Nonparametric Risk Bounds for Time-Series Forecasting
Authors: Daniel J. McDonald, Cosma Rohilla Shalizi, Mark Schervish
JMLR 2017 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We motivate our techniques with and apply them to standard economic and financial forecasting tools a GARCH model for predicting equity volatility and a dynamic stochastic general equilibrium model (DSGE), the standard tool in macroeconomic forecasting. We demonstrate in particular how our techniques can aid forecasters and policy makers in choosing models which behave well under uncertainty and mis-specification. We now show how the theorems of the previous section can be used both to quantify prediction risk and to select models. We first estimate a simple stochastic volatility model using IBM return data and calculate the bound for the predicted volatility using Corollary 18. Then we show how the same methods can be used for typical macroeconomic forecasting models. |
| Researcher Affiliation | Academia | Daniel J. Mc Donald EMAIL Department of Statistics Indiana University Bloomington, IN 47404; Cosma Rohilla Shalizi EMAIL Mark Schervish EMAIL Department of Statistics Carnegie Mellon University Pittsburgh, PA 15208 |
| Pseudocode | Yes | Algorithm 1: Kalman filter. Recursively generate minimum mean squared error predictions b Yi using the state space model in (1). ... Algorithm 2: Steady-state approximate filter. Recursively generate approximate minimum mean squared error predictions b Yi using the state space model in (1). |
| Open Source Code | No | The paper does not provide any explicit statements about releasing source code or links to a code repository for the methodology described. |
| Open Datasets | Yes | We estimate a standard stochastic volatility model using daily log returns for IBM from January 1962 until October 2011 (n = 12541 observations). ... To estimate the parameters of this model, we use four data series. These are GDP yt, consumption ct, investment it, and hours worked nt which are from the Federal Reserve Economic Database. |
| Dataset Splits | No | The paper discusses 'training error' and 'in-sample forecasts' but does not specify how the datasets were split into training, test, or validation sets, either by percentages, sample counts, or predefined citations for reproducibility. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware used for running its experiments. |
| Software Dependencies | No | The paper does not specify any software dependencies with version numbers. |
| Experiment Setup | No | The paper describes model estimation methods such as 'maximizing the likelihood' and 'maximizing a penalized likelihood', and specifies parameters for its theoretical bounds (e.g., ς = 9.62, φ = 0.996, σ2 η = 0.003, µ = 538, a = 11, d = 2, M = 0.1, η = 0.05). However, it does not provide specific experimental setup details like learning rates, batch sizes, number of epochs, or optimizer settings for training the forecasting models. |