Framing Mean-Reversion Forecasts for Stocks and Portfolios

Summary

The document poses a research problem about predicting short-term price reversals in individual stocks and in a portfolio. The author observes that some stocks rise and then reverse, with the apparent pattern clearer in some names than others. They suspect supply becoming saturated relative to demand may be involved, while also recognizing that broad market moves affect the observed behavior.

The questions distinguish single-stock reversion after accounting for market conditions from portfolio-level reversion, where aggregated returns are the focus. They also ask how to express forecasts with confidence probabilities or intervals, including when future demand is unknown. No predictive method, data, empirical results, or definitions of reversion are supplied. Any investigation would therefore need to specify the horizon and reversal event, separate market-wide effects from stock-specific behavior, and account for the dependence among holdings before assessing portfolio uncertainty.

Key ideas

• The document asks how to forecast short-horizon reversals in individual stocks and baskets.
• It identifies broad market movements as a potential confounder of stock-specific reversion.
• A proposed supply and demand explanation is presented as an intuition rather than tested evidence.
• The request seeks probabilistic forecasts but provides no model, data, or definition of a reversal event.

Tags

# How to predict a portfolio's reversion?


# How to predict a portfolio's reversion?












Sorry if this has been asked before. I've been baffled by a question I'm facing.

Assuming I know there are some certain demands for some stocks in near future, and I put them in a basket as a portfolio. What I found is in short period, prices of some of them keep increasing and then at certain point revert. This reversion behavior is very clear for some symbols but less clear for others. General feeling is it has something to do with the saturation of the supplies to the demands. It's also coupled with the fluctuation of the overall market.

My perplexities are:

- How to predict the reversion of single stocks with some confidence probability/internal, with overall market condition excluded?

- How to do the same for the portfolio? After all, returns are measured by the aggregation of the portfolio.

- If it makes it easier to remove the assumption of known demands, then in general, how to predict a single name and a basket's reversion with confidence probability/interval?

Kind of lost where to start. Some leads or links of papers will be enough to point to some directions.

Thanks!

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.