Wednesday, September 2, 2009

Computation, finance and engineering

Moral: Wherein we consider the American dream and success which can have many definitions, most of which would include money which then brings in finance. In engineering, it has to do with coming up with new things that will marvel the world.

---

The ACM Communication editor-in-chief (Moshe Y. Vardi) reminds us (The Financial Meltdown and Computing) that the current state of affairs in the economy has been heavily influenced by computation. That is, we could not have the 'markets' as they are now without computer and network assistance. Where would high-frequency trading be without the resources they are exploiting?

Moshe mentions a series of shocks that we've seen the past 25 years, starting with October, 1987. That black monday was caused, some say, by program trading gone awry. Since then, various control mechanisms have been put into place to limit decline. Why isn't there one on the upside? Of late, especially with the higher speeds and derivative trading, do we know the effects? By the way, that is not rhetorically meant, as we'll dive into the guts as we go along here.

Then, Moshe mentions the LTCM mis-deal. This could be funny except we did not seem to learn from the situation. That Nobel winners were involved is something to look at, in depth (gosh, we need a sandbox for the heavy players). Needless to say, 'moral hazard' was bailed out.

Now, of course, we have the current mess. There are many culprits and factors. And, we have to add quants to the list, those who ignore complexity and ought to know better.

We can use the ACM Communication to look at a couple of the issues related to the theme. Computation is not the only problem, as we have seen malfeasance, ignorance and other bad traits come into play. Yet, there is a lot to learn by applying the metaphors involved.

--
The Status of the P versus NP Problem (P not equal to NP)

The discussion about this started as a Computer Science problem but now is on the Mathematics prize list. The gist of the concept is that a solution to a problem is easier to check than to find. That is, you can tell if something will work by trying it. Now, if you don't know the something that is needed, how do you find it? Of course, others may have worked out the problem. That is what true pioneers do.

Yet, what if the something doesn't quite work. You see, any change to the situation related to the problem could invalidate the solution that you have plus put you back into facing the hard issues again. Hence, people do not like change; one motivation for small perturbations is the hope that small changes do not make things worse.

These blogs have used undecidable several times as the means to denote not knowing until after the fact whether something can be done. It's the heart of problems with earned value and of many of the harder problems that are part of our daily existence.

The usage of undecidable can be waved off as academic (see reaction to Vienna School), yet, it is more than that. Also, that the quants run off after solutions that gather money solely while arguing that their measure spaces are well-behave enough to ignore these issues only means that they do not care about the wider effects which have some impact on us all.

Many system problems (including the numerically based) are of the same ilk in terms of difficulty. The mere act of bringing in computational support opens up the related can of worms.

-- AI and hive mentality

Ah, as Jaron says, we see humans "bending over backward, sometimes making ourselves significantly stupider, to make an algorithm seem smart." Just like we have seen in the current mess, he says. I ask, is that which the quant focus takes similarly myopic?

As we've said, leveraging is almost a perpetual motion analog (which is something from nothing), and we know how physicists react to such proposals. Why then is it allowed with money? Like the tranche idiocy?

Same, too, the thing called lean has undecidable states. Management handles these issues by scrapping people, whereas one would hope that we could take a more mature approach (to be defined).

So, how will we ever know if we obtain collective intelligence? In making that determination (underdetermination is the operative concept here), would we have "collectively become idiots?"

Remarks:

03/15/2015 -- Finally, getting around to the pending business.

01/15/2015 -- This week, this post is getting read. Great! Nice little piece of work (kidding, in part) so many years ago. ... At last, a series that will establish the basis and extensions, as required. We are going to go back to some simple and come forward to the modern, complicated economy. Why? My long chain of ancestors (inherited via Prof. Lucio Arteaga) is one motivation.

03/03/2014 -- Acknowledgements, including math pedigree, will be expanded.

12/06/2013 -- If only Ben would put a shot across the bow. He's helped the chimera unfold in unhealthy ways. He could, at least, say a mea culpa.

08/07/2013 -- Over the weekend, Motley Fool had an article that asked how high could the DOW go to which we made comment (post that was precipitated by the article). So, here are a couple of things to discuss, especially since there seems to be some worry of the taper, of late. First. How do we determine the price at which loss is guaranteed? You see, losses like we've see with the two recent swoons (May/June/July, and the one of the past couple of days) are not shared by all sellers. That is, losses that are not paper only, since losing ill-begotten gains is not a real loss. The loss manifests when you sell. So, as a market tumbles down, those who can still make a profit can bail out to a certain point. That is, they can sell before they lose what the put in. Again, ignore ill-begotten gains, please. If we look closely, that price is not as low as we might think. In a sense, equity holdings will lose money. All the talk about the equity markets rising, over time, do not consider the accumulative losses during the period, nor do they look at the side-effects of inflation (and structural changes like we see now where most jobs are essentially glorified indentured slavery). I know, the trick is to use constant dollars; yet, the total picture is not painted. How do we do that? In short, this problem applies to bond markets, too. Second. For anyone to sell there has to be a buyer. But, to buy, there has to be a seller. Prices goes up as buyers try to get in the game. As we saw with FB recently, prices can jump quite a bit (leaping for the stars - usually without a tether). As markets, like  the DOW, go up and up, there has to be some implosion point (as in, the thing collapses due to the unnatural states that accumulate). Let's say that you run out of buyers. Well, that was handled, in the older times, by the specialist who was the buyer of last resort. Guess what? With these modern schemes, we have seen where there was not a buyer (as in, the game runners want it one way: their continual gain - loses go to the public). So, with no buyer of last resort (and, Ben is being just that for bonds), how long can the market just sit there and churn? Now, there are many ways that people might decide to sell. The fall in price, like the going up, will be faster if there are no buyers to catch the thing. How far can we fall? Guess what? Now, they stop the stupid market. Why not let the stupid thing go down to some low number, say what it was in 1990? No, the thing has to ratchet down. Supposedly, that would remove insanity and allow jaw-boning time (by all of the idiotic heads that yap every day) to bolster people's confidence to get back in to the game (which, we know, as ca-pital-sino, is a loser's game). Or some such. ... Now, we could bring in a third (and more) thing related to the churning caused by all of the algos (errant computation) and other crap (greedy worldviews). Stock would increase in value with real accumulation of wealth (whatever that is) over time. However, how can we see such with the dark pools (idiocy) and other murkiness (just the whole thing of perturbing trades that are meant to lure the foolish to their financial death)? Ben, address that, please.

07/30/2013 -- The future: economy and technology.

06/11/2013 -- CDOs and tranching, once again.

05/14/2013 -- Still the most-read article (post). Ought to bring this up to date. Those who play the games are into computation, heavily. Why not? It seems easy; too, they're allowed to fill their pockets. Congress doesn't understand, evidently. Do auditors? Of course, I've not made it explicitly clear what is the problem (will be working on that). So, we're May, 2013. Ben has a 15K DOW as an accomplishment (congratulations are in order). And, he has not unwound, though we've asked for this for years now. Guess what? We're back to the heavy leveraging idiocy.

10/13/2012 -- Homo heuristicus.

03/23/2012 -- Renewal of the idea (and related energies) via Cooper and CiE.

03/15/2012 -- Okay, might have used incomputability (see post on Alan M. Turing) but stand by the context, the issues, and the need for resolutions. Wake up, quants (you, too, Ben).

09/20/2011 -- This will be used in our constructive effort.

04/03/2011 -- Tis tranche and trash.

02/26/2011 -- Technology will be important. But, for the benefit of people.

01/19/2011 -- For the most, things are dire, not by necessity.


12/15/2009 -- Requiem for the dollar (WSJ) and responses.

11/30/2009 -- From 'Our basis' can grow a whole bunch.

09/09/09 -- We'll need to look at UUUN, as a framework.

09/03/2009 -- Jaron asks: How far back in history toward the stone age will people have to devolve in order to find a way to make a living when fabricating robots are that good? Will people be forced by the market-place to work the fields, as academics did under various Maoist-type regimes? Well, hard, manual work can be good for the soul. Actually, some of the imbalances that we see from abstractionism and gamism would diminish if we had requirements, such as mandatory civil service, of the hands-on kind. that expanded experience.

By the way, it turns out that determining value, either fair or earned, is NP which leads to essentially undecidable states.

Modified: 03/15/2015

No comments:

Post a Comment