Day traders versus algo traders

[bigbadwolf]

An article in today's NYT:

The problem is that stocks aren’t bodies and their motion is subject to forces Newton could never have fathomed. Some of those forces are hard for the Today Trader duo to fathom, too. Mr. Gomez says that day trading has become far trickier in recent years because of the rise of robo trading — the use of computers to automatically buy and sell huge numbers of shares in superfast bursts, based on algorithms.

Big, muscular Wall Street veterans like Goldman Sachs have the money, smarts and brute power to dominate this computerized battle, and many day traders may not even be aware how outgunned they now are.

About the most Mr. Bettinger will say about day trading is that it’s a “tough gig.” “You’re competing against mega-institutions that are trading in hundredths of a second.”

It's no contest. It's similar to a human player -- even a grandmaster -- facing the latest generation of computer engines (say Rybka 3). The humans can't contest the brute calculating power the engines have and that power trumps even the advantage in judgement and assessment the humans may have. The edge of calculating power becomes even stronger at quicker time controls. Even Kasparov will be outplayed in a five- or ten-minute game.

 

[Joy Pathak]

In the U.S., high-frequency algorithmic trading firms represent 2.0% of the approximately 20,000 firms operating today, but account for 73.0% of all equity trading volume.

There is all this talk about human intuition...and so on. But at the end of the day, the true brute force of algorithmic trading is unstoppable.

 

[winstontj]

I know some pretty good "gut" traders that do OK. That said, Algo trading will easily earn more returns simply because a "gut" trader can't watch as many stocks.

Totally different trading style though.

 

[AlexandreB]

In the U.S., high-frequency algorithmic trading firms represent 2.0% of the approximately 20,000 firms operating today, but account for 73.0% of all equity trading volume.

Hey Joy Pathak,
I've heard those figures before (and believe them) but I can't find the source. Any clue?
Thanks

 

[Joy Pathak]

Hey Joy Pathak,
I've heard those figures before (and believe them) but I can't find the source. Any clue?
Thanks

I don't remember exactly where I got it from. I just took it off Wikipedia right now. But like you said, I have just seen those figures tossed around quite a bit in articles before. Never saw a clear source, but the numbers made sense so I never bothered investigating further.

 

[bigbadwolf]

Totally different trading style though.

The trading style of the bots remains a mystery (except to their designers). In chess, the engine allies calculating power with a relatively rudimentary evaluation process. The evaluation process throws some variations into the bin (thus pruning the explosive jungle of variations that would stress even an engine) and provides a rudimentary evaluation at the end of any particular calculated line. A master can't hope to match the calculating power but has a more nuanced positional feel. Thus in a particular position, the master may start with, say, five candidate moves, quickly dismiss two, and calculate the three remaining at greater length, generating a tree of variations. At the end of any node, he will use his positional feel to arrive at an assessment. This whole process is conducted solely to decide on the next move. In the case of an engine, it more than compensates for its poor positional feel by lengthier and more accurate calculation. But at least we understand what the chess engine is doing. I have no idea what trading bots are doing, what algo traders work on.

 

[winstontj]

HFT is such a broad term that it is very hard to outlay even the basics but it is true that it takes up that much of the volume. On a manual scale a trader may be able to develop a strategy that works but he/she can only monitor a small number of stocks. On an automated scale the algo will be able to apply the same strategy to the entire universe of the NYSE (or more).


In most cases it is very simple logic/programming combined with statistics versus a quant/math genius. The quant/math genius comes in to optimize strategies to get percentages in the traders favor.