Forums  > General  > QuantFinance is dead  
     
Page 1 of 2Goto to page: [1], 2 Next
Display using:  

nikol


Total Posts: 850
Joined: Jun 2005
 
Posted: 2019-09-17 15:09
Why quant jobs in finance no longer need finance knowledge

GIGO era is coming.

Reminds "The Decline of the West" by O.Spengler

Jurassic


Total Posts: 276
Joined: Mar 2018
 
Posted: 2019-09-17 15:31
What a waste of my time reading that article was

nikol


Total Posts: 850
Joined: Jun 2005
 
Posted: 2019-09-17 21:47
Spengler's name is a nickname for predicting things which do not happen.

Lopez made a bold statement. Let's see.

pj


Total Posts: 3469
Joined: Jun 2004
 
Posted: 2019-09-18 06:08
What nikol said.

I thought that quants usually dabbled in structured products
not in technical analysis. Cool

@Jurassic,
come on, the article wasn’t that difficult to read.

The older I grow, the more I distrust the familiar doctrine that age brings wisdom Henry L. Mencken

nikol


Total Posts: 850
Joined: Jun 2005
 
Posted: 2019-09-18 07:59
I have to admit I added a bit of "pepper".

Lopez says mathematically loaded analysts will be in demand, but they don't need the knowledge about finance, hence QuantFinance is dead slogan.

This statement brings new opportunities to the new generation of financial quants.

ronin


Total Posts: 512
Joined: May 2006
 
Posted: 2019-09-18 13:45
And suddenly, AQR's performance makes sense...

"There is a SIX am?" -- Arthur

chiral3
Founding Member

Total Posts: 5101
Joined: Mar 2004
 
Posted: 2019-09-18 14:06
I haven’t read Lopez. Maybe I should, but I haven’t. Funds tend to house gadflies and interesting people. I don’t want to name names, because some of these people are our friends, but sometimes funds keep people around that have interesting ideas and write books because, at one end, their ideas are novel and challenge the norm, and at the other they serve as moonshots.

It’s naive to think that things will stay the same and foolish not to adapt. The criticism that math-types don’t understand business is as relevant now as it was in the 1980s. Maybe the modality is different: now it’s ramming data through gigo algos, then it was pricing rate derivatives. In the end all new employees aren’t going to understand the business, regulatory environment, how people make money. If someone spends two years as a low level quant, some years in academia, then writes a book like “Weapons of Math Destruction”, just to pick one, it’s virtually guaranteed that the person doesn’t understand what they are talking about.

Nonius is Satoshi Nakamoto. 物の哀れ

chiral3
Founding Member

Total Posts: 5101
Joined: Mar 2004
 
Posted: 2019-09-18 15:20
I had to cut that post short because of a meeting...

It’s not that “quantfinance is dead” as much as risk taking is dead and the regulatory environment is devastating. I would also argue that traditional risk-shifting and financing strategies have been disintermediated by VC and private equity, although I am seeing a resurgence of structured finance a la 2000’s. This is primarily out of necessity in the hunt for yield via liquidity and not credit risk. However, the days of quantfin re exotic derivs that ran from about 1980 to the crisis is probably dead forever, and that’s a good thing. The bigger issue is that the opportunity that it morphed into is being strangled by global QE and austerity measures.

I am in general bearish on this front and think that the innovation that is suppposed to thwart a Malthusian collapse is being subverted by our inability to take risk, and that most innovation is being realized in mindless commercialization of, basically, entertainment. The reality is we bought a $100 bond for our kids that will pay $95. The money has to come from somewhere and it’s not from paying the Kardashian family or their close relative, Ivanka Trump.

Nonius is Satoshi Nakamoto. 物の哀れ

EspressoLover


Total Posts: 384
Joined: Jan 2015
 
Posted: 2019-09-18 17:23
> I would also argue that traditional risk-shifting and financing strategies have been disintermediated by VC and private equity,

This is a very under-appreciated point. It's not obvious, but I think in the past ten years private equity has really eaten our lunch.

Much of the raison d'être of high finance comes down to the stylized fact that investors, for both institutional and psychological reasons, tend to be much more risk-averse than is rational. To the extent that you can "fool" them into ignoring their fear of risk, you can create real-value. That's the main insight of the classic noise trader paper.

Quant alchemy's typically accomplished this by transmuting risk into some form that's obfuscated from the constraints. The pension mandate says you have to be 50% in investment grade bonds? Okay we'll just pool and tranche this bucket of high-yield junk until it technically meets the rating agency requirements. You feel uncomfortable buying stocks on margin? No problem, we'll just build this weird option that's effectively 200% leveraged with this esoteric barrier clique that's essentially meaningless but lets you sleep better at night.

Private equity's found a simpler solution to this problem: just eliminate liquidity and transparency altogether. For the irrationally risk-averse, those features are actually bugs. In the long-run private equity returns are pretty identical to small-cap stocks. But on a month-to-month basis the portfolio valuation can be massaged to smooth out the volatility. Look at how much VCs bend over backwards to make sure their companies never have a down round.

If you look at an entity like Softbank it looks pretty similar to a Bush-era SPV. A sprawling nexus of contingent claims, obfuscated exposures, and opaque ownership backed by the aura of geniuses behind the curtain. Basically a way to transmute underlying investments in very risky ventures into a respectable-looking package that a pension manager would be comfortable with. When someone like PIF makes a $45 billion allocation to Vision Fund, that's pretty likely taking food out of quant mouths.

Good questions outrank easy answers. -Paul Samuelson

chiral3
Founding Member

Total Posts: 5101
Joined: Mar 2004
 
Posted: 2019-09-18 18:30
“Private equity's found a simpler solution to this problem: just eliminate liquidity and transparency altogether”

Right, and part of the way that they were able to do that was by leveraging to the hilt while regulators were (are) looking the other way while feasting on the souls of the banks. I’ve been in meetings with the top levels of the big (US) PE where they propose things that sound illegal; until I remind myself that they have a really great relationship with the regulator that looks the other way. If you go to an investor presentation for any of the big, say, three PEs, it’s virtually impossible to wrap your mind around the complexity of their company, and they aren’t using quants to figure out what happens when correlations go to 1 (how do you cross a Ukrainian wind farm with CLOS, a movie theatre, and a tax arbitrage in xccy swaps across ten countries?). They will either be the smartest of the dumbest people in the room at the end, although with this protracted credit cycle who knows when the credit impairments will materialize.

Good example was this morning: PE buys entire bond sale for it’s own LBO!!!! OPM is an art that sounds illegal. How do those fees add up? Problem isn’t with the PE. If I had no moral compass and I was let lose in a whorehouse with someone else’s money I’d fuck a bunch too. Problem is there are no parents around.

Nonius is Satoshi Nakamoto. 物の哀れ

nikol


Total Posts: 850
Joined: Jun 2005
 
Posted: 2019-09-18 19:46
@ronin: And suddenly, AQR's performance makes sense...

Applause


@chiral3: The money has to come from somewhere and it’s not from paying the Kardashian family or their close relative, Ivanka Trump.

The very heart of that.
What we observe right now is that money are inhaled and exhaled to/from the system without asking anybody around ("Those who you voted for has given us the mandate, not you. It is in the design of the system us to be independent from you. Prepare for the next move!")

@EspressoLover

> lets you sleep better at night

Nice comparison of derivatives to sleeping pills. I am afraid you are right.

It crosses with my view that trading book is just a forward with complex underlying which is an obfuscative wrap over original cash flows and intentions of the client processed through back-to-back transactions.

I get a sense over years, that all that FinQuant profession is needed only to make that wrap.


Coming back to Lopez, I would lower my initial sarcastic bar, because I sense his feeling of frustration and being mad at "Black boxes" who took "his money". He is in the state of "dark irony and humor".

kloc


Total Posts: 16
Joined: May 2017
 
Posted: 2019-09-18 20:52
@EspressoLover/@chiral3: these 3 posts about PE are probably the best summary of the curent state of affairs that I have seen so far. Kudos...

Maggette


Total Posts: 1167
Joined: Jun 2007
 
Posted: 2019-09-19 08:11
My pen and paper math skills suck, but I read parts of Lopez book and was not convinced. For example in Ch. 5 IMHO he seems to mix up the time series concepts of memory and stationarity.... Which are kind of important to distinguish if you want to forecast stuff. Memory IMHO is something like mutual conditional information (entropy) , the stationarity is a property of a stochastic process.
That might sound like bickering over terminology, but to me it has practical implications.
There are some interesting things in his book... But I somehow have doubts he really developed strategies

Ich kam hierher und sah dich und deine Leute lächeln, und sagte mir: Maggette, scheiss auf den small talk, lass lieber deine Fäuste sprechen...

rod


Total Posts: 392
Joined: Nov 2006
 
Posted: 2019-09-19 08:13
Lopez de Prado lists his perfect MENSA score on his LinkedIn profile...


Maggette


Total Posts: 1167
Joined: Jun 2007
 
Posted: 2019-09-19 08:25
He us smarter than me... But who isn't

Ich kam hierher und sah dich und deine Leute lächeln, und sagte mir: Maggette, scheiss auf den small talk, lass lieber deine Fäuste sprechen...

chiral3
Founding Member

Total Posts: 5101
Joined: Mar 2004
 
Posted: 2019-09-19 11:17
Don’t be too hard on yourself Maggette. What you may not have in Mensa credentials you probably make up for with employability, balance sheet, and the ability to actually build and deploy things that don’t belong in a toy chest.

Nonius is Satoshi Nakamoto. 物の哀れ

deeds


Total Posts: 459
Joined: Dec 2008
 
Posted: 2019-09-19 11:30

@chiral3, what does OPM stand for in your 2019-09-18 18:30 post?

deeds


Total Posts: 459
Joined: Dec 2008
 
Posted: 2019-09-19 11:37

@Maggette

Have not read Lopez, but saw him speak so i may not have a great picture. Mandelbrotian, Talebbic. One or two important observations in a souffle of exaggerated application with some convenient ambiguities in order to ratchet up significance. One of the biggest services is to shine a diffuse light into an area perhaps worth exploring.

Agree on stationarity / memory (in addition with entropy approach). seems part of a larger issue about model choice and design

chiral3
Founding Member

Total Posts: 5101
Joined: Mar 2004
 
Posted: 2019-09-19 11:43
Other people’s money. It’s a strange thing: people will try to make other people’s money work in ways they will not attempt with their own capital.

I was writing quickly before. I didn’t mean to suggest that more regulation is better. The government can’t attract and retain talent. It’s like EPA being able to attract the best chemists compared to industry - not possible. However, there’s been no competition, which is what was needed. The best case was bringing a knife to a gunfight. I’ve been in situations that were identical but went to a PE for what what reason? And if they make a mistake there’s a free sovereign (Saudi?) put that will pay some reasonable amount of downside. It’s just an artificial regulatory construct that limits competition.

Nonius is Satoshi Nakamoto. 物の哀れ

deeds


Total Posts: 459
Joined: Dec 2008
 
Posted: 2019-09-19 12:14
@c3 agree

- i remember reading the FT articles when in the UK in 2005-2007...every 6 months or so the structured talent drain was explicitly described as quants moved from rating agencies and regulatory to funds...starting salaries for PhD graduates of GBP $500 K (at 2 USD for each GBP)...

- in PE case, the mechanism of regulation is audit, SEC, PCAOB...SEC regs seem currently aimed at mutual funds (though they bite all RIAs), and valuation is around what to do when markets become illiquid in order to manage impact of dilution upon redemption. PE investments are designed to be illiquid, but there is no specific ASR type communication around them and PE firms do not take these regs seriously.

- agree about risk appetite or even diligence w OPM. i work in "helping" PE funds estimate fair value for financial reporting. if at all possible, fund managers want to hold at cost until time to raise further funds or exit, with no sensitivity to wider market movements or even their own exit market. frequently they do not have available the terms of the equity securities they own (usually preferred stock with interesting features) nor the purchase agreement / articles for the investee company. They think of their investments as though fully diluted common claims improperly using transactions in preferred as reference points (this has been written about in the last two years in the popular press after a study suggesting most unicorns are overvalued by a factor of two...wework?)

- agree about sovereign or even greater fool put...(in the dark, it seems the population of greater fools is boundless)




nikol


Total Posts: 850
Joined: Jun 2005
 
Posted: 2019-09-19 12:55
The "illiquid design with limited competition" has generated very awkward "innovation" - ICO's

In attempt to escape SEC eyes crypto-entrepreneurs attracted pools of money with 90% probability (at this level it is a guarantee!) of losses.

What else we have?

Sisqo


Total Posts: 10
Joined: Jul 2010
 
Posted: 2019-09-19 12:55
I asked my Lopez De Prado a question at a conference last fall which was meant as a layup and an invitation to elaborate. He did not understand the question and instead of asking me to clarify he jumped in and, smiling at me, proceeded to give an unnecessarily long nonsense answer. So take from that what you will.

nikol


Total Posts: 850
Joined: Jun 2005
 
Posted: 2019-09-19 13:00
@Sisgo

I had the same personal impression. But still his statement (on top of the thread) is about to be valid.

If true, this brings forward the comparison with monkey able to print "War and Peace".

Sisqo


Total Posts: 10
Joined: Jul 2010
 
Posted: 2019-09-19 13:26
I don't know enough to say whether crowdsourcing is the future, or just the very recent past.

Crowdsourcing feature engineering seems doomed in the medium term, unless mechanical Turk is cheaper than code that does feature discovery in an automated way.

Crowdsourcing forecasts, on the surface seems like more value-added, but stacking multiple models is completely automatable, and exploring the space for models and parameters is doable now at great expense, but increasingly something you can do with meta optimization and attendant code.

The point about offloading the computational costs onto your quant's computers is very valid, but there is a chicken and egg problem - if the crowdsourced quants are performing tasks best done by algorithms, maybe the raw computational cost does not dominate the total value of the exercise.

This is jumping ahead slightly, but I think this concept of cleanly delineated feature extraction, forecast, and portfolio building seems slightly outmoded to me. It is convenient to ignore reflexivity, and state "as long as our trades are less than x percent of volume" then reflexivity can be neglected, and I conjecture that many market participants makes this conceptual tradeoff.

But maybe a more advanced methodology is one that explicitly models market participants as dynamically interacting according to their own time-varying aggregate function over external data and variables internal to their fund? It seems to fit reality more cleanly and while I am no great modeler or anything else, whenever I consider going in this direction, I decide it's too hard, but maybe more openness in the statistical learning community and more collective experience will yield some very large scale reinforcement learning approaches that more directly map to how most order book type markets actually operate?

I could probably have said that in one sentence.

TonyC
Nuclear Energy Trader

Total Posts: 1317
Joined: May 2004
 
Posted: 2019-09-19 14:00

> Good example was this morning: PE buys entire bond
> sale for it’s own LBO!!!! OPM is an art that sounds illegal.
> How do those fees add up? Problem isn’t with the PE.

I must admit I haven't been following too closely, so which deal was that?

did they actually charge some equity client a fee, and then turn around and charge
their mezzanine financing client a fee for the loan that made the equity deal actually possible?

flaneur/boulevardier/remittance man/energy trader
Previous Thread :: Next Thread 
Page 1 of 2Goto to page: [1], 2 Next