Wednesday, November 28, 2007

Some unique factors for your next quant strategy

It's been a hard year to reside in quant-land. The ignominy of being the smartest guy in the room and yet underperforming day-traders must gall those financial market participants with more degrees than the Sahara in the summer. Even Macro Man, whose "degree count" is closer to absolute zero than the desert, has seen his relatively simple quant strategies get absolutely butchered this month.

The problem, it appears, is that everyone who is using a model looks at the same things. Goldman Sachs, whose Global Alpha fund is reportedly suffering even more than Macro Man's beta plus portfolio this month, said in August that they'll need to include some more "unique factors" in their model construction. This strikes Macro Man as an eminently sensible suggestion. To help his quant-y friends, and to ensure that Aston Martins stay in the proper hands, he offers herewith some suggestions for unique factors to include in future-generation models. Astute readers should feel free to contribute their own suggestions below.

1) Neptune's moons. Has your 2007 performance left your investors out in the cold? Enlist the aid of the coldest thing in the solar system, Neptune's moon Triton! If the number of Neptune's moons visible from an earthbound telescope is even, go long stocks. If the number is odd, go short stocks. If your telescope is broken, go flat.

2) What book are you reading? If it is non-fiction, that suggests that you are interested in facts and analysis. Construct a bottom-up model that trades primarily from the long side. Are you reading fiction? You are creative and imaginative. Sell a basket of stocks short, contact a friendly journalist, and make up salacious rumours about each. Cover your shorts when you finish the book. If you don't or can't read, trade foreign exchange instead.

3) Flip a coin. An oldie, but a goody.

4) What colour shirt are you wearing today? If it is a primary colour (red, blue, yellow), go short equities. If not a primary colour (everything else), go long.

5) Football. Everyone's kids plays football (i.e. soccer) these days, even in America. If your kids score two or more goals combined this week, go long equities next week. If they score one goal, go flat. If they don't score, go short. If you don't have kids, become a broker- you'll be up all night entertaining clients , but at least you can sleep in at the weekends.

6) Use a dartboard. For those who manage Asian equities, where stocks are typically listed by numbers rather than mnemonics, construct a long/short basket by throwing darts at a dartboard. Not only will this give you an uncorrelated strategy, but the practice you get should enable you take take money off your friends at the pub on Friday night.

7) Use sunspot activity as a variable. Actually, maybe this isn't so unique after all.

8) Use your social network! Find the largest person you know. Ask him what his favourite food is. Go long the commodity used to make that food. Then find the most petite person you know. Go short the commodity used to make their favourite food. Rebalance once a month.

9) Use Mr. Brain! In yesterday's comments section, Macro Man inadvertently noted that he used "mr. brain" to develop the simple VIX model that he had written about. Friend-of-the-blog Charles Butler helpfully found the link above. Macro Man isn't sure what Mr. Brain actually does, but is confident that it can only be of use in constructing financial market trading models.
10) Use realistic transaction cost/correlation assumptions. Yes, it may be more mundane than the previous nine, but on current evidence it might be the most unique of the bunch.


t said...

How about numerological cycles?

The classic Armstong cycle of 8.6 years, based on pi?

Ah, not unique, you say.

Well then, I hereby propose the hitherto unknown and unique t cycle based on the beautiful number e,
of 2.718281828 years, which has proven to accurately predict the rise and fall of the Roman, Grecian, Egyptian, and Hindu ancient civilisations, as well as that of Atlantis. There is a secondary cycle based on the imaginary number i which is the king of them all, being (*ahem*) imaginary. And the cycles are all related, since e^(i pi)=-1. So elegant it must be true.

Still the Feb 27th prediction is a little freaky...

Anonymous said...

Ah kudn't read wurds too good, but pictures tole me Loosers go short and Wynners go long. Did Ah get it?

Quarrel said...

I've started buying the ones that are going to go up, selling the ones that are going to go down. It's working a treat.

I might go write an ebook about it - look for it on ebay soon!


2and20 said...

If you don't or can't read, trade foreign exchange instead.

LOL! i read extensively and like to think I make well though out trades in general, but i NEVER get FX right, now i see what I'm doing wrong!

Charles Butler said...

Mr. Brain, I believe, is the physical manifestation of a black box, perpetual return system as it undergoes an entropy event.

Alexander said...

What about the Google-correlation investment thesis? You pick two words, the first is the commodity you care about, the second is a truly toxic word or person. You type both words into google and record the number of results. Then take the first word again and run it against a truly awesome word or person. Again, record the number of results returned by Google.

Then you take toxic results and divide them by awesome word results to get the Google-Derived-Short-Long Ratio, or the GDSLR. The ratio dictates how short/long you should go.

For even better results, you can use a basket of toxic words and a basket of awesome words, which will help eliminate some of the wobble in the GDSLR.

By the way, I am not a registered investment adviser, so use the GDSLR at your own risk.

Macro Man said...

Bravo! Keep 'em coming...

Anonymous said...

macro, what do you think of bill gross predictions of fed funds cuts and a possible '2 handle for fed funds'?is he out of his mind, it seems to me the fed might do it one more time maybe twice but sooner or later they will tell the market to f themselves, they seem to have started to process of getting the governors mouthpieces out in hawkish tones to prepare for just that

Macro Man said...

I think PIMCO is the ultimate "get what you pay for" shop. If you pay fees and buy their funds, you get a pretty good bond fund (disclaimer: I used to own Gross's fund for my personal account but sold it a few years ago). If you pay nothing and listen to what Gross says on TV, you get atrocious market-timing advice and Fed calls.

It's a fairly poorly-kept secret that Mr. Gross uses his TV appearances to push the market his way...if it doesn't work, he gets out of his position.

Anonymous said...

Oh Mr Maquereau Man, thay tuk all mah munny today, pliz sho moor pikters 2morrow pliz. Ah need hep. Butt bettor pikters so Ah kin win some, okay?

U.Unknown said...

1 1 2 3 5 ...

is all you need to explain everything ;-)

Ok...maybe 42 as well