• C++ Programming for Financial Engineering
    Highly recommended by thousands of MFE students. Covers essential C++ topics with applications to financial engineering. Learn more Join!
    Python for Finance with Intro to Data Science
    Gain practical understanding of Python to read, understand, and write professional Python code for your first day on the job. Learn more Join!
    An Intuition-Based Options Primer for FE
    Ideal for entry level positions interviews and graduate studies, specializing in options trading arbitrage and options valuation models. Learn more Join!

future of algorithmic trading

ranald

Joined
4/11/08
Messages
10
Points
11
The more i read about algorithmic trading, particularly HF trading, the more I fear that market prices deviate from what they should be... it all seems so artificial. Does anyone else suspect some serious regulation on algo trading to be taking hold at any point in the future?
 
When everybody begins to do the same thing, there is often a crisis ahead.
 
Primitive algo trading got blamed for the crash of 87, and we still have a legacy of obsolete and/or dumb rules from that, coupled with a lack of any idea of what rules would be good for the future.

Puco has a point about crowd momentum being an inherently dangerous thing, but I believe that not only will proprietary AT grow as a % of volume, it will come to dominate trading.
This is because components of AT are applicable even when we are dealing with investment strategies that have days or even month cycles, not just milliseconds.

Several firms now offer the part of AT systems that minimise market impact as a service. Timing a trade to be at the "right" point over a multi day period follows from that.

Ironically, algotrading may actually be partly driven by regulation.
In Europe there is a directive for "best execution" that requires managers to demonstrate that they got the best price for their clients.
Automation makes this easier, and just as importantly, you have an audit trail that supports the case that you obeyed these rules.
In all markets there are rules about market manipulation, and trade automation is on hand to help you avoid being seen as breaching those regulations.

Also there is more regulation on risk, and more invasive risk management to the point where no trader can possibly understand which trades he is "allowed" to do at any point in time, and the periods for exposure limit updates will move towards being near real time, so that one minute you will be allowed to do a trade, the next minute you will not, then back again, or not.

Humans will struggle to cope with that.

So my view is that regulation will act to reduce "seat of the pants" trading by humans because it is harder to prove that it obeyed the rules.

I recall some years ago a semi-serious suggestion that the US tax code be abolished, and replaced with the software that implements it. A law is of course just a program, so why not cut out the error-prone task of converting sloppy English to code ?

The tax law would be whatever the tax system did, we're almost there already, would save quite literally billions, and probably be fairer.

At one point the British Nationality Act was released as an expert system, and now the points based system to get a UK work permit is in effect a computer system with the people merely providing input.

Isaac Asimov would be so happy.
 
In terms of electronic/algo trading, I believe that one can get sustainable advantage if he can simply execute trades faster than his competitors. But I kind of suspect one can get sustainable advantage simply by "smarter" algorithm ... especially given many participants know roughly what the others are doing and, more importantly, it is hard to argue a person is definitely smarter than others and he is also consistently lucky in trading.
 
zhouxing is right that sustainable advantage is usually impossible.

AT mostly works by exploiting inefficiencies in the market, and by their nature, they go away. I disagree however that competitors asccurately know what you're doing.

Certainly that happens, but the dynamic nature of both the systems and the market make that hard, and potentially dangerous to act upon.
 
True, it is not easy for competitors to know exactly what you are doing (especially given you may continuously polish your ideas). But competitors in the same category may know roughly what you do. And, interestingly, I think many people in the same field may even independently come up with similar, though not exactly the same, ideas because they may try to exploit / solve the same problems.
 
ranald said:
Does anyone else suspect some serious regulation on algo trading to be taking hold at any point in the future?
Click to expand...
No kidding. We have a NYT article on this "fast machines" of Wall Street. We have a US senator demanding some investigation into this. And everything is right after someone "stole" the code from Goldman Sachs money making algo machines.
This does not bode well for hifi trading, at least for the short term. In the long term, I don't know. One thing I know is that when something becomes main stream, you no longer can make money (or maybe you can but a fraction of previously).
Too much hysteria in politics today to have a stable market.
 
Andy said:
One thing I know is that when something becomes main stream, you no longer can make money
Click to expand...


the more people know about it, the more people will "emulate" such thus carelessly using it. But I must say it takes experience to really make money using AT.
 
Back
Top