High Frequency Trading algorithms are not exotically advanced machine learning techniques, they’re shockingly simple

View Comments

Here’s an interesting little weekend item on High Frequency Trading algorithms, in which the writer says that the best HFT firms now trade in around 2 microseconds (“context: sound takes about 150ish microseconds to travel from the back of your throat to your mouth”) in signal to trade time, and that doing advanced calculation work on a computer takes a lot longer than that… and thus

  • The strategies employed by the most advanced HFT firms are shockingly simple in terms of Maths or Stats: think exponential moving averages, with a handful of if statements and safety checks…
  • when faced with the decision to produce a reliable, stable strategy that provides good liquidity, or simply the fastest adequate one possible, speed wins every time.

More here

 

Author: Eamonn Sheridan

Eamonn Sheridan worked with Bankers Trust Australia for 13 years as a Spot foreign exchange dealer, trading across all major currencies and all time zones. He rose to a Vice President position, running spot operations during the busy European time, leaving the bank just prior to it being sold to concentrate on running his own business in the ‘real world’! The markets, however, had him hooked – he continued to trade equities, CFDs and then on to futures, giving him broad experience across financial markets. He is now active in FX and equity index futures as well as writing for ForexLive™. Eamonn is a graduate of The University of Melbourne in Australia and lives in New South Wales.

9 Comments

  1. Complete nonsense and from an anonymous source!!!! Next you’ll try to tell us that microsecond computer news analysis does not occur, and there is no such thing as computer programed stop loss hunting, and brokers do not enter trades against their clients, and there really is a Tooth Fairy.

  2. not sure about advance thing…i’ve been experimenting MT4’s homegrown Expert Adviser fer quite sometime now with consistent profit. I believe algos are the same thing albeit with a little bit more math and programming sophistication. If a restaurant line cook like me can do it, how hard can it be, rite?
    …..no? ;)

  3. If algos where so sophisticated and brilliant why would they need to be faster than the competition if they were already smarter ?
    We all know that they’re winning by rigging the game, not by playing by the rules…

  4. Even if trades are scheduled every 1 second or 10 seconds there s in any case a queue FIRST IN FIRST OUT so speed to jump in the queue before your competitor still exists , what is currently under deployment is to apply a random delay of 1 to 3 milliseconds to every trader to crumble down the speed race , Biggest banks are working to get this in few months , get ready!!!

  5. Interesting. However, it doesn’t explain why the vacancies I’ve seen at at HFT firms such as Tradeworx (http://www.tradeworx.com/) have required an M.Sc/PhD level in maths/stats or some related quantitative field. If the algorithms are so simple why the need for such high levels of quantitative knowledge?
    Further, I do happen to own an algorithm based trading system. It is not HFT based and may not be as sophisticated as an HFT algo but it was developed by a trader with a very strong quantitative background. It certainly was not simple do develop and is the result of two years of backtesting.

  6. Different people have different strategies. You can take low latency execution advantage from different ways. You can analyze HFT data at different levels and methods. No ways are impossible.

  7. High Frequency Trading algorithms are not exotically advanced machine learning techniques, they’re shockingly simple., 60 minutes had a segment on trading computers used by the banks and trading houses.They take the brightest and best from the colleges to write the trading programs for them.It isn’t a simple thing to write these programs. Some programs are event driven ,so the time you hear about the
    news the trade is already made.Then are banks of computers that even make small trades that add up to big profits.I had posted before ,there is off the New Jersey turnpike in a nondescript building the size of almost three football fields of associated servers and large banks of computers that do HFT trading.Approximately up to 60 % of the trading volume is done by HFT and Banks computers.They are in charge ,you the investor and trader are just along for ride wherever it takes you.

  8. IMO (based on what little I know about the subject), the actual algorithms used to execute trades are very simple. However, this type of trading is still far from “simple.” For example, the research used to chose what algorithms to use at any given time may (in some cases at least) be quite complex.

  9. Why people are still misconceptions with high frequency trading strategies and low latency strategies ? High frequency trading strategies are far more broad than low latency strategies.

Top

© Copyright 2014 ForexLive™  |  Advertise With Us  |  Login To Comment  |  Sitemap

HIGH RISK WARNING: Foreign exchange trading carries a high level of risk that may not be suitable for all investors. Leverage creates additional risk and loss exposure. Before you decide to trade foreign exchange, carefully consider your investment objectives, experience level, and risk tolerance. You could lose some or all of your initial investment; do not invest money that you cannot afford to lose. Educate yourself on the risks associated with foreign exchange trading, and seek advice from an independent financial or tax advisor if you have any questions.

ADVISORY WARNING: FOREXLIVE™ provides references and links to selected blogs and other sources of economic and market information as an educational service to its clients and prospects and does not endorse the opinions or recommendations of the blogs or other sources of information. Clients and prospects are advised to carefully consider the opinions and analysis offered in the blogs or other information sources in the context of the client or prospect's individual analysis and decision making. None of the blogs or other sources of information is to be considered as constituting a track record. Past performance is no guarantee of future results and FOREXLIVE™ specifically advises clients and prospects to carefully review all claims and representations made by advisors, bloggers, money managers and system vendors before investing any funds or opening an account with any Forex dealer. Any news, opinions, research, data, or other information contained within this website is provided as general market commentary and does not constitute investment or trading advice. FOREXLIVE™ expressly disclaims any liability for any lost principal or profits without limitation which may arise directly or indirectly from the use of or reliance on such information. As with all such advisory services, past results are never a guarantee of future results.