<t>"other observable market data supporting a valuation technique"This part does seem like you could drive a tractor trailer through it, so I'm not even sure you would even have to go the correlation route for baskets of stocks. Couldn't you just look at the historical volatility of your basket, the...

C'mon it's a certified market price. If regulators aren't going to give you a break on that one, then they might as well just pull the plug...

<t>Probably I'm missing the point, but if you want to know your correlation risk, alter the correlation matrix uniformly (+/-.1 pts in correlation). This tells you your risk. It does not tell you what your limit should be. For this refer to management, who will probably refer the question back to yo...

<t>1) If you have multiple entities, book the product or products outside the U.S. (Don't know if that works or not, but it might be worth a try!!)2) *If* you have only a few long-dated options (and long-dated options are your problem) and *if* you don't mind telegraphing to 2 or 3 large corporate b...

Well you have two possibilities: market price and theoretical price. Market price only works if your derivatives are exchange lookalikes and extremely liquid to boot. That leaves theoretical price.

<t>I swear this is a true story...I once held a position on a stock where my gamma was about half the daily volume. It was a plus position, and this went on for several months, so naturally the historic vol started tending to the single digits. Just after the maturity, I left on vacation for a week,...

I would say it shouldn't matter. If it did, people would simply adjust their behaviour by dissimulating the "real" order in the book, to take advantage of any increase in probability.

I know my bank got pressure from regulators to put VaR limits on specific option books, so I don't find it inconceivable that other banks might have caved in.

FDAX: no problem, on quantos you're right.

<t>As you know correl is not linear, so reserving at 100% correl when the correl is 85% is diffierent than reserving at 70% correl when the correl is 55%. It's not just 15 itty-bitty correl points. Reserving at 100% when the correl is 85% is equivalent to reserving at infinite volatility when the vo...

<t>FDAXHunter: Um, perhaps I missed something, but I think the original poster said nothing about USD/JPY and Nikkei 225. It was you who brought in this example. To abstract from one case during one period to the idea that you should mark every correlation to -1/+1 (for reserve purposes) is, I think...

<t>QuoteOriginally posted by: kwoksun2000Any of you have an idea on how to go about estimating dividend policy across different companies? I realize that many companies actively pursue a dividend smoothing policy. This throws off my model since it makes dividend increases and timing relatively inela...

<t>+1/-1 is too much for a sensible reserve, if e.g. an auditor is going to be looking at it. There is of course no standard, so you have to use your judgment what level you think you can justify to whomever you're going to have to justify it to. Phrasing it in terms of standard deviations (1, 2, or...

From the client POV you are probably paying through the nose for a CPPI.From the investment bank POV you have to make sure you are handling the crash risk correctly.

<t>"Probably the most contentious assumption would be the volatility. I'd ideally want to see an assumed volatility that is less than the implied volatility of traded options or historical volatility, to reflect that the options are freely hedged with treasury stock."Market (implied) vol seems the m...