QuoteOriginally posted by: Gmike2000BIS and Basel Committee are bureaucrats, not traders. This discussion is funny, because it has no point. VaR is used by people who do not actually sit at the trading desk "front lines" where six sigma events are witnessed at least 10 times a year (today the bund future for no particular reason fell more than a figure within less than a minute...if you had a stop loss order sitting just below the market, you got shafted...I want to see your VaR forecast on that one!). I learned VaR under Philippe Jorion himself, and implemented a complex VaR system at a major investment house in my previous life, front to back, so I suppose I know VaR quite well. It is a good measure in the sense that it when you track it over time, you can get a rough indication of whether you are taking more or less risk, and you can decompose it to see where the risk supposedly comes from. So it is a quick and convenient measure to be used by upper management and other "decision makers" (or by regulators, shareholders, etc). However, VaR is completely inadequate for forecasting loss percentiles....forget about it. It also is inadequate for daily use on a trading desk, where you need to see the uncorrelated sensitivities for each position or risk factor.The reason there is so much debate on VaR is because it is being wrongly marketed as a way for forecasting losses. That is like saying Viagra adds 5 inches to your penis. It does not do that, but it still has its merits (for those who need it)....First off, for someone with ED, Viagra probably adds something close to 5 inches to their flacid member.Secondly, this is sort of the point I was hoping to make. I think that VaR has some merits, and in an imperfect world, it has uses as a policy tool for bureaucrats. I guess, though, that it's fundamentally easier to poke holes in a theoretical risk measure than to suggest a new one that outperforms it.Maybe I'm too dumb to understand the subtleties of VaR, but if that is the case, how the hell is a bureaucrat supposed to understand it? Or a salesman/woman? Or a flow trader? Or a financial journalist reporting on the markets? Or the first-year finance undergrads to whom it is taught?VaR is supposed to be well-liked by these people because it is easy to understand, not the other way around. You simply can't have it both ways--either it's uncomplicated or it's not. It can't simultaneously be too simple and too complex for the same person.

QuoteOriginally posted by: ppauperQuoteOriginally posted by: TraderJoeQuoteOriginally posted by: ppauperso we've got nay votes from bb1220 and TJ for VaR, and yes votes from the Bank for International Statements and the Basel Committee on Banking Supervision.of course VaR has shortfalls (for example, bb's list and many more would be listed and discussed in any standard text on risk management) and part of effective risk management is knowing the limitations of any tools you use. When there are failures, more often than not it is because the tools are being inappropriately applied by folks who lack the intellectual capacity to understand themLimits need in place for the various risks banks and traders take, and unless and until we come up with a better measure of risk, VaR is the gold standard.VaR, like Black-Scholes, provides an unrealistic and overly simplified model of the real world. Constant vol anyone?you continue to demonstrate that you lack the intellectual capacity to understand even rudimentary risk management.And you continue with the insults. I'm glad you're not managing my money. Ha!

QuoteOriginally posted by: Gmike2000I learned VaR under Philippe Jorion himself, So that means you did not attend a top 10 school?QuoteThe reason there is so much debate on VaR is because it is being wrongly marketed as a way for forecasting losses. If it's not good for forecasting losses -- what is it good for?

QuoteOriginally posted by: KackToodlesQuoteOriginally posted by: Gmike2000I learned VaR under Philippe Jorion himself, So that means you did not attend a top 10 school?QuoteThe reason there is so much debate on VaR is because it is being wrongly marketed as a way for forecasting losses. If it's not good for forecasting losses -- what is it good for?Exactly!

QuoteOriginally posted by: TraderJoeQuoteOriginally posted by: ppauperyou continue to demonstrate that you lack the intellectual capacity to understand even rudimentary risk management.And you continue with the insults. I'm glad you're not managing my money. Ha!you don't have any money, TJ

QuoteOriginally posted by: KackToodlesQuoteOriginally posted by: Gmike2000I learned VaR under Philippe Jorion himself, So that means you did not attend a top 10 school?QuoteThe reason there is so much debate on VaR is because it is being wrongly marketed as a way for forecasting losses. If it's not good for forecasting losses -- what is it good for?forecasting risk !

QuoteOriginally posted by: ppauperforecasting risk ! Does risk have any realization? If not, then how can your boss know if your forecasts are any good? It's easy to forecast something that nobody can check later to see if you were right or not.

QuoteOriginally posted by: KackToodlesQuoteOriginally posted by: ppauperforecasting risk ! Does risk have any realization? If not, then how can your boss know if your forecasts are any good? It's easy to forecast something that nobody can check later to see if you were right or not.Although its true that a single event provides no basis for assessing a single risk forecast, one can assess a population of forecasts over a population of futures. If future drawdowns exceed the expected VaRs too often, then one can say the VaR process was optimistic. And if one is willing to model the form of the distribution of tail events, then one can assess the quality of a VaR even in the absence of events on the wrong side of the VaR expectation. Likewise, any forecast of expected future volatility can be tested against realized volatility. It is true that normal VaR provides no true worst case estimate of losses -- it only gives a confidence interval that losses won't exceed some value. For worst case losses, one needs methods such as conditional VaR or expected shortfall. But it's not clear to me that expected shortfall is meaningful or computable if the true future higher moments of the distribution are undefined or potentially astronomical (e.g., the expected shortfall for a Cauchy-distributed returns). There will always be events that break the system (see the bogey-man thread) which mean that one always faces a threat of total loss (or worse if one is a housing lender that gets stuck with a negative recovery ratio on a mortgage due to the intervention/lawsuits by local governments).

Bear Stearns <=> Enron. No amount of VaR or any other risk management would have prevented its collapse. Haha!

QuoteOriginally posted by: Traden4AlphaAlthough its true that a single event provides no basis for assessing a single risk forecast, one can assess a population of forecasts over a population of futures. too bad your boss has to assess your compensation based on a single event. what is he suposed to do? take your word for it and give you a bonus based on an imaginary "population"?

Just saw this now, interesting column by Michael Lewis, at the end the question is the same that torontosimpleguy raised:What Wall Street's CEOs Don't Know Can Kill You: Michael LewisThe final paragraphs:QuoteAt this point you have to at least wonder if Wall Street firms should be public companies. Their complexity renders them inherently opaque. Investors are right now waking up to this fact: They will demand to be paid for opacity, and also for volatility. The firms have been revealed to be so treacherous in bad times that the only way they survive as public companies is to make outrageously huge sums in good times. That is, as public companies, to be economically viable they are likely to be socially problematic. If they aren't about to go under, they are making so much money that everyone else hates them. Something is about to give.

In a free market people should be allowed to invest in whatever they want - if they want to take the risk in investing in a Broker Dealer they should be allowed - I don't need to tell anyone on this forum about the relationship between risk and return - if you invest in these firms you are taking a risk easy as that. I like Lewis a lot, but not sure he has got it right here.<rant>VaR makes me laugh. It is one of the best cases of someone inventing a tool to help understand very complex risky portfolios (i.e. a firms position in the market).Then a load of people who aren't as bright saying they understand it and then using it as a bible and an excuse when they get shafted by the market... I can imagine the conversations of the CFO saying to the risk guys: But VaR was 10 MM, how did we loose 100 MM??? It is a typical case of people (who should know better) not understanding the difference between probability and certainty. As my Philosophy of Science lecturer said: "Probability is great - I am always right - 80% chance of rain, and it doesn't? Well it was obviously in the other 20%."</rant>

I agree with rmax and disagree with Lewis.IBs should be public companies because it forces more transparency and accountability on the firms than if they were private. Although one could attempt to replicate that transparency and accountability with government oversight, I'd argue that such oversight would be inferior on three dimensions. First, it's the counterparties to the IB that should have oversight on the IB, not the government. Second, government oversight would both stifle innovation and miss novel threats. Investor/analysts embedded in the financial services industry surely have a better understanding of what's happening inside IBs than would a Washington/London/Brussel-based bureaucrat. Third, public investors (with money to lose) have much more incentive to understand the IB than would any bureaucrat (with a life-time job).At some level, ALL companies have aspects of opacity that investors and CEOs don't or can't understand. Obvious examples are big conglomerates (e.g., GE) that have so many components as to be opaque (and any of those components could become the basis for a disaster that damages the core). But I'd argue that even "simple" companies can be opaque to the extent that the essence of their success is some non-obvious strategic or cultural factor. What made companies such as Dell, Nokia, or Southwest Airlines succeed? Although a lot of people have written a lot of words "explaining" the respective successes of these companies, it's obvious from the diverse and changing verbiage that no one really knows. Certainly, no one really knows if these companies will continue to succeed which makes their futures opaque. Yes, exotic derivatives can make a firm opaque but so can some ineffable, potentially ephemeral, esprit de corpLewis does raise a good point about the speed of Bear's death but he draws the wrong conclusion from it.The key is to realize that the revenues, costs, assets, and liabilities of any company are governed by unpredictable processes (I call the processes unpredictable because they aren't 100% deterministic, aren't 100% stochastic, aren't 100% chaotic, aren't 100% complex adaptive but may be some mix of all four.). Moreover, the earnings and equity in the company are governed by differences among revenues versus costs, and assets versus liabilities, respectively. Obviously, if we have a difference between unpredictable processes, then we face some chance that that difference could become less than zero which leads to losses (revenues < costs) and bankruptcies (assets < liabilities).Bear died so quickly due to the potential speed of the zero crossing in the difference between these unpredictable processes. In the months, weeks, and days leading to the bailout, Bear lost revenues and assets much faster than it could shed costs and liabilities. This problem is NOT unique to Bear or to IBs. To the extent that the revenues of any company can go from 100% to 0% of normal in a few days because it's lost the trust of its customers (and those customer have ready alternatives), any company can suffer Bear's fate. I'd wager that many companies face the problem that the time frame on costs & liabilities is longer than the time frame on revenues and assets -- that revenues & assets can disappear much faster than can costs and liabilities.If anything is the root cause of this, then it is the hubris of leverage and a naive trust in liquidity. Leverage reduces the difference between assets and liabilities as a function of the magnitude of assets and liabilities. With leverage, it takes a much smaller burble in those unpredictable processes to create a zero crossing that puts the firm in bankruptcy. Trust in liquidity also plays some role because a company will believe it can survive a draw-down on the revenue/asset side of the business if it believes it can depend on liquidity to help it readjust its balance sheet and cash flows.

Losers.

were these guys MDs or VPs? i don't see any photo of their wives... are they young trophy wives?