How Factor Models Inform Allocation and Risk Management

Summary

The document distinguishes factors as systematic sources of risk from signals that directly generate alpha. In response to questions about discovering and trading factors, it presents value, momentum, and quality as examples of factors identified through accumulated academic research. The answer cautions against treating factor modeling as a recipe for finding profitable trades.

Within the arbitrage pricing theory framing, apparent alpha after accounting for a chosen set of factors may indicate that relevant sources of risk have been omitted. The response instead describes factor exposure as useful for portfolio allocation and risk management, with allocations adjusted across factor styles as market conditions change. It gives no quantitative selection rules, trading example, performance evidence, or specific method for estimating exposures, so it offers a conceptual distinction rather than an actionable strategy.

Key ideas

• Factors represent systematic sources of risk that investors may be compensated for bearing.
• Value, momentum, and quality are examples cited from established research.
• A model’s apparent alpha may reflect omitted risk factors under the stated framework.
• Factor exposures can inform asset allocation and risk management.
• The document provides no implementation recipe or empirical performance evidence.

Tags

# Alpha generation and factor models


# Alpha generation and factor models












I have studied factor models in a very introductory manner, going through there Fama-French model and then APT. I understand the concept of decomposing returns into factors, but I don't understand how funds can trade on the basis of these factors. My question is then twofold:

1) How can you find factors? Say we are trading Equity or Commodities, from the time series of companies or of commodity prices, how do we proceed to quantitatively delineate factors to trade on?

2) Once we have these factors, how can we generate alpha from them?

I know successful models are in proprietary hands, but if anyone has an example that resembles as closely as possible what is usually done in practise, that would be very appreciated.

Thanks

## Answer by Chris (score 1)

https://quant.stackexchange.com/a/49635

You don't 'find' factors. Factors are systematic sources of risk the market rewards investors for holding. They're based in research done by academics over the years and include things like value, momentum and quality.

You don't generate alpha from factors. In fact, using the APT framework, with a given set of factors, if your model suggests the existence of alpha, it means you're missing risk factors.

In short, factors aren't really a 'trading' mechanism...they're more commonly used in asset allocation (eg, being more heavily allocated to quality and value in a downturn and momentum and low volatility in an upturn) or in risk management.

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.