QuoteIf you want, I can send you a couple of very good references. n'hésitez pas à nous en faire part. merci.

- frenchyWill
**Posts:**59**Joined:**

Je vois qu'il y a quelques français par ici (ou du moins des francophones) : nous finirons par conquérir toutes les salles des marchés So, for the papers, the best ones I've read on the computation of risk measures using heavy tail distributions are :-"Extreme Value Theory for Risk Managers", McNeil (1999)-"Estimation of Tail-Related Risk Measures for Heteroscedastic Financial Time Series", McNeil and Frey (2000)and I strongly recommend this book : "Quantitative Risk Management", McNeil, Frey and Embrechts (2005).

Last edited by frenchyWill on June 29th, 2010, 10:00 pm, edited 1 time in total.

If anyone is seriously interested in having a good model of asset price distribution, including a quantitative model of fat tails, based on elementary principles governing market inefficiency, pm me.

- Beachcomber
**Posts:**140**Joined:**

There is a lot of information at www.gloriamundi.org on this stuff.Many papers on extreme value theory and fat tail risk. Can't vouch for overall quality; I've only read a couple of papers.

QuoteOriginally posted by: FermionIf anyone is seriously interested in having a good model of asset price distribution, including a quantitative model of fat tails, based on elementary principles governing market inefficiency, pm me.No model is much better than any other. Prices move a lot. To say one model has better detail, or is more accurate so far as giving probabilities, is over-confidence.If you have a bigger margin, you will have a lower chance of going bust. But nobody can say what that chance is beyond the obvious. More specifically, nobody can consistently mass-produce measurements of that chance any better than the obvious.If one guy's tail is fat 10% deeper than another guy's, neither is right or wrong. One guy is safer than the other, but not more right than the other.Aren't I fucking deep today?

QuoteOriginally posted by: farmerAren't I fucking deep today?Oh wait, I am deep. It is some kind of epistemiological thing or something.

QuoteOriginally posted by: farmerQuoteOriginally posted by: FermionIf anyone is seriously interested in having a good model of asset price distribution, including a quantitative model of fat tails, based on elementary principles governing market inefficiency, pm me.No model is much better than any other. Depends on how much "much" is and for what purpose. A model that has a power law will be better for many purposes than one that doesn't. And having the right power is even better.QuotePrices move a lot. To say one model has better detail, or is more accurate so far as giving probabilities, is over-confidence.Or having a better model. Of course, no model will give an accurate mean, since that depends on unknown information. But a model that has the right shape will give better consistency of derivative valuations in a hedged portfolio than one that has the wrong shape.QuoteIf one guy's tail is fat 10% deeper than another guy's, neither is right or wrong.....Aren't I fucking deep today?If some one has a black hat and someone has a white that, then neither is right or wrong. Now that is deep.Of course, if black hat's win money and white hat's lose money, then it's a different story.

QuoteOriginally posted by: FermionOf course, if black hat's win money and white hat's lose money, then it's a different story.We are talking about risk management, not options trading. Even if we were discussing options trading, I have never seen a study that showed people who sold deep options all the time were net losers. Can you show me one, a study showing people who used one model or another for the tails were net losers? If a guy sells options and makes money every year, some arguments about probability are meaningless.

QuoteOriginally posted by: farmerQuoteOriginally posted by: FermionOf course, if black hat's win money and white hat's lose money, then it's a different story.We are talking about risk management, not options trading.You can talk about what you like. Why would you assume that risk management involves a portfolio that does not include derivatives? (It's not only plain vanilla options that depend on future price distribution.)QuoteEven if we were discussing options trading, I have never seen a study that showed people who sold deep options all the time were net losers. That's may be what market-makers do. They compensate for risk with the bid/ask spread. Quote Can you show me one, a study showing people who used one model or another for the tails were net losers? If a guy sells options and makes money every year, some arguments about probability are meaningless.Until they blow up. A rational person would not merely be worried about whether they won or lost, but by how much and what their risk of blow-up was. Of course, if you're trading with someone else's money and you get a fat bonus if you win and they take all the loss when you blow up, then trade away!The same goes with any derivative trading. If you have someone else's money to trade, then you are made.

Hi folks, I just want to get a single number "modified standard deviation" which captures fat-tail risk, etc.Any thoughts? Thanks

- frenchyWill
**Posts:**59**Joined:**

What do you mean by "fat tail risk" ?

QuoteOriginally posted by: frenchyWillWhat do you mean by "fat tail risk" ?the blow-ups... the tail is fatter the normal distribution...

- frenchyWill
**Posts:**59**Joined:**

So, by "fat tail risk" you mean kurtosis...

Last edited by frenchyWill on July 14th, 2010, 10:00 pm, edited 1 time in total.

Last edited by quartz on August 8th, 2013, 10:00 pm, edited 1 time in total.

I think the most practical solution is to invest in a simple trend-trading portfolio. This would be one that is expected to lose money in the average year, and does not do a lot of fancy stuff which increases profitability, but may miss trends. An all-in-all-the-time system like Dunn Capital would be good. He could customize it to weight stocks and interest rates in a way that matches your portfolio.Then your tail risk would be converted almost completely, in my opinion, into a known sum. It would be the expected losses each year in the trend trading portfolio, maybe 5%.You are not trying to get rich with deep OTM options or something. You are taking a more direct approach than position sizes for maximum allowable loss, combined with stop loss and liquidation rules. All risk management is basically trend trading. Having a specialized trend portfolio could perhaps reduce the management complexity of risk control, and the resulting risk of risk-system failure.Instead of bothering your mortgage-instrument structurers or whoever about risk every fucking day, you just reduce their salaries 5% and blow it in a trend-trading system.

GZIP: On