The problem is not the VaR, I have used the bootstrap method actually. The problem is the series. The portfolio is mainly benzene, the series is really volatile and the quality of the data I have is dubious.

I have a window of 5000 observations and the distribution is indeed quite fat tailed

Statistical models such as...? Any suggestions? Thanks!

<t>I have computed VaR for a portfolio of assets ( commodities: oil and derivatives). The problem is the VaR value is quite volatile. It can be one value today and double that the next day. ( The assets prices are volatile too). I managed to compute a different and more stable VaR that is basically ...

<r>Rmax,I hope this solves your issues with brenthttp://<URL url="http://www.platts.com/IM.Platts.Content/InsightAnalysis/IndustrySolutionPapers/olmi003_datedbrent.pdf"><LINK_TEXT text="www.platts.com/IM.Platts.Content/Insigh ... dbrent.pdf">www.platts.com/IM.Platts.Content/InsightAnalysis/IndustryS...

The discussion was more of a theoretical one so "why people care about Brent" is not really the issue. Plus, even though it may be less than 1% of the total supply, Brent is the Baseline for pricing most oil and oil derived commodities.

Don't worry I got that we were having a theoretical discussion I just wanted to say that the spread has been more volatile in the last years so I wouldn't put my money on a model for this spread

Sorry, i uploaded the image here

<t>Well, as you said we have to define arbitrage. You made the initial supposition that the spread should be 12$ which is not dependable. It could be more accurate to bet on the spread rather than betting on the futures (as brent and wti are highly correlated so your position is somehow hedged) but ...

again, arbitrage is not possible as they are different products. You would have to convince your counter parties that the product has the same chemical specifications.

As they are different products and the transportation costs are significant I don't see any arbitrage opportunities here.

It is useful to use lag operator polynomials for understanding VAR or VEC models, for example

<t>I had the same problem.The slight difference between 50D and 100% Moneyness is expected and here iswhy:delta = N(d1)where N(.) is CDF function for normal andd1 = log(forward/strike) / (vol * sqrt(expiry)) + vol * sqrt(expiry) / 2For 100% moneyness, we have forward == strike and thusd1 = vol * sqr...

You can read all about the problems of scaling VaR in Carol Alexander's Book- Market Risk Analysis Vol IV- Value at Risk Models