Hi, I am interested in getting to know some commonly used term structure models used in the industry. In general I found two parallel approaches, work by Schwartz et al that focuses on stochastic modeling of the relevant factors, and some papers on PCA modeling. Did not fnd something that tries to combine the two, not even sure if there is value.Any experts out there who can shed more light, I'd be very grateful.Thanks.

Stochastich convenience yield models (e.g. Gibson-Schwartz) can be useful if your goal is to price options. I'm not sure if there is something like a (de-facto) industry standard, options markets are quoted in premium, not volatility points.PCA might have it's uses for either risk management or forecasting purposes, though admittedly I haven't been involved in either for myself.Maybe you'll elicit more responses if you can explain what your goal is (e.g. pricing options, VaR computation, forecasting,...)? Also I remember this question being asked before, did you try the search function?

thanks a lot bramJ, the goal I would say is definitely to forecast, especially the far out contracts that are illiquid, the goal is to understand what drives their price process.why do you think pca may be better than a structural model approach like that of gibson, is there a way to combine the two?thanks a lot again.

also this seemed interestingwww.utstat.utoronto.ca/sjaimung/papers/VAR-FPCA.pdf

No body, i am sure there are experts out there

really now, as BramJ mentioned, there is no point in term structure modelling in commodities unless you want to price exotics. There is nothing that drives the price "process" of illiquid points on a wti futures curve, apart from fundamentals. Term structure modeling is a crude approximation of the dynamics of the curve that can help simulate american/european/asian option prices for far out futures contracts, while potentially maintaining their volatility and correlation structure.

ok thanks, I am sure there are players out there who successfully make money trading the back months which are as we discussed fairly illiquid, they are maybe using some kind of a factor model to do that, so wasn't sure what the class of these strategies are, and what kind of models work.

QuoteOriginally posted by: Caesariareally now, as BramJ mentioned, there is no point in term structure modelling in commodities unless you want to price exotics. There I have to disagree with you. There are quite a few vanilla options traders who take a view on longer term, which involves some kind of analysis of term structure (with varying degrees of formality).

thanks tw, would you be able to elaborate more on what they do, any mathematical details you could share?

Well, without wanting to be too explicit, front vs back vol spreads are a fairly standard strategy.I'm sure you know all the standard models by which a volatility term structure might arise, so it shouldn't be too difficult to see how such spreads can perform under the different assumption implicit in these models and the implications for volume weighting them. QuoteOriginally posted by: deskquantthanks tw, would you be able to elaborate more on what they do, any mathematical details you could share?

GZIP: On