article on algo trading

here's an interesting article on algo trading written by professional mathematicians and the nerd leinweber. it is show how computer beat the human with algorithm by use various price analomies

algo have edge because of firepower, and human overtrade and give money to broker

it is also mention visualizations, like nyse markettrac, which i think, can make look in the pair trading, when one stock go away a lot, trade it against the dia etf maybe

 

Did the algo beat the benchmark? That's the "important" thing. If not, just get an index fund.

 

An algo is as smart as the person that wrote it. Yes, they are required to do arbitrage or pair trading, but this doesn't ensure profitability.

Look up how quant hedge funds have performed in 2009 thus far. Renaissance is the big one and their main fund has been devastated.

 

chess programs on a pc beats workld champion. Unthinkable 20 yrs ago. Do not underestimate program trading. Computers trace everyone's orders, mark winning traders, copy their trades and make them into losers because of the size, trace market emotions via orders, chart requests, blogs, ....

One must be aware of all that to beat them and make profits. By having many systems, different timeframes, changing approach, you minimise chance of beeing finished off. And of course playing againt majority.

Yes, watch for these rare occasions when they are on backfoot ! And press HARD !

 

IMHO, an algo can trade better than the person who wrote it, simply because it will be disciplined and will always follow the rules.

 

Algo reminds of people trying to turn lead into gold, alchemists.

Data is lead, algo is a magician wannabe. Gold simply exists. Lead is fatal.

 

Love this chart from the article:


Here is a snapshot of a dynamic view of a market with limit order book transparency
from Oculus Information’s Visible Marketplace (c. 2004 Oculus Information). Layers on
the book show grouped sizes at each price. As with all of these graphics, the dynamics
and drill-down-to-details ability are content-rich aspects that don’t translate well to the
printed page.
(Oculus is glad to have us use this. Contact is Bill Wright,
bill.wright@oculusinfo.com)

 

Orignal view of the market book..

 

we don't know how simmons is trying to play this game because he is very secretive . .. i've heard him say in his testimony to congress on hedge funds that they're basically trying to be market neutral . . .

you're right, maybe they're not any better in the long run, than anyone else, who is not automated . . .

BUT one thing to take away from the article, i think, is that, like in war if there's information asymmetry - you know more about the enemy than he does about you - you have an advantages.

that's why i'm disturbed by these automated algo maneuvers to gain access to the order book, to get a snapshot of where things stand, intraday

i mean retail day traders don't have than kind of information. so they know less compared to algos. i don't understand how this order book stuff works, but it must be pretty important, if people get hysterical trying to get access to that information.

i've heard somewhere that about 30% of daily trading volume in the es is automated blackbox. so it's not that you're playing only against the machine. it's just that machines have access to order book information, and their time horizon is milliseconds.

another thing to take away from the article, i think, is visualizations. they are endorsed by a very sophisticated mathematicians, who knows what he's doing. i personally always thought they were a toy, but now i'm thinking they give you some edge, in how to interpret price data. as i said earlier, i think it has some application in pairs trading (finviz maps, markettrac, both FREE), probably.

because you can literally visually see by eyeballing market if stock go away very far, and rest of market not go away, and then have something to think about for a potential pairs trade. of course it should be used in conjunction with various statistical.

but if, on a particular day, a dow jones component stock gains 10%, and the dow gains 1%, this may be a good initial filter to look at the trade: short that stock, long DIA (the dow etf).

the dow has a relatively low volatility (~25%), compared to other things, including its component stocks. so if you go short a high volatility stock, that has just made a 1.5-2.0 daily standard deviation spike, your chances should be, probably, better than average. provided you have looked at the correlations, cointegration stuff, and ratio mean etc.

so i am think this and it is good articles

 

Ms Varima-Garch,

I will admit that the visualization is very impressive.

But I wouldn't be too fearful about the order book. The main reason is because it doesn't say who is buying or selling. It's not only day traders participating. What about the investors? Pension funds and mutual funds or the big players. The heavy volume could just be longer-term investors buying/selling for their own reasons... they don't care about support/resistance levels and momentum. Now with the growth of ECNs all the bigger players are better able to disguise their footprints... it's a fancy game of poker. Block trades have always been misleading too.

It is a tiny edge in my opinion... one that gets neutralized after the cost of doing business.