, yes it can be. Validating that it is a good idea or that it works, that is a different question.

EDIT

As far as I have been able to determine, nobody has done that in the literature; or, if a paper exists, then the search terms obscure it. As I said, you will have to build the functional form yourself and test it.

If you can explain to yourself how the volume data would impact the price data, then you can test a form.

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.