Code covered by the BSD License

5.0
5.0 | 1 rating Rate this file 14 Downloads (last 30 days) File Size: 1.19 MB File ID: #42280 Version: 1.0

# Monte Carlo example of the Multi-Factor coupled Commodity Forward curves Simulator

### Ahmos Sansom (view profile)

Implementation of the Multi-Factor multi commodity forward curve simulator

File Information
Description

1. Introduction

The attached matlab code simulates future coupled forward curves based on the Clewlow and Strickland model detailed in [1]. The aim of the code is to highlight how Matlab's optimisation routines can be used to improve the calibration process.

2. Running the code

The script:
MultiFactorExample.m

can be run to show how to the model is initialised that will output several figures highlighting simulations and validation.

The main engine determining the simulated forward curves:

MultiFactorForwardCurveSimulator.m

3. References

[1 "Multi-Factor Mult-Commodity models & Parameter Estimation Processess,” John Breslin, Les Clewlow, Chris Strickland, Daniel van der Zee, Lacima, 2008.

Required Products Econometrics Toolbox
Optimization Toolbox
Statistics and Machine Learning Toolbox
MATLAB release MATLAB 8.0 (R2012b)
Other requirements May have some memory restrictions
Tags for This File   Please login to tag files.
Comments and Ratings (2)
15 Jan 2014 Ahmos Sansom

### Ahmos Sansom (view profile)

Thanks for the comments, I'll check the simplification! There is spot price model in the technical note by Blanco and Pierce in Energy Risk, May 2012 that described how forward curve simulations can be used in a mean reverting spot model, i.e. the prompt month is used to mean revert to a spot price.

I'll contact you with further info and probably post the model when I convert to Matlab...

Comment only
11 Jan 2014 Martin

### Martin (view profile)

Excellent submission.

A quick suggestion would be that in computing the optimum correlations Matrix, a cleaner way of doing this is [Inverse of Va]*[Covab]*[Inverse of the Transpose of Vb]. This gives the exact solution without needing the optimisation.

One quick question if I may. If the modeller wishes to extract the spot price, is this as simple as simulating the price path off the simulated prompt month with a contract expiry of now?