Title:

Optimal Prediction Pools, CQER Working Paper 09-05

PREVIOUS | NEXT

Optimal Prediction Pools, CQER Working Paper 09-05

DATE: October 26, 2009

PART OF: AUTHOR:

ABSTRACT: A prediction model is any statement of a probability distribution for an outcome not yet observed. This study considers the properties of weighted linear combinations of n prediction models, or linear pools, evaluated using the conventional log predictive scoring rule. The log score is a concave function of the weights, and, in general, an optimal linear combination will include several models with positive weights despite the fact that exactly one model has limiting posterior probability one. The paper derives several interesting formal results: For example, a prediction model with positive weight in a pool may have zero weight if some other models are deleted from that pool. The results are illustrated using S&P 500 returns with prediction models from the ARCH, stochastic volatility, and Markov mixture families. In this example models that are clearly inferior by the usual scoring criteria have positive weights in optimal linear pools, and these pools substantially outperform their best components.

Download (pdf) View Full Text

Share this page:


Diversity is critical to the Federal Reserve, and we are firmly committed to fostering a diverse and inclusive culture throughout the Federal Reserve System. Collections within FRASER contain historical language, content, and descriptions that reflect the time period within which they were created and the views of their creators. Certain collections contain objectionable content—for example, discriminatory or biased language used to refer to racial, ethnic, and cultural groups. These viewpoints and attitudes are inconsistent with our values, but the original descriptions are retained to ensure that they are not erased from the historical record.
Save & Share
less

Federal Reserve Bank of Atlanta. "Optimal Prediction Pools, CQER Working Paper 09-05" in Center for Quantitative Economic Research Working Papers (October 26, 2009). https://fraser.stlouisfed.org/title/7165/item/656234, accessed on December 2, 2023.

More Information
less

NOTE: JEL classification: C11, C53

Key words: forecasting, GARCH, log scoring, Markov mixture, model combination, S&P 500 returns, stochastic volatility

SUBJECTS: THEME:
Metadata (XML)
Metadata (JSON)