Financial Risk Manager

‘Bogus’ Risk Models

2008-11-10T15:22:52-08:00

Interesting article on risk management models:

http://www.bloomberg.com/apps/news?pid=newsarchive&sid=aW2ByfpGZflA

Business schools should teach empirical analysis and drop risk-management models that failed to foresee the worst market declines since the Great Depression, according to “Black Swan” author Nassim Taleb.

Universities teach mathematical theories such as the Black- Scholes model of pricing options that don’t express risk properly, Taleb said in a Bloomberg Radio interview. Options are derivatives that give the right but not the obligation to buy an underlying security at a set price and date.

“Recent events have proved that all risk management was wrong,” Taleb, a former options trader, said. “We need to do something drastic immediately to stop quantitative risk managers from inflicting more damage.”

Academic teaching resists change because tenured professors aren’t forced to adapt or subject to the same standards that the market imposes on Wall Street, Taleb said.

“People are learning the wrong thing,” he said. Rare and unforeseen events are known as “black swans,” after Taleb’s book, “The Black Swan: The Impact of the Highly Improbable.” It was published in May 2007, about three months before the credit crunch rocked global markets and led banks to announce almost $700 billion of asset writedowns and credit losses.

‘Bogus Matters’

“You have a business school establishment that’s completely disconnected from what’s going on” because it depends too greatly on equations, Taleb said. “We should learn to abandon these bogus matters, and instead of using a computer we should look at the world with the naked eye.”

Taleb, who once taught a graduate course on how models can fail in quantitative finance at New York University, had the biggest payday of his life during the stock market crash of Oct. 19, 1987, when his options exploded in value.

This year, investors advised by Taleb gained 50 percent or more this year as of Oct. 14 after his strategies for navigating big swings in share prices paid off amid the worst stock market in seven decades.

Universa Investments LP, the Santa Monica, California-based firm where Taleb is an adviser, has about $1 billion in accounts managed to hedge clients against big moves in financial markets. Returns for the year through Oct. 10 ranged as high as 110 percent, according to investor documents. The Standard & Poor’s 500 Index lost 39 percent in the same period.

Record Volatility

U.S. stock swings as measured by the Chicago Board Options Exchange Volatility Index, or VIX, rose to the highest in the 18- year-history of the gauge last month, reaching 80.06 on Oct. 27. The index measures the cost of using options as insurance against declines in the Standard & Poor’s 500 Index, which tumbled 17 percent in October for its worst monthly drop since 1987.

Even the hedge-fund firm co-founded by Myron Scholes, who won the 1997 Nobel Prize in economics for his development of the options pricing model with Robert Merton, wasn’t immune to this year’s market turbulence.

Scholes’ Platinum Grove Asset Management LP temporarily stopped investor withdrawals from its biggest fund after it lost 29 percent in the first half of October. The decline left Platinum Grove Contingent Master fund with a 38 percent loss this year through Oct. 15, according to investors

GARCH(1,1) Estimating Gamma, Alpha, & Beta

2008-11-13T15:53:01-08:00

David,

When using the GARCH(1,1) model, what is the best method to determine the weights for gamma, alpha, and beta? I watched your video over EWMA, and noticed that you recommended setting lambda to .94, but was wondering if this is an appropriate weight for beta in the GARCH(1,1) formula? If we do so, then gamma becomes small, and the forecasted rates more persistent. Any resources or papers over this subject would be helpful. Thanks in advance! I really appreciate it.

Closed form

2008-11-12T18:16:00-08:00

Hi David,

What do we mean by closed form and reduced form as mentioned for many methods. Like BScholes is closed form ...

Pls suggest.

Rgrds
OM

explanatory power?

2008-11-12T10:23:27-08:00

hi,

what is explanatory power??

and one simple vocabulary question,

the portfolio is unwound -> what is the meaning of unwound ??

quite stupid question,,

thanks!

suk

EVT

2008-11-11T23:03:41-08:00

Hi David,

Q36 in the practice exam 2008

Which of the following statements about Extreme Value Theory (EVT) and its application to value at risk
are true?

It mentions in answer for this that “For empirical stock market data, standard value at risk estimates at the 95 percent
confidence level tend to be fairly accurate, and generally only becomes inaccurate at the 99.5 percent
confidence level and beyond"… I am not clear how can estimate be inaccurate at the 99.5% level ?

Regrds,
OM

SER (a.k.a. SEE) - How to determine n?

2008-11-11T16:19:54-08:00

In the following formula, how do you determine n in the question below? The question doesn’t mention the sample size. The answers seems to indicate that n = 3.

SEE = (SSE/(n-2) )1/2

10. Paul Graham, FRM® is analyzing the sales growth of a baby product launched three years ago by a regional company. He assesses that three factors contribute heavily towards the growth and comes up with the following results:

Y=b+1.5X1+1.2X2 +3X3

Sum of Squared Regression [SSR] = 869.76
Sum of Squared Errors [SEE] = 22.12

Determine what proportion of sales growth is explained by the regression results.

a. 0.36
b. 0.98
c. 0.64
d. 0.55

Answer: C

Coefficient of Determination i.e. R2 explains proportion of variation explained by the regression.

R2 = SSR/SST
SEE = (SSE/(n-2) )1/2
SST = SSR + SSE.

Therefore, SSE = 489.29, SST = 1359.05, R2 = 0.64

Platykurtic and Leptokurtic

2008-10-29T23:50:43-08:00

Hi David,

Can you provide provide a figure showing when and how they cross each other wrt to Normal distribution.
T distribution is a leptokurtic....does it too have thinner tail than normal ? In your spreadsheet I can see it to cross normal just once making it to have fatter tail than normal.

Thanks
Sipani

Square Root of Time Rule Problem

2008-11-09T09:31:02-08:00

I’m having problems applying the square root of time rule for this problem. I converted all the portfolios vars to yearly vars, so that I can compare “apples to apples”:

Var2 = Var1 * alpha * squareroot(Time2 / Time1)

So for portfolio #1: 10 * 2.33 * squareroot (252/5) = 165.41

I did this for all the portfolios, but I could not get the order in answer A below. Am I applying the square root of time rule correctly here?

104) Rank the following portfolios from least risky to most risky. Assume 252 trading days a year and there are 5 trading days per week:

Portfolio / Var / Holding Period Days / Confidence Interval
1 / 10 / 5 / 99
2 / 10 / 5 / 95
3 / 10 / 10 / 99
4 / 10 / 10 / 95
5 / 10 / 15 / 99
6 / 10 / 15 / 5

a) 5,3,6,1,4,2
b) 3,4,1,2,5,6
c) 5,6,1,2,3,6
d) 2,1,5,6,4,3

ANSWER: A

Multiple regression (Multicollinearity)

2008-10-29T21:49:57-08:00

Hi David,

No linear relation between variable but can they be non linearly related say ....X1 = b1*X2^2

Thanks
Sipani

Random Walk/Mean Reverting (Quant A)

2008-05-28T21:10:27-08:00

Hi David,
Thanks for your reply on my previous question.

Another request - Will you be able to upload a spread sheet example to illustrate Randon walk/Mean reversion.?

Regards,
Sathya