Predicting the price of crude oil

madness

I am a (mathematical) physics student doing a project on predicting the price of crude oil, and was hoping someone here could help point me in the direction of some good resources (books, websites, journals etc). I am especially interested in resources which model things mathematically. The project outline is very vague, and I'm really just looking for somewhere to start at the moment. Thanks.

CRGreathouse

http://en.wikipedia.org/wiki/Efficie...ket_hypothesis
http://en.wikipedia.org/wiki/Technical_analysis

John Creighto

I'm not an Economists but I'd use basic principles of supply and demand. However, for the demand equalization I'd use a linearized utility function constrained by the GDP.

The Unitlity would be Given by the following equation:
1) [tex]U=U_o + (\nabla _x U)(X-X_o)[/tex]

Where:
[tex]U[/tex] is the Utility
[tex]U_o[/tex] is the utility at the point of linearization.
[tex]X[/tex] is the quality of goods purchases
[tex]X_o[/tex] is the quality of goods purchases at the point of linearization.

For [tex]X_o[/tex] I'd use trends in demand.

Since we can't measure utility perhaps set it to one and estimate [tex]U_o[/tex] and [tex](\nabla _x U)[/tex] from historical data.


The Supply funciton is more straight forward:

2)[tex]S=S_o+ (\nabla _p S)(P-P_o)[/tex]
[tex]S[/tex] is the supply
[tex]S_o[/tex] is the supply at the point of linearization.
[tex]P[/tex] is the price of the good.
[tex]P_o[/tex] is the price of the good at the point of linearization.

I'd Base [tex]P_o[/tex] on the price given in the futures market. I'd set [tex]S_o[/tex] based on the growth projection given by industry (i.e. planed future projects)

[tex](\nabla _p S)[/tex] is the gradient of the supply with respect to price. I'd probably just estimate this from historical data although alternatively one could look at how the marginal cost of production varies with the price and remember that the in equilibrium the marginal cost of production should equal the expected rate of return (say about 8%). Also remember that oil companies usually plan production based on a lower cost of oil then the market cost.


The GDP is given by:

3)[tex](GDP)=(GDP)_o+(X-X_o)^T(P-P_o)[/tex]
[tex](GDP)[/tex] Is the GDP
[tex](GDP)_o[/tex] Is the GDP at the point of linearization.

Again I'd set [tex](GDP)_o[/tex] based on the growth trends in GDP.

And the total value of demand is given by

4)[tex](P^TX)=(P^TX)_o+(X-X_o)^TA(P-P_o)[/tex]

equation 3) must equal equation 4)

[tex]A[/tex] reflects the marginal savings rate. Eignvalues greater then one suggest negative savings and eigenvalues less then one suggest positive savings. Note that a Positive savings rate means a negative money multiplier and a negative savings rate means a positive money multiplier.



So my basic proposed model would linearize about the trends and then check based on supply and demand weather we expect these trends to continue.

Edit: I think my proposed method is kind of like a predictor corrector method. Also notice the sensitivity of the equations to A. A small change in the savings rate could dramatically effect the equilibrium.

madness

Thanks for the replies, unfortunately they aren't exactly what i was looking for. The aim of the project is really to conduct some kind of literature review on the subject and compile a report based on various sources. The first reply is probably not mathematical enough to be acceptable for this project, and the second hasn't cited any academic sources. I'm finding it very difficult to actually find any detailed mathematical models used to predict the price of crude oil. I'm sure they are out there somewhere...

mgb_phys

Anybody with an accurate model for world oil prices wouldn't be publishing it - they would be sitting in a house made of diamonds while Bill Gates and George Soros mowed their lawn.

Then on top of regular supply-demand laws you have the political aspect. What if Obama dissed Celine Dion at the inauguration and Canada imposed an oil embargo while Venezuela invaded Iran.

madness

I didn't say I needed a perfect model to predict the price of crude oil. I'm looking for mathematical models, which I assume exist and have been published, otherwise my supervisor wouldn't have set the project. Supply and demand, the role of speculators etc are all things which can be modelled mathematically, and the reliability of such models is to be evaluated as part of the project.

CRGreathouse

Yeah, that was the reason I left a lin to the EMH in post #2.

Proton Soup

good luck with your math and market theories. oil isn't a free market.

http://en.wikipedia.org/wiki/OPEC

madness

So everyone agrees that I have been given an impossible project then? It is not even possible to find someone who has attempted to model oil price trends mathematically?

mgb_phys

Over the short term supply-demand equations will work. oil will generally track other economic indicators since it's demand is linked to industrial output (to fuel factories) and wealth (to run cars).
Longer term there will be hiccups due to artificial shortages (the 70s oil embargo following the Arab-Isreali war) and new discoveries (the North sea or oil sands coming on line)

madness

I have to stress again that finding a model which works in the long term in not my aim. I need access to any work in this area which I can evaluate and compile a report on.

mheslep

Madness - The US EIA appears to be far and away the leading voice on short term oil predictions; a task such as you have described must start there.
http://www.eia.doe.gov/emeu/steo/pub/gifs/Fig1.gif
They steadfastly predicted falling oil prices last Summer when oil was topping $150/bbl and the manic were predicting $200/bbl. EIA predicted the current oil surplus; they didn't capture the sharpness of the decline, but that surely must be assigned in part to the credit crash. EIA representatives also testified as much to Congress last Summer (Rep Markey's committee) and held to them despite incredulous responses from Markey.
http://www.eia.doe.gov/pub/forecasti...teos/jun08.pdf
They're now predicting a price rise to $50-60/bbl by 2010.

In particular, you'll want to dig into their petroleum supply and demand model equations here:
http://www.eia.doe.gov/emeu/steo/pub...nt/textsu.html
http://www.eia.doe.gov/emeu/steo/pub...nt/textpt.html

stewartcs

Here is some historical price information that may help.

http://www.oilnergy.com/1obrent.htm#since88

CS

Potential

T. Boone Pickens may provide the information you are after. He should have a public web site which is also advertised on TV occasionally. In order for him to discuss oil topics with congress, surely he had to invest personal funds in public research. If so, these documents should be available, maybe due to the Freedom of Information Act.

Phrak

How does one model OPEC inclinations and middle east termoil in a predictive manner?

Potential

One possible way would be to implement an artificial neural network to find non-linear patterns of historical oil data.

DrClapeyron

You could take into account such things as US food exports to the middle east, weapon sales to the middle east, and short and long term US Treasury interest rates. I would put emphasis on interest rates and financial derivatives in general.