Skip to content
All library documents

How Econometric Models and Neural Networks Relate in Forecasting

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

Summary

The discussion asks whether econometric time-series models and neural networks can be compared as competing approaches to forecasting, including by measures such as error and accuracy. Its central explanation is that the categories overlap: familiar econometric models can be represented as restricted neural networks, including networks with no hidden layer and matching inputs.

Neural networks allow a broader range of model structures, but that flexibility can make them harder to train and more prone to overfitting. The answer also cautions that a study comparing one particular network configuration with traditional time-series models does not establish which entire model class is superior. The document gives a conceptual distinction rather than empirical results or a general performance ranking; conclusions depend on the specific models and comparison setup.

Key ideas

  • Econometric time-series models can be viewed as special cases within broader neural-network model families.
  • A neural network can use no hidden layer and the same inputs as a linear econometric model.
  • Neural networks offer more modeling flexibility, with added training and overfitting challenges.
  • Results from comparing one network configuration with time-series models do not settle which model class is generally better.

Tags

Machine learningStatisticsBacktesting
Full text
# Econometric vs ANN models for forecast?


# Econometric vs ANN models for forecast?












I hope this is an appropriate question for this forum... for me it is an obvious query since it intrigues me for a long time.

Ok, assume there are 2 distinct classes of models: econometric (AR, MA, ARIMA, ARCH, GARCH, EGARCH, TGARCH, ...) and neural networks (MLP, RBF, BPTT, TDNN, Elman, NARX, ..., I'm putting even SVM and SVR into this group).

I know it is a broad subject - depends on the market, assets, for a start... but under what conditions one is better than the other? Is there a general consensus over this? In terms of MSE, R2, accuracy and so on? Is it fare to compare them? Does it make sense?

I've seen many studies doing this kind of comparison, here is an example (sorry for this being biased towards one side). But none summarizing previous conclusions on this topic.

Finally, what is your experience with both of them? Do you have other articles running this kind of test (even if not published)?

Thanks in advance, DBS.

## Answer by Ram Ahluwalia (score 6, accepted)

https://quant.stackexchange.com/a/2696

They are not mutually exclusive. For example, the class you refer to as "econometric" are simply linear regression models that include as factors prior returns or residuals of the return series sometimes with weightings on the observations.

You could easily design a neural network with no hidden layers and the same inputs. So each of the econometric models are special cases of neural network models. Neural Networks offer a broader class of modeling options although for the same reason they are more difficult to train and avoid overfitting.

In the paper, the authors are not comparing neural networks as a class to the econometric models. They are comparing a very specific neural network configuration to traditional time-series models.

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.