Skip to content
All library documents

Approximate Dynamic Programming for Portfolio Choice

Article Quant Q&A · Author: donpresente
Read the original

Summary

The document points readers to approximate dynamic programming as a way to solve finance problems involving decisions over time, focusing on dynamic portfolio choice. Its central method is value function approximation: simulate possible paths and approximate the value of future choices to make a high-dimensional optimization problem more tractable.

It identifies research on return predictability, optimized portfolio weights, value function recursion, and state variable decomposition as relevant examples. It also recommends background in optimal portfolios and numerical function approximation, and notes that the broader dynamic programming literature may help. The post is a brief reading guide rather than a tutorial: it gives no implementation, algorithm steps, empirical results, or comparison of methods, and the respondent cautions that the cited work reflects a limited focus rather than a comprehensive survey.

Key ideas

  • Approximate dynamic programming can address sequential portfolio choice problems.
  • Value function approximation provides one route to solving dynamic portfolio decisions through simulation.
  • Related research studies return predictability and recursions based on portfolio weights or value functions.
  • State variable decomposition is cited as a numerical improvement for consumption and portfolio problems.
  • The references form an introductory path, not a complete survey or implementation guide.

Tags

Portfolio constructionMachine learningStatistics
Full text
# Do you have a good application example of Approximate Dynamic Programming?


# Do you have a good application example of Approximate Dynamic Programming?












Have you ever tackled a finance problem with Approximate Dynamic Programming?

I have only used dynamic programming for simple examples like a optimal extraction in mining.

- Do you have canonical reading material that you recommend?

- Do you have any code implementation example (in any language or pseudo code)?

## Answer by Bob Jansen (score 5, accepted)

https://quant.stackexchange.com/a/16541

I totally missed the coining of the term "Approximate Dynamic Programming" as did some others. Also, in my thesis I focused on specific issues (return predictability and mean variance optimality) so this might be far from complete. That's enough disclaiming.

Let's start with an old overview: Ralf Korn - Optimal Portfolios. Kenneth Judd - Numerical Methods in Economics gave me some good background on approximation of functions. You may not need it.

Brandt, M.W. et al. - "A Simulation Approach to Dynamic Portfolio Choice with an Application to Learning About Return Predictability" does what it says and uses value function approximation to do it. Van Binsbergen, J.H. and Brandt, M.W. - "Solving dynamic portfolio choice problems by recursing on optimized portfolio weights or on the value function?" improve upon this idea. Garlappi, L. and Skoulakis, G. - "Solving Consumption and Portfolio Choice Problems: The State Variable Decompostion Method" provide important numerical improvements.

The literature of Bertsekas should be of interest as his papers are often cited. However I haven't looked at them yet. I hope this interests you, if this is not what you meant, do tell.

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.