Friday, November 20, 2009

Commodities corner: Sugar futures bubble ready to burst - an update



Over the past few months, sugar has been on accelerating upward spiral, hitting a 28-year high at USD24 cents per pound. On September 1rd In analysis posted on this blog I stated that sugar prices exhibit a “bubble” characteristic. Fitting LPPL model I came to conclusion that bubble get into critical zone. The critical zone describes the maturation of a systemic instability forewarning of an inevitable crash. Since the beginning of September the sugar prices abandoned the „super-exponential” growth pattern and start to widely oscillate which goes along the prediction of the LPPL model





Here you can download a PowerPoint presentation. This presentation is aimed to show a more detailed analysis of the sugar#11 futures contract prediction and the methods used to make and test them. Specifically they are LPPL model, LPPL fitting procedure

6 comments:

foodjunkie said...

Hi,
Great post. I always read your posts with great interest. Thank you very much for the attached presentation.

I am just working my way through Sornette's Book (Why Stock Markets Crash). How did you find it? Any other must-read papers on this topic. I've looked at some of Sornettes published papers.

Working for a penison fund in New Zealand. SO quite a long way from Poland :).

Do widzenia (have a good polish friend so am familiar with a very limited number of words.)

Konrad Cheers

Piotr Chwiejczak said...

Kia Ora
I’m really happy that you like the blog. Speaking of “must read” I would suggest reading some of the Sornette (et all ) paper links here.
http://arxiv.org/PS_cache/cond-mat/pdf/0210/0210574v1.pdf
http://arxiv.org/PS_cache/physics/pdf/0301/0301023v2.pdf
http://arxiv.org/PS_cache/cond-mat/pdf/0209/0209591v2.pdf
http://arxiv.org/PS_cache/cond-mat/pdf/9901/9901268v1.pdf
http://arxiv.org/abs/cond-mat/0503607
http://arxiv.org/abs/cond-mat/0209065
http://arxiv.org/abs/cond-mat/0109410
Good luck in your research. Let me know if you need some more informations
Ka kite ono
Piotr
P.S. ( I also have some good friends in Kiwi-land)

Konrad Schaefers said...

Pretty good! That's more than I know myself :).
I'm originally from Germany so Kia Ora is all I know.

I will have a look at those papers, thank you very much for that.
I will probably always have more questions thogh :).

I find it very interesting that these models have not received more attention in the mainstream economic circles. It seems the economic academic establishment is the most arrogant and least open to new ideas out of all the sciences.

The general way to approach a model or theory would be to observe, hypothesize, and test, test, test, looking constantly for better alternatives or a way to disprove the theory/model. WHereas in economics the aaproach is to pick a theory, disregard all of the alternatives and in general ignore empirical evidence which does not support the theory.

If you look at the economic 'models' they are incredibly simple. Unbelievable reallly givne that an economic/financial system is probably one of the most complex systems out there. Yet somehow a simple demand/supply model is the solution to all problems. Scary in a way.

Anyways. I am working on trying to implement Sornettes models in Matlab, so that will probably keep me busy.

I heard Sornette himself is working on a big project to automatize the LLPL framework for a large range of securities, indices, etc (~50k)

Have you heard of the chaosmonitorblog.com?
Similar approach although much more short-term oriented. WHat are your thoughts on their approach and the blog itself.

Dobze, (is that how you spell OK?)
will post some more comments in the future. Had been reading your blog for a while now.

Konrad

Piotr Chwiejczak said...

I do apologize that i left you comment unasserted so long but I’m quite busy with my own project. My advice is to use patternsearch(), ga() and fittype() functions when implementing LPPL in matlab. It is not a very thoughtful solution but the LPPL equation is quite complex and sometimes you need to use brute force instead of M-N nonlinear algorithm. If you need any more guidance please write on my priv email address

Konrad Schaefers said...

I can't find your email address on your blog. please email me at kschaefers(at)annuitas.co.nz.

Thanks

Piotr Chwiejczak said...

Piotr.Chwiejczak@gmail.com