Comparison of Portfolio Trading and Testing Platforms?

Discussion in 'Trading Software' started by tickzoom, Jun 24, 2009.

  1. A search on elitetrader for a specific and recent thread that compares portfolio trading platforms turned up nothing.

    So several people interested in portfolio trading systems have created a draft comparison chart.

    Please check out the portfolio trading platform comparison chart at:

    http://www.tickzoom.com/wiki/TickZoom/Comparison

    Each platform has pros and cons as well as different pricing so it will be very useful for ET users to get a clear, honest comparison in this thread. Any corrections and suggestions get updated accordingly to the chart at the link above.

    So far, the only platforms found that have any serious portfolio features are:

    TickZOOM
    Portfolio Maestro
    Trading Blox
    NeoTicker

    What others can be added to the list?

    NOTE: The chart above originated from a chart on wikipedia. But it had many other types of software including libraries which aren't exactly platforms.

    Plus it lacked a column to compare whether platforms have portfolio testing or trading features.

    Sincerely,
    Wayne
     
  2. Dennis sent this via email to update the MultiCharts info in the comparison.

    Thanks, Dennis!

     
  3. Dennis, I'm updating it for MultiCharts and have questions:

    By the way, I suggest you update the WikiPedia chart.

    First

    Based on your note above, I am moving MultiCharts up with the portfolio software and put yes everywhere for MultiCharts except for the following:

    Options: Blank
    Tick Data: No
    Open Source: No
    Last Release: Blank

    Can you fill in the blanks please? So, does it support options? And what's the last release date?

    Also, please verify all the yes and nos.

    Second

    We want to change the requirement for qualifying to automated trading to fit what most automated traders expect. So does MultiCharts fit the new definition below?

    Most automated traders have the following 4 requirements for truly automated trading:

    1. Software can start up and run without a GUI interface whatsoever. In other words, it starts the portfolio, strategies and connections purely from configuration files.

    2. It must have robust logic for restarting and resetting in the event of broker connection or network service interruption or even hardware reboot.

    3. During any restart, it must be able to reload gaps in tick and resynchronize the positions and orders in the account with the current portfolio status.

    4. The automated server must have some other facility and/or API for monitoring its status which includes trades and performance stats. That facility could be a separate GUI, a web interface, file, or database output and preferably all of the above as choices.

    Question

    Dennis, most of the platforms that claim to support automated trading never of satisfying some of these requirements for fully automated trading.

    So we plan to add another column entitled "Fully Automated".

    Does MultiCharts fit the following fully automated requirements?

    MultiCharts vs. TickZOOM

    By the way, I used multicharts also, and it's a lovely system. Especially the elegant and direct support for EasyLanguage.

    It has awesome discretionary trading features.

    In contrast, TickZOOM has zero discretionary features and focuses only on perfecting the features for total automation of portfolio trading.

    Sincerely,
    Wayne
     
  4. Pippi436

    Pippi436

    "Must be able to process at least 10 million ticks in less than 60 seconds to qualify for a yes"

    Process how is the question i guess. You mean just load a tick-chart with 10m ticks or actually do something with them? For a meaningful comparison there should be a simple test-case + same computer specs.
     
  5. byteme

    byteme

    LOL. Exactly. He just doesn't get it.

    He keeps regurgitating these figures as if they are supposed to mean something in a vacuum.
     
  6. Pippi, you can always be counted on for excellent suggestions.

    Rules for Benchmark

    The benchmark must load and run at least 10 million ticks in a historical test of a single strategy using hourly bars (built from the ticks "on the fly") and generate a reasonable number of trades. It must generate a chart of the hourly price data and all the trades generated plus some stats and trade reports. The 10 million ticks must cover at least 3 years of time span.

    The system must truly use the tick data during processing to provide accurate (not simulated) entry and exit prices for the limit orders (strategy to use is below).

    All of the above gets including in the processing time.

    If a platform can apply multi-threading and multiple-CPUs, that is fine and must be stated in the result.

    TickZOOM Result

    The TickZOOM test above was done using a single CPU on the quad core with the following specs from a Dell Inspiron 530 machine:

    Intel® Core™2 Quad processor Q8200 (4MB L2, 2.33GHz, 1333FSB)

    The TickZOOM test was performed using the following strategy code:

    Code:
    public class BenchmarkSimpleStrategy : StrategyCommon
    {
    	public override bool OnIntervalClose()
    	{
    		Enter.BuyLimit(Bars.High[0]);
    		Enter.SellLimit(Bars.Low[0]);
    		return true;
    	}
    }
    
    How the strategy works

    The default bar interval is hourly and so this strategy OnIntervalClose() gets called at the end of each hour to make trading decisions.

    TickZOOM builds hourly, daily, session, weekly, monthly and yearly bars from the tick data (not required for a benchmark).

    That should be enough info for another system to run a benchmark. If more info is needed, let me know.

    Wayne
     
  7. dgloba

    dgloba

    Wayne,
    Thank you for the consideration. I’ve updated the wiki chart.
    1. Options are not supported.
    2. I would say yes rather than no, because I applied a simple indicator that calculates Close*2 and it takes 90 seconds to process 10 millions. It is important to mention that I run an old single core PC with 3Ghz with a hyper threading architecture. I know for sure the new processors at least 1.5 quicker or more. So you decide.
    3. Open Source: No
    4. Last Release: 05.23.2009

    MultiCharts auto-trading must be controlled thru GUI. It has robust connection loss detection and it allows you re synchronizing of strategies’ positions. We don’t have position/order tracker yet, but it will be introduced this summer. We do have API that allows trader to control auto trading and write proprietary trading gateways.
    Strictly speaking MultiCharts doesn’t fit all your criterions so you may set “No”in your table, but I’m not sure it is correct choice, because MultiCharts is used by hundreds of fully automated traders. Some of them run strategies 24/7.
    Thank you!
     
  8. dgloba

    dgloba

    P.S.
    I tried to do the same exercise on another PC which is old too, but better than mine.
    It is AMD Athlon(tm) 64 X2 Dual Core Processor 3800+ 2.01 GHz, 1.5 Gb RAM

    The same calculation (Close*2) takes 42 seconds. I think the latest PC will be quicker...
    So hardware is a very important factor in this consideration..
     
  9. Hi,
    I must confess I'm a bit confusd by your message here. On one hand you claim to be comparing portfolio trading and testing platforms but I see entries in this thread relating to tick data throughput, ie. you're adddressing intra-day timeframes.

    Given that, why do you include the likes of Trading Blox in your list? It is eod only and so you're not comparing apples with apples.

    That said, I will suggest another to add to your list - until it is properly rationalized:

    Tradersstudio:

    Going by the columns of your table...

    1. Yes
    2. Yes
    3. Yes
    4. I think it can do this, I don't trade stocks
    5. No
    6. Yes
    7. No - They have auto-execution under development I believe
    8. Don't understand difference between this and previous column.
    9. No - A Real-time version is under develpment. I don't know when it will be released.
    10. No
    11. No - An API is provided
    12. Can't remember last release date - late last year I think.

    However, I think this type of check list is pretty pointless. How a platform fulfills these requirements is fare more important than whether it has them or not. And flexibility, ease of use and programmability are critical too. Trying to make this a kind of black & white beauty contest is ultimately self defeating...


    Thx
    D
     
    #10     Jun 25, 2009