Day Trading Systems

Discussion in 'Trading' started by Tahoe, Nov 21, 2001.

  1. neo_hr

    neo_hr

    Hi

    Heres the code for the Oddball system :

    { Data1 = Either the SP Ccash S&P 500 index tracking stock or the S&P 500 commodity futures contract. Data2 = Advancing issues of the Nyse. * this was added after article was released to be sure that a buy-sell does not occur on the closing bar. }

    if time >830 and time<= 1500 then begin

    { based upon (GMT-06:00) Central Time (US and Canada) - note you may have to adjust for your time zone. }

    Inputs: RL(7), BZ(3), SZ(1);
    If ROC(Close Data2, RL)>BZ Then Buy;
    If ROC(Close Data2, RL)<SZ Then Sell;
    end;


    Could anyone of you fellas explain in ENGLISH what does this mean and how could anyone make use of it??? I mean where do I put that code? (Hrm, why did they disclose that if it still works???):confused:

    Alex
     
    #141     Nov 27, 2001
  2. rickty

    rickty

    Alex,

    You didn't copy the code fully. It should be:


    { Data1 = Either the SP Ccash S&P 500 index tracking stock or the S&P 500 commodity futures contract. Data2 = Advancing issues of the Nyse. * this was added after article was released to be sure that a buy-sell does not occur on the closing bar. }

    if time >830 and time<= 1500 then begin

    { based upon (GMT-06:00) Central Time (US and Canada) - note you may have to adjust for your time zone. }

    Inputs: RL(7), BZ(3), SZ(1);
    If ROC(Close Data2, RL)>BZ Then Buy;
    If ROC(Close Data2, RL)< SZ Then Sell;
    end;

    This is written in Easylanguage, which is the name of the computer language that
    is used in Tradestation. Tradestation is a product that allows for trading system
    testing.

    Text between "{" and "}" is simply comments that Tradestation ignores but is useful
    for a human reading the code. The logic behind the program was disclosed fully
    in the Active Trader article. The program uses 3 variables RL, BZ and SZ which
    are set to 7, 3 and 1. These were parameters that I'm sure Mark Brown optimized.
    ROC refers to the Rate of Change indicator (a standard indicator)where:
    ROC = ( Al/Al[n] - 1) * 100,
    where Al = latest number of advancing issues of the NYSE
    Al[n] = number of advancing issues n periods ago. Here n = 7 is used.
    Brown uses a period length of 1 hour.
    So the code says: when the rate of change of the advancing issues from 7 hours ago
    is greater than 3 buy the S&P; and when it's less than 1 sell it.
    I hope this helps.

    Richard
     
    #142     Nov 27, 2001
  3. The aforementioned "systems" and the related algorithms have been floating around for some time in one form or another. I especially enjoyed a recent "**** Trade" presentation where they showcase a more simplified version based on "tiki" and net upticks vs. net downticks ("proprietary" in their view).
    ....anyway....since there is no exact science to how many "bars" "time frames" "ticks" or anything else that will work all the time, we must temper the results with some objective intervention prior to automated order entry (IMO). And, yes, these human-assisted automated programs are getting better!
     
    #143     Nov 27, 2001
  4. MarkBrown

    MarkBrown

    (Hrm, why did they disclose that if it still works???)

    because i am generous and have plenty to spare. i sincerely hope you enjoy it and profit by it. there are free email alerts and one year since release stats there on the web site. the signals are also posted on the web site in real time, all for free.

    mark brown

    http://www.oddballsystems.com
     
    #144     Dec 17, 2001
  5. lassiter

    lassiter

    My current data feed only provides me with 2 months of historical price data on 60 minute charts.

    Does anyone know of a site/vendor where I can obtain hourly historical price data for the past year?

    I would like to take a closer look at the OddBall sytem and its performance.

    Thanks.

    lassiter
     
    #145     Dec 17, 2001
  6. rickty

    rickty

    lassiter,

    I was just thinking of doing the same as you. Quote.com
    has about three years of historical intraday data. I wonder
    what is the symbol for the NYSE number of advancing issues?
    I checked out Mark Brown's web site hoping to find his
    email address for some guidelines here but I couldn't find
    one that I saw to be his personal email address.

    So, does anyone know what would be the symbol for
    NYSE # of advancing issues in Quote.com?

    Richard
     
    #146     Dec 17, 2001
  7. rickty

    rickty

    #147     Dec 17, 2001
  8. jeffm

    jeffm

    For a number of good discussion on oddball, do a search of the omega list archives at:
    http://purebytes.com/archives/

    There are a multitude of tweaks you can apply to change oddball around. Whether or not the system is still tradeable for your personality will vary, naturally.

    I believe the symbol on qcharts for NYSE Advancers is QC:ADVN.NY.

    Just a heads-up...you may have a hard time duplicating exactly the trades listed by mark brown on his website. There are a variety of reasons for that. For example, one reason is that quote.com's ADVN.NY data does not exactly match esignal's $ADV. I tested oddball over a period of time using ADV data from both vendors and got very different results. The signals themselves were pretty close. But an hour difference here or there adds up to a whole lot of variation in the results.

    This is not an endorsement, just a note of curiosity: esignal's $ADV produced much better results than quote.com's ADVN.NY.

    -Jeff

    PS: Just a note for anyone who sees this under "Daytrading Systems"...Oddball is an "always in" system that trades on 60m bars. Its not a daytrade system by any stretch.
     
    #148     Dec 17, 2001
  9. I will second what jeffm said, you will get different signals using QCharts. I tried it today and couldn't duplicate his signals. This is a common problem for data on items like advance/decline, trin, tick etc., and it can be an expensive one. I had a system that worked beautifully on Futuresource data but killed me using PC Quote. If you compare feeds side by side you will see wide disparities in data for non price items.
     
    #149     Dec 17, 2001
  10. MarkBrown

    MarkBrown

    actually it does take allot of adjustments to get things just right. like setting the ts server to time stamp using local time and not exchange time. and i have my chart offset set to -1 because i am in central time zone. i also use an atomic clock to set both my computer time and the data feed server time stamps. you can also modify the session times to begin for example at 9:00 and end at 3:00. fully ignoring the sessions opening at 8:30 and the close at 3:15.

    but the basic thing is what the number of reported advancing issues are at the top of the hour. that is the number you need, every hour on the hour.

    for a daytrading twist of oddball just program it to close all trades at the end of the day. it beats most every daytrading system i have seen. imagine this system trades an index that it does not receive it's buying and selling information from. is that odd or what?

    i have just updated the stats for the year since i have released this system. and those stats are based on the cash not the futures, which moves much bigger. i did not want to hype the performance, it speaks for itself. the trades are there to help people get the hang of the thing without having to maintain it. i sincerely hope that it helps a few aspiring mechanical systems traders have hope that they can build something that will not fall apart.

    this is a very, very tough system to trade. it does not employ stops, trades overnight on and on are the dangers. yet look at it year after year after year.

    ps i saw where someone posted the cta's with a 10 year track record ect. i do business with at least one of the top most performers. if you want to understand the type of background these models have i came from a cta with a 15 year track record with never a loosing year and out performed the sp every year. i used a million dollars a year budget and developed all these models on a ned davis program called technaylzer and a cray t-90 computer. 10 years i spent as a full time mechanical model developer.

    the toughest thing was to program the ndr code into sleasy language for ts for public use. i have the hang of it now though it's called trail and error, forget the logic. i have never developed anything that works using ts or any other program, that would hold up. the ndr program has some tricks and well worth the hundred grand a year it cost (for just the software) only 5 of these are out being used by people like tutor jones ect.

    mark brown:cool:
     
    #150     Dec 17, 2001