Using Cross-Asset Price Models to Fill Asynchronous Market Data

Summary

The document considers whether predicting an asset's current price from other assets has practical value when the market already reports that price. The responses identify data alignment and missing observations as possible uses. A model could estimate unavailable prices to backfill or interpolate time series, or help align closing prices for assets that trade in different time zones.

At tick level, assets update at different times, so cross-asset information might estimate a price before that instrument records its next trade. The discussion suggests performance may depend on how frequently the target asset updates, but provides no tests, model details, or measured results. These estimates are substitutes for missing or asynchronous observations, not a reason to disregard available market quotes. Their accuracy and suitability for trading would need to be assessed for the intended data frequency and use.

Key ideas

• Cross-asset models can estimate prices when target-asset observations are missing.
• Price estimates can help backfill or interpolate market time series.
• Different trading hours can leave daily closing prices asynchronous and difficult to align.
• At tick level, a model may estimate a price before the asset records its next update.
• The document gives possible applications but no empirical validation or performance results.

Tags

# Is there any utility to being able to predict an assets current price?


# Is there any utility to being able to predict an assets current price?












I was playing around with some models, and I'm able to predict a stock's current price based on the current prices of other stocks. This model is extremely accurate, although I can't see any use of this. If I wanted to know a stock's current price I would just grab it from the market, not derive it from other stock prices.

Does anybody know if a model that does this has any utility at all? Thanks.

## Answer by Kermittfrog (score 3)

https://quant.stackexchange.com/a/71043

You can provide estimates for missing data. Backfilling / interpolation of market time series is a common problem in the industry.

## Answer by Pontus Hultkrantz (score 1)

https://quant.stackexchange.com/a/71051

Adding on to Kermittfrog's answer: it might be that you only have cash close prices for stocks that trade in different time zones, hence you only have asynchronous daily prices which you might need to make aligned for various purposes.

If the model works on tick level and since stocks do not tick synchronously, you can predict the next price on stocks that have not yet "ticked". It could be that it works better (or worse) on stocks that have fewer price updates per time unit.

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.