JPM-Hit by the limits of statistics?

thetrader's picture




We are amazed by the many articles discussing (and giving advice)  the JPM “perfectly hedged” trading CIO book, by mainly journalists that don’t understand risk, trading nor (the) greeks. JPM, claiming to have one of the world’s most sophisticated risk management, has apparently lost at least 2 billion USD. What actually happened will probably never hit the media. While journalists debate whether or not to drop VaR as a risk measure, we think it is appropriate to review an old article from Nassim Taleb, one of few clever minds when it comes to understanding “real” risks. From Taleb’s Fourth Quadrant, via Edge.

Statistical and applied probabilistic knowledge is the core of knowledge; statistics is what tells you if something is true, false, or merely anecdotal; it is the “logic of science”; it is the instrument of risk-taking; it is the applied tools of epistemology; you can’t be a modern intellectual and not think probabilistically—but… let’s not be suckers. The problem is much more complicated than it seems to the casual, mechanistic user who picked it up in graduate school. Statistics can fool you. In fact it is fooling your government right now. It can even bankrupt the system (let’s face it: use of probabilistic methods for the estimation of risks did just blow up the banking system).

The current subprime crisis has been doing wonders for the reception of any ideas about probability-driven claims in science, particularly in social science, economics, and “econometrics” (quantitative economics).  Clearly, with current International Monetary Fund estimates of the costs of the 2007-2008 subprime crisis,  the banking system seems to have lost more on risk taking (from the failures of quantitative risk management) than every penny banks ever earned taking risks. But it was easy to see from the past that the pilot did not have the qualifications to fly the plane and was using the wrong navigation tools: The same happened in 1983 with money center banks losing cumulatively every penny ever made, and in 1991-1992 when the Savings and Loans industry became history.


It appears that financial institutions earn money on transactions (say fees on your mother-in-law’s checking account) and lose everything taking risks they don’t understand. I want this to stop, and stop now— the current patching by the banking establishment worldwide is akin to using the same doctor to cure the patient when the doctor has a track record of systematically killing them. And this is not limited to banking—I generalize to an entire class of random variables that do not have the structure we thing they have, in which we can be suckers.

And we are beyond suckers: not only, for socio-economic and other nonlinear, complicated variables, we are riding in a bus driven a blindfolded driver, but we refuse to acknowledge it in spite of the evidence, which to me is a pathological problem with academia. After 1998, when a “Nobel-crowned” collection of people (and the crème de la crème of the financial economics establishment) blew up Long Term Capital Management, a hedge fund, because the “scientific” methods they used misestimated the role of the rare event, such methodologies and such claims on understanding risks of rare events should have been discredited. Yet the Fed helped their bailoutand exposure to rare events (and model error) patently increased exponentially (as we can see from banks’ swelling portfolios of derivatives that we do not understand).

Are we using models of uncertainty to produce certainties?

Full article here

A few Taleb videos worth reviewing here

Chart VaR distribution via Wikipedia.


Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.
jmc8888's picture

The funny thing is nothing they tell you is real.  It's just a forward guess. In reality, they are guessing on 0 or 1 events.  Thus when the entire practice is spent inbetween 0 or 1, nothing it points to is really real. 

Most places focus on being able to do the 'math' of it.  I had classes like that.  But I also had a great one taught by one of the best.  It wasn't about the math, it was about the interpretation.  Basically how to interpret it, is, it's full of shit.  It may hint strongly or not strongly, but never forget it only hints.  It never tells you anything real. 

We base just about everything in this world on 'hints' or 'guesses' called facts by dumbasses.  It's everywhere.  Medicine.  Wall Street.  Judicial system.  Science...errr PSUEDO SCIENCE.  Because there is no science that uses statistics that's real science. 

A million other places.  All of it taken as facts, when really it's fiction...that may be right, but not because the math told you.

People that believe in statistics are certifiable morons.  Don't be one.  Realize the sophistry inherent in a flawed process, and realize it's just a hint/guess, and lambaste the idiots that see them as facts.   Pass it along.

Glass-Steagall based on real limits not a system of sophistry based on guesses.

LawsofPhysics's picture

Economics is NOT a science.  It is a CON.  This article is a waste of time.  Atlas is indeed shrugging and will continue to do so until the fraud is prosecuted and the rule of law is re-established.

sof_hannibal's picture

it's a soft science (or taken to an extreme sense, it is pseudo, as in some of it isn't really science-- it's conjecture) -- like psychology, sociology, philosophy, and even psychiatry...

i.e. PhD or a PhD in Econ... = means, Doctor of Philosophy (i.e. Doctor of Soft Science)...thus, someone with a PhD in Econ has a Doctor of Philosophy is pretending to know what they are taking about...

Centurion9.41's picture

Yep, and that's the lie that needs to be brought into the light of day to vaporize.

The fact is the switch to a fiat currency was deliberately made to allow the banksters to slowly take the economic life from the masses in exactly the same way one slowly turns up the heat and boils a frog.  That's another dirty little truth, the masses are not Sheeple but rather little green frogs.



brodix's picture

Face it. When you have hundreds of trillions of "liquidity" chasing a few trillion of viable investment, it's all a bubble and those trying to keep it from popping have to stuff cash in any and all orifices. When one blows up, they can only patch it with more "liquidity" and the problem only gets bigger.

The Alarmist's picture

It's what you get for dealing with a log-stable oriented problem using tools modelled under a log-normal distribution. If VaR were modelled under log-stable, the risks would be significantly larger, so large in fact that a number of players might actually quit the game ... can't have that now, can we? 

ebworthen's picture

The fatal flaw of statistics is that it strives to eliminate outliers.


sof_hannibal's picture

corzined bitches... buy gold; retain your value

The Alarmist's picture

"The fatal flaw of statistics is that it strives to eliminate outliers."

The beauty of it is that one man's outlier is another man's alpha.

disabledvet's picture

no we are not creating "models of certainty to create certainty" we are using "models of risk to create risk." And now we can point to Jamie Dimon...who was using his "PREVENTER OF RISK DEPTARTMENT" PERSONALLY to "CREATE RISK." And of course "there was a putcsh attempt against Jamie for doing this" and "they were fired, Jamie lost a couple of billion bucks and we'll see what happens on Monday." Of course "now we understand why there was a putcsh and why it failed." They were right he was wrong...wrong wins again. The "post failure news conference" on the other hand...

tedstr's picture

I agree.  You don't have to be a banker to get this.  It is the story of every business scandal.  BUt the lack of controls and process at these banks is amazing.  What is this about the 10 trader blowup we've seen in the past 6-7 years.  You could set your watch to it.

narnia's picture

There's so much credit in the market backed by worthless collateral, no one has any idea where bid and asks end up when the fiscal & monetary stimulated music stops.

Gully Foyle's picture

I feel there is a very tangible link between the "vaporized" money at Corzines MF Global and this missing amount, which is just about the same.

In February, news came that large portions of the money had been traced to banks like JPMorgan Chase in the United Kingdom, as well as to other customer's accounts.

The development, announced this week by the trustee tasked with returning money to MF Global customers, suggests that a substantial sum of client funds is still sitting at JMorgan. The statement from the trustee, James W. Giddens, said that he and JPMorgan “are presently engaged in substantive discussions regarding the resolution of claims.”

The final $680 million or so was transferred to other financial institutions with which MF Global did business, including a substantial portion that went to JPMorgan.

GMadScientist's picture

Allow banks to model their risk, and you'll find out that they can game modeled risk (as usual, to their detriment, and everyone else's as well).

automatons with business suits clinging black boxes,
sequestering the blueprints of daily life
contented, free of care, they rejoice in morning ritual
as they file like drone ant colonies to their office in the sky

there is an inner logic,
and we're taught to stay far from it
it is simple and elegant,
but it's cruel and antithetic
and there's no effort to reveal it

GregGH's picture

Thanks Trader -- enjoyed the Taleb video links you psted at bottom of the article ... fragility is the new buzzword ... sure explains a couple of computer projects I have had to endure .... economies of scale ...ha ..  doh ! 

Umh's picture

The models work fine until something not modeled occurs.

Can't find someone to take the other side. The debt can't be repaid due to bankruptcy, war or government intervention. The model will work out, but you need the money sooner than that.

Somehow a 95% confidence level seems like a better bet than it is to most people.

GeoffreyT's picture

Oh for fuck's sake... people with the barest scintilla of a grasp of statistics sieze on every fuckup by the BSD's as some vindication of a theory of tails so fat that they would have Queen writing a new song - but it's hard to fit "leptokurtic girl" into a scansion.


What happened here is NOT a "failure of VaR" (or, to update bullshit by about a decade, ETL/PC). I am prepared to bet half of my blood that if you looked at the actual portfolio that caused the implosion, it will be so far off its mandate that you might as wel lcall it a bet book.

This is the thing, right... you can have all the 'process' and 'risk management' in the world, written on paper and handed out to potential suckers... but if your OPERATIONAL side doesn't IMPLEMENT the fucking risk-management approach, you're likely to find your sphincter expanding rapidly and painfully at some future date.


I saw this as a mutual fund analyst in the early naughties: I was assessing a manager whose 'Dividend Imputation Fund' contained a significant-shareholder stake (i.e., more than 5% of the entire float) in a hot new-telecomms company that had no prospect of ever turning a profit, let alone paying a fully-franked dividend (which is the point of an imputation fund). BUT... the "bumph" distributed to researchers made a great deal of how they combed the investible universe for high-yield stocks paying fully-franked dividends.

Prior to this point, I had actually done a decent job at trying a valuation of the company, and under very vanilla assumptions I came up with a small negative number, plus or minus 3/5ths of fuck-all. In otehr words, it was objectively identifiable as a "hollow log" as we say in Straya.

I "aksed" the manager why he had 4% of his imputation fund in such a single pointed high-risk bet - onew that appeared to have zero underlying value. His answer: "Murdoch's in it; Packer's in it. That's all I need to know". (Packer is the other half of the Oz billionaire-press-mogul world).

The company was One.Tel, and it went toes-up four weeks later... which was three weks after I put a big fat "SELL" on Australia's most popular Imputation Fund (which fell 8% in the week thereafter), and two weeks after the newspapers were full of news that the fund manager in question had 120million One.Tel shares and was DESPERATE to offload them.


So fucking get over yourselves... it's not a failure of the analytical method, or some nuance in the statistical properties of financial markets (not in this case, anyhow). It's just that some confidence trickster convinced his superiors to fuck off and leave him alone, while he quietly went about losing hundreds of millions of other people's dollars and hiding the fact (probably with complicitly from half the risk-management team).


Here endeth the lesson.

Centurion9.41's picture

"it's not a failure of the analytical method, or some nuance in the statistical properties of financial markets (not in this case, anyhow). It's just that some confidence trickster convinced his superiors to fuck off and leave him alone, while he quietly went about losing hundreds of millions of other people's dollars and hiding the fact (probably with complicity from half the risk-management team)"

Geoffrey, F U and your lesson moron.  Of course methods and math don't matter when you have F-wads lying.  The point is that the industry is filled with morons, of which you were obviously by your own admission one, who may have a soul but don't have the brains or education to realize they don't know jack-sh1t.

The ONLY way to PROVE the game is rigged by BS math/science is to show factually that the math, which started off as legitimate, has gotten so far off its original limits and assumptions to be garbage.

So take your "don't blame the math" BS and shove it up your permanently expanded sphincter.  And if you know so much, then join the team and start pointing out the BS rather than looking to cover the @sses of your old sphincter brothers by blaming a couple of their leaders.

When a soldier is given an order to kill innocent women and children, he is not let off the moral hook because some officer told them to do so.  In fact, the officer who isn't there has more moral cover than the guy sitting there pulling the trigger.  And if it comes to killing innocents or being shot, guess what sport?  Man up, say no and get shot.  At least then you can meet your maker with honor and integrity. 

So man up.



disabledvet's picture

You need to re-read your Scottish Renaissance. Adam Smith was writing a book on PERCEPTION of reality when he died. Too bad he didn't finish it. "Could have changed history" as they say. In other words "how do i know what i'm looking at is what i'm looking at?" The rumor is that the money lost (so far) on this trade is almost exactly equal to the amount of money Jamie Dimon had as cash on hand in...ahem..."his bank." Sounds like his risk was well known to him then! The question of course is "what did HE see in the trade" is it not since "the risk is not the risk of the formula but the risk in you and me and me." PSYCHOLOGY folks...the whole "WE ARE NOT ORDINARY" thing. And i AM NOT saying there's a problem with the bank...not that i would be belly up to the bar on Monday either. HISTORY teaches us "mistakes are ALWAYS repeated." So we excel, we stand out...but once we really believe "we're different"...well, so long as it's a "we" i guess...

GeneMarchbanks's picture

Well said, the met-empirical doesn't exist as a concept. Only that which has happened before will happen again on top of the entire cosmos being 100% model-able. Thanks for the knowledge and the continued advancement of dogma despite the facts.

As long as you continue to not let facts alter your mind clarity, you'll be fine. Or you might be a dangerous idiot.

Just keep repeating 'VaR is science'.


GeoffreyT's picture

That's not what I said (anybody who can read above about a 7th grade level already knows that... but that's less than 25% of the population, so the odds are against it I guess).

What I said was that it is not shortcomings in the risk-management model that cause these blowups - it is the fact that the actions of agents in these incidents follows no risk model at all.

It does NOT follow from that, that you can sensibly infer that I claim that the risk models are represenations of reality: I simply claim that they are not to blame if they are not followed. (They might have done a worse job... unlikely).


VaR is basically a pedagogic tool, and anyone who implements vanilla VaR as if "the map is the terrain" is going to end up being able to re-bore their asshole by shoving a ham up it and pulling out the bone. It's almost as stupid as thinking there's a risk-free asset, or that asset prices are continuous, non-jumping and Brownian, or that Markowitz EV is the shit, or that the historical return covariance matrix is an unbiased estimate of the future return covariance matrix. Nobody in their right mind thinks any of those things, and we diminish the debate by pretending that anybody does.

Likewise, statistical ingoramuses think that frequentists (of which I am NOT one - I'm a dyed-in-the-wool Bayesian) claim that the data-generation process is constant (or even piece-wise stable)... that's the most ignorant view imaginable. Why the fuck would GARCH and its bastard children - M-GARCH, STARCH and so forth - even exist if the DGP was stable in 't'?


Seriously... don't try to hold up the pathetically-low-level statistical pseudo-analytical chicanery of VaR as if it's one of the Holy Relics of the professional analysis crowd. It isn't: it's for CFA 2nd year students, and MBA/2nd-year-undergraduate (but I repeat myself) Finance and Accounting classes.

And don't try to pretend that shit that happened before is not a useful tool in trying to establish Pr(Shit happens Again) - if you do, you just mark yourself out as a bit of an airhead. To pretend that every new second is "tabula rasa" is to go naked into the world second by second - you would start your journey from your bedroom to the dunny as if it could literally be in any direction. But you have a probabilistic model in your mind (with a very tight distribution) that says that you will maximise the likelihood of pissing at the right bowl if you turn right at the bedroom door instead of left.

But using frequentist models where

Pr(Shit Happens Again) = (Number of Time Shit Happened in the Past)/(Number of Observations) + (FudgeFactor)

is useful if you have some idea of the sampling distribution of the mean, and the environment that generated the data and Pr(the future environment will be "similar enough").

Anyhow... this is sort of my 'thing': expectations formation and what-not. We can argue about this shit for MONTHS and I promise you, I will never tire of it.

sof_hannibal's picture

GeofT, I agree with you in theory... but a few things (and I will comment on the statistics as well, as what you discount or overlook is systemic ERROR in my opinion),

ok, so -- first point 1) -- what then or what next? The problem with "anything" is/or comes down to not following the rules-- communism, socialism, capitalism, and God for bid, libertarianism-- all these things in theory work or sound great (Marx's theories read beautifully), but the problem always is that nobody follows the rules (especially considering that although "society" tends to balance the needs of the many, individuals always seem to be striving/ hoarding/ concentrating power for themselves, friends, or family (i.e. survival of the fittest, i.e. a dog eat dog world; so this appears to be a case of society vs. individual survival...thus, society creates rules (and enforcement agencies to "protect" the greater good)-- ok, and then the rules are just regulated/ enforced with greater oversight, but then the elite players just trick fuck the system (lobbyist, monopolies, GM dismantling the LA subway system in 1950s, etc.)... history proves this idea correct time and time again... So, now what? what do we do about it? I don't know and I don't have the answers, but I think calling this stuff out (like JPM being a lying sack of shit)-- this becomes important-- that is, getting the people/ masses/ sheeple aware of the truth or the risks...Now do you have any other suggestions on how to fix the system? Just have people [magically be more honest]?

2) point -- Taleb's theories can be shot down somewhat easily, as he does get to carried away with possibilities/ or the theoretical...but, I like this quote from his: "Now the problem: Parametrizing a power law lends itself to monstrous estimation errors (I said that heavy tails have horrible inverse problems). Small changes in the "alpha" main parameter used by power laws leads to monstrously large effects in the tails. Monstrous..." ~ ... conversely, from Wikipedia, this is an interesting comment/ criticism of his: from~ Aaron Brown, a finance professor at Yeshiva University, said that "the book reads [Black Swan] as if Taleb has never heard of nonparametric methods, data analysis, visualization tools or robust estimation."

BUT, I think that last comment/ critic misses the point, in that statistics can be very "Robust," however, Taleb's point is that they are not applied or rather we are not prepared for extreme outcomes as we do not apply these tools (stats) in a robust manner/fashion (i.e. that is Taleb's "4th quardant," that is where statistics/probability fails; but his ideas are just that (in the 4th quadrant), they purely theoretical; whereas-- what stats does right (the or Taleb's: 1st Three quadrants) are pretty scientific and accurate, reliable, valid, etc.; albeit abstract at times when removed from the "lab" or the model and then applies to the "real world."

In my personal experience, I see in clinical trials (and RCTs for medications) all of the time abusing the "science"-- with things like data mining, data dredging, stepwise (regression) error, thus, resulting in the inability to estimate appropriate degrees of freedom, causing vast increases in error, and then decreases in understanding of what "actually" works, and finally, resulting in an overall decrease in the "power" of said outcomes/models/ statistical tools/ and whatever scientific theories (or pharmaceuticals) are being applied or tested...

And, what is behind this all of this-- it's money (and what do you think JPM's stretched idea of hedging where small gains do not outweigh the extreme risks-- it's about making money)-- for me, if you put a drug to market, you need it to beat placebo or you lose big (so you pretty much have to make it beat placebo for instance)... there are so many tricks to make said "drug" beat placebo-- I can tell you this, for a fact, most SSRI anti-depressants are no better then Placebo, but the stats are juiced, manipulated, changed, altered, put in a black box, and the results come out significant (and we make big money). The science works-- that is true. The science isn't wrong, it is just wholly applied in an inappropriate way (i.e., that is what I surmise as Taleb's main point-- we use this science (stats-- which are robust)/ these tools (probability, clinical significant, etc.); however, we use them in an UN-or-NOT-robustness fashion-- to pretty much get the results we have-- to get paid and tick the system)...

so, you can sit there and debate VaR, but mind you -- 700 trillion in world wide derivative(s), garbage bubble collateral financial devices will burst-- at some point (maybe not today, maybe not next week-- but yes, the system is flawed)--and this bubble bursting will be epic and wreak havoc on the global financial system... I recommend preparing yourself in a "robust" fashion for said potential "black swan event" -- (the world will probably not end), but for a few years I would have enough gold, water, silver, shotgun shells, canned food in you garage/ basement, master bed room, car, etc. to at least barter your way to survival over the next several years, until the system rests, and we just do it all over again...or consider self unrobusted/corzined...vaporized...

totally unrelated, did anyone catch that JPM moved another few billion into their litigation fund/ account at the same time they announced this 2 billion loss?

Precious's picture

Hey everybody.  It's the anonymous fucking genius.  He spends his time trolling sites telling everyone how smart he is.  That's how he became the most intelligent person on the planet.  

Get right in here.  You might never get such a chance to see the world's smartest man in action, queer mask and everything.

GeneMarchbanks's picture

What I said was that it is not shortcomings in the risk-management model that cause these blowups - it is the fact that the actions of agents in these incidents follows no risk model at all.

It does NOT follow from that, that you can sensibly infer that I claim that the risk models are represenations of reality: I simply claim that they are not to blame if they are not followed.

Doesn't that make the entire exercise of having models completely meaningless? If a RM model doesn't have (because it can't) the risk of correct human implementation risk within it, then what good is it?

We can argue about this shit for MONTHS and I promise you, I will never tire of it.

Really?! Your ten paragraphs of zealotry on a very simple matter haven't given you away at all but rest assured I won't be the one to engage you in an intellectual mutual masturbation about basics of risk.



sof_hannibal's picture

classic quote: "dyed-in-the-wool Bayesian" ; likewise, can I be bathed in the rivers of a "Markov chain Monte Carlo" (table)

sof_hannibal's picture

good point, and let's not forget what we are debating about...the debate is really about laboratory science vs. how does stuff actually works or plays out in the real world...

as earlier posts have noted, in JPM's (hedged); not-hedged, prop trading desk CIO theory there is always someone to take your bet...however, in the real world, the casino places limits on it; or reality places limits and there are no counter parties... who is going to take the other side of your bet (well, we all know that the FED has become the counter-party of last resort); thus further distorting risk models.

thus, in this case, "risk" via VaR or otherwise cannot be properly assessed... i.e. GARBAGE in, GARBAGE out (the first rule of stats for dummies)...and the 2nd rule is-- don't forget about-- Thursday night statistics, which is a term for arbitrarily setting up rules, numbers, etc.... why is it a 95 percent confidence interval ? or a 99 percent confidence interval; or why not a 99.5 or 99.9 percent... why, because we all got together on Thursday night and decided it should be 95...

also, as numerous others have noted, it's a matter of transparency vs. regulations... and so as long as human's lie or are untrustworthy, models of risk calculation will never fully be accurate. i.e., ERROR (what ever that may be-- unconcious, sub-conscious, intentional, unintentional, or procedural); and don't forget, people will always try to trick fuck the system for a little more fiat...

Seorse Gorog from that Quantum Entanglement Fund. alright_.-'s picture

So, according to GT, LTCM didn't follow their risk management, 'because Murdoch and Packer were in'. To hell with the Black-Scholes model!




GT must've been wanking when he wrote all that...

Nobody For President's picture

No, that is not what he said - those were two completly different examples with different risk models - his point being that in both cases, the risk model that was suppossed to be governing was not being followed. Though even following Black-Scholes, LTCM may still have blown up with the Russian default/black swan event - but going overboard on Russian bets made it a sure thing.