Review Omnitrader

Discussion in 'Trading Software' started by Privateer, Jul 16, 2002.

  1. This is to post my thoughts about OT 2002 EOD version.
    Warning : it's a very elaborate rating.

    Part 1 :
    First time I came across OT was in early 1999. I use only the EOD version for stocks.
    Current release is OT 2002 Version 4.

    OT is imho a very unique program, offering a wealth of possibilities for a technically oriented trader to work with. However, contrary to what Nirvana says, it is not a beginners tool imo.
    One should have a thorough understanding of how the indicators used in OT work, what OT does with these indicators, how the voting process works and can be influenced and how the various other functions work together in order to get the best out of it. After that objective is reached, OT offers the user a very much organized way of managing his trading. It prospects for interesting signals, helps to evaluate and plan a possible trade, and offers the tools to manage the trade after a position has been taken.

    Especially the 2001 / 2002 versions offer some valuable tools I missed in the past versions as a multiday swing trader. After working with their 2002 release for quite a while, my previously ambivalent opinion towards OT has changed to a more positive one .

    I think, a review of this software deserves a bit more room then just a single line or two.
    Let me just say, that I'm in no way related to Nirvana. I have used and tested several of the higher priced software packages as well as cheap TA software. OT is , among others, still part of my arsenal of a few trading tools I kept after all this testing, while I skipped using most products- simply they didn't fit to my trading style or because of a lack of capabilities.
    The following expresses my opinion on OT, release 2002 EOD version - used for short-term trading ( 1 day - 3 weeks ).

    As Nirvana stresses several times throughout their good education files, OT is NOT
    an automated trading system !
    Although all those performance statistics generated by OT might let you think otherwise.
    Don't trade OT's signals just as they are fired - evaluate thoroughly any trade signaled by OT before you pull the trigger ( as you would probably do it with signals generated by any other trading software, I guess ).
    Afterwards, use the wealth of tools to manage the trade ( trendline S&R line alarms, 8ths. Tool, S&R, levels, automated P/L stops, portfolio-functions etc.)

    OT is not a black or grey-box TA-software. The user has access to all of the 120 daily and weekly indicators used by OT. He can change their parameter settings and select & deselect the various system for use in his profile.
    He can choose from a variety of customizable entry & exit conditions as signal confirmation or as triggers for a trade. The user can opt for delayed trade entries & exits to add a more realistic scenario to the backtesting ( one enters a trade ususally the day after a signal has been fired, not on the same day, since the EOD data wasn't available for backtesting until the markets close ). He can also use customized commission-scales and available portfolio-capital for the backtesting.
    He can not develop and apply his own new indicators in OT. Thus, OT is certainly not a developers preferred tool !

    In OT 2002, all indicator (also called "trading-systems" ) signals can now be used as additional confirmation-filters for OT's generated voted signals.
    Weekly signals can be used as confirmation for daily signals and the user can determine the level, to which extend and which of the weekly systems OT shall use for this confirmation process, aside from those choosen by OT diuring the backtest-process based on their performance.
    In addition, one can define stop & delayed entry rules and the type of signals used by selection or customization of the trading model used for each profile ( a profile is the list of stocks / funds etc. you want to let OT work on )

    OT comes with a number of pre-defined profiles and trading-models.
    One can setup an almost unlimited number of own profiles.
    Since the OT 2001 version, several chart annotation tools have been integrated, among others automatic trendline-break alarms, 8th. tool etc.
    Fibonacchi and S&R levels, long, medium and short-term trendlines as well as several candlestick patterns are calculated automatically and displayed on the charts ( daily and weekly mode )

    Back & Forward testing ( - results ) :
    Probably the most discussed subject when people start to rate OT. One has to understand the philosophy behind OT in order to make a valuable judgement on OT's usefulness in back and forward-testing. Otherwise, the ( changing ) results may lead to misunderstandings.
  2. Without going into much detail about the technology used for OT's testing process on more than 100 indicators, let me say, that OT's signal generation process is based an the so called "adaptive reasoning model - ARM " - meaning, that the character of any trading-market changes over time and therefore technical indicators for analysis of this market have to be adapted or even replaced by other ones over time in order to apply only those which really work under current market conditions.( Standard setting : maximum of 6 of the best performing systems, but the user could push OT to use all available system/indicators for this process, although the results are usually worse then with a capped number of used systems )

    This philosophy alone would be a subject for a discussion, separately from OT's performance.

    However, if a technical oriented trader subscribes to this philosophy ( that any single system does not work in all time frames and under all conditions ) he might find a useful companion in OT, since this software is dedicated to do exactly this - finding always the best systems out of the 120 built in for the back-tested time-frame and with regard to user defined additional entry and exit rules ( or any of the standard rules preset by Nirvana in various models ).

    OT does this automatically, each time the user runs a backtest on a set of data, and comes up with a so called voting line signal, which comprises the overall consensus on up to as many systems as you like to include in this voting process with regard to their respective performance filters settings, if the user applied any.

    It is recommended by Nirvana, to run the backtest every 10 - 20 days, in volatile markets even every 5 - 7 days. The backtest-period should not be too long , in order to have OT find those systems which work best at current market conditions. Since the automated selection process includes an evaluation of any systems performance over a given period of data, it might very well be, that i.e. a STO indicator, which works very well good in trading range market and fails in trending markets, receives an underperformance rating when tested on 10 years worth data, simply because the longterm trend has been up in the markets. As this STO indicator fails to deliver good results on the 10 years data, it will not be considered for the further voting process in OT - although it might be, that it is the best indicator on current market conditions.

    One has to understand thoroughly, what exactly happens when OT selects certain indicators and what's the reasoning behind it.
    This has nothing to do with optimizing ( curve-fitting an indicator settings to a given set of data ).

    If, i.e., a market would have had 10 years of sideways action, no trending-systems, other than ST trending, would probably be included in the voting process, because Med - LT trending systems perform poorly in sideways markets. But almost all oscillating systems could be found among the best performers over time and thus used by OT in the final voting process.

    Well, I hope you got the picture about how OT generates signals by applying the ARM's process.
    The generated trading statistics show, how well any choosen set of indicators and the single indicators in the set would have worked over time on any tested market.

    OT selects automatically different systems for any backtested market, based on their overall performance during the choosen backtest-period.
    I.e., if you have 1000 stocks in your profile, OT could theoretically select 1000 different combinations of the 120 indicators, for each stock a different combination - whatever worked best for each stock in the given time-frame.
    This whole process takes only a few minutes on 1000 stocks and say 2 - 5 years of daily data. I don't know of any other programm which can do this in such a short period of time.

    If OT isn't able to find any working systems on particular stock, which match your performance criteria - no voting signal. Also, not every indicator gives a signal each and every day. If OT finds confirming signals after backtesting and taking a trade, it will display them together with an explanation, which systems fired a a continuation signal and to what extend this kind of signal can be considered as valid with regard to it's historical performance ( advisor rating, systems peformance rating )

    By applying certain filters, a trader can choose to have only certain signal-types displayed and / or used for creeating the voting signal - oscillators, trending signals, candlestick patterns - whatever - so he doesn't have to rely on OT's automated voting process only.
  3. Now - how good are those signals ? Well as a first step, a trader could use OT's Game-mode in order to test his personal abilities to interpret trading-signals and charts and to manage a trade successfully. Afterwards he has a good impression how he would have performed as a trader trading OT's signals.
    This game mode is an excellent tool for honing ones trading / chartreading and trade-management skills. Try this one out first, when you do a test of the program. Chances are, when you fail during the game-mode, you will probably not make it in reality-mode. ;-)

    I tested the various OT releases more than extensive over a period of almost 3 years now.
    I applied back & forwardtesting, changing / customizing the various trading-models, indicator parameters etc. on thousands of US stocks. High volume, low volume, volatile, less volatile, high priced, low priced. I used up to 10 years of worth of daily data for many of those stocks.

    My current standard profile consists of about 800 stocks, ETF's and some major market and sector indices, 70% Nasdaq Stocks, 25% listed and 5% ETF's / Indices.

    Here some general conclusions :
    OT ( 2002 EOD ) works best on finding appropriate systems on more volatile stocks ( many of the well known volatile Nasdaq subjects among them ).

    When talking about forward testing in OT, one has to understand, what's meant by this :

    If you define a backtest period of say 500 days and a forward test period of 250 days, OT goes 750 days back, tests all systems of on the first 500 days. The combination of systems found to be the best performing one for this time frame will then be applied unchanged on the remaining 250 days worth of data.

    When I apply very long backtest and forwardtest periods ( say 5 years back, 5 years forward, or 7 years back, 3 years forward ) OT's performance reports ( very detailed, btw ) show, that on about 30% of all stocks in the profile, the average backtest and forwardtest hitrates ( meaning profitabel signals ) tends to be in the 50% range.
    Why only on 30% of all stocks ?
    Simply because many of the stocks in my current profile don't have a 10 years trading history and can't therefore be tested on such a long timeframe correctly by OT.
    Now how about profitability of those signals ( APR - annualised percentage ) ?
    This depends obviously very much on the underlyings overall performance during the selected backtest period and, to a certain extend, on your personal OT settings for stops, entry delays etc. and can thus vary considerably. I apply always two stops for testing : Reversing signal and fixed stop of 2 times the ATR, based on the gainside of a trade.

    When backtesting on 7 years of data and forwardtesting on 3 years of data for the portfolio of stocks mentioned above, you'll see typical results of 20% - 50% APR for the backtesting period and 40 - 80% APR for the forward test period, when filtering only those stocks, who do have a positive performance on back & forwardtesting.
    About 40 - 50% of all stocks, having 10 years of data, tend to generate positive results on both - back & forward testing.
    So in total, about 80 to 120 stocks out of 800 show positive back & forward test results. It can be assumed, that the systems choosed by OT seem to work on those particular stocks reardless of market conditions with an average hitrate of 50% and an average APR of about 20 - 50%. Peaks are hitrates of 70 - 80% and APR's of + 100%

    Such a test comprises of 18.000 - 25.000 trades taken and documented by OT over the course of 10 years data. Testreport is up to 85 pages long - not quite useful for posting here.;-)
    So this whole back& forward testing has probably some statistical value.

    When testing over much shorter time frame, more stocks can be tested by OT, simply because there's more data available. Interestingly, when I tested the profile on 4 years back and 1 year forward, the percentage of stocks with well working systems in both timeframes didn't change very much compared to the 10 years test ( 7 / 3 ). But since more Nasdaq stocks are now taken into the equation, the overall APR results increased significantly for those stocks showing positive results. Also, the hitrate increased a bit to 60% on average.
    For many Nasdaq stocks, OT found systems generating much more than APR 100% in the backtest- period ( not surprisingly ) but also many who showed + 100% in the forward test.

    If appyling even shorter timeframes, say 250 or 125 days, overall hitrate increases significantly, while APR remains in the 40 - 100% range - the moves worth in USD in any stocks are less dramatic these days then in 1999 til 2000.

    So after all of this testing, I would say, that for those stocks showing consistent & good results in back & forward testing, the signals generated by OT are of some significance. and could be considered as valid options for entering a trade.
    The next step is, to trade only those stocks who showed the best results overall - this might be a list of only 30/40/ 60 stocks in the end, out of 800 in the initial profile.
    How to keep those systems ?
    Well, usually, OT's task is, to find always the best systems. Therefore, backtesting should be run very few days..
    But, once I have completed a longterm back&forwardtest, selected only those stocks which worked well, I don't need to have OT to check every other day for better systems.
    Thus, I simply copy those stocks, including all settings, in a new profiles and set the the repeat backtest option on 250 days = no further backtest on these stocks for the 250 days.
    I still run the backtests occasionally on the original profile, simply because over time, OT may find system combinations in back & forward test on stocks which performed poorly in the first place.

    Why is that so ?
    As already explained, OT uses several filters to work with it's systems in order to generate signals. One of these filters is the lookback-period, used in signal generation.
    It's usually set on 5 days and last 3 weeks. Even if a system is has a good performance in APRor hitrate, if it has not fired any signal over the last 5 days for daily systems or last 3 weeks for a weekly system OT does not include this system in it's filter for the current voting line signal.

    Of course, the user can change this lookback periods according to his needs.By changing the look-backperiod however, the current hitrate can also be influenced negatively by considerung old signals for a new vote.
    The 2nd. major filter is the so called "Advisor rate". Standard settings are 60% or 70%. Look in OT's helpfiles, how the advisor rating is derived from the various systems performances over time and how it influences OT's selection process for systems to be considered for generating the voting signals.

    This should give you some insight, of how to tweak OT in a way, that selected systems don't change.
    Either by simply disabling Advicor-rating cutoff or by using higher or lower percentages.

    Another way to influence test results ( and thus performance ) is, to have certain systems always on the voting line, overriding OT's systems selection. This can be achieved by selecting certain indicator signals as entry or exit filters. Unless this particular indicator doesn't fire a signal, OT will not enter a trade, even if the voting line signal clearly indicates a long or short trade.
    Furthermore, you can force certain or all systems used in OT to be considered in the signal generation process, by simply select the force option in the trading-signals panel.
    If you'd do that for all systems OT offers, the backtest process is much slower simply because OT applies no past performance filters anymore and has to calculate the overall signal based on 120 different signals, not on 6 - 8 selected ones. Results are not necessarily better as with unforced, filter on mode.

    There are many more ways to "customize " OT to the users needs, but be aware, that every change in your settings will have a certain impact on the performance results ( but can also add some more realistic views on them )

    Summary :
    OT can be a very versatile trading tool for the technical oriented trader , who does not want to develop his own trading systems ( which would be not necessarily better than the ones included in OT btw. ).
    It is easy and convienient to use and comes with a comprehensive manual ( online and offline ), allowing the user to get used to OT in very short time frame.
    However, due to the complex way OT selects tradingsystems and generates signals, a user should have some profound knowledge about TA already, in order to be able to intepret signals and charts succesfully.
    OT offers quite a lot of features to manage ones trading effectively ( portfolio, P/L stops, alarms ) and even to train ones TA skills ( Game & Labmode ). It's automated charting capabilities makes it easy to focus on most important aspects of chart ( trendlines in various timeframes, candlestick-pattern recognition, S&R, Fibs , P/L stops, Alarms etc. )

    With a decent background in TA and some indepth knowledge about OT's internal process, a trader can focus on what's most important - trading, without having to think to much about how to select the best indicators, backtest all day long etc. ( paralysis through analysis syndrom ).
    Therefore, my rating is thumbs up for OT ( 2002 EOD version ).
    OT 2002 release 4 ist stable and works well under Windows XP.
  4. Bsulli


    for intraday futures trading analysis? I've sat thru a webex demo , but the gentlemen doing the demo was a swing trading stock guy and completely avoided doing the demo with streaming realtime data which bother me greatily when he knew going into it that I only wanted to see emini's. The only intraday reviews of the product I've been able to find are mostly comments about the signals being late. I would like the opinion of someone who uses the Ominitrader intraday on futures if your out there and would like to comment.

    Looking for an unbiased second opinion. Thanks in advance.


    There are things about the product I like.

    Privateer thanks for your posts.
  5. Hi Bsulli,

    sorry, but can't help you on that. I don't use OT for Futures and not the realtime version. Tested their RT version briefly when it was brandnew , together with RT quotes, which happend to be a desaster at that time anyway - so I wasn't really able to put the blame on Nirvana.

    I know from the OT website, that there are several people mentioned using OT RT for futures. Maybe you can mail Nirvana that you want to get in contact to those users.

    In their newsletter for the 2002 version, the mentioned some significant improvements for the RT profiles ( new / other indicators, faster signal generation etc. )

    I'd say, the best way to check whether OT is right for you or not is, to use their 60 day free trial.

    And indeed, there are certain aspects in OT which prevented me from dropping the program into the trashbox as stated in my review above.

    What's interesting is, that there are literally no usergroups on the web to be found for this software. There were 2 or 3 attempts in the past to open such forums, but with the exeption of a few posters in the startup phase, these forums dried up quite early after they started.

    On the other hand, when Ed Downs calls for seminar, he seems to get a large audience. Also their Nirvana's userclub seems to have a long list of international members - despite the hefty entrance fee of several thousand bucks.

    From time to time, I get the User's club's news letters as a teaser to join, but I'm happy with what I have and see no need to spend more money then necessary on OT. It does what I expect it to do and that's about it. Whether a signals perfomance is 5 or 10% better due to the ARMS II model, which is free for the Users club members, doesn't matter that much to me.

    One has to evaluate any trade carefully before stepping in. Whether one applies OT's signals or any other indicator. I don't interprete too much into technicals - it's certainly necessary to a certain extend, but there's no such thing as a 100% sure trading-system - otherwise, the one who used it first would be a billionaire in no time and nothing is left for the rest of the market ;-)

    BTW, I don't subscribe to Ed's philosophy of the "1% - 5%" daily gains vs. buy & hold either.

    Theoreticaly, when starting with 10K capital and compounding the gains, you'd have the market in your pocket after only a few hundred trades.

    Unfortunately, reality is far from that. So I use OT and some selected other tools only to prepare myself for trading and to manage the trades. I don't expect OT to take the trading decisions for me. I'm the one who might loose some money - not Nirvana - whatever capabilities OT ( or nay other software for that matter ) might offer.

    Although, I must admit, some system-combinations found by OT during those extensive tests were really intrigueing me to follow simply their signals. But I'm not a robot either, so I didn't do it.
    And actually, OT's signals were right in the end and I wasn't - but I would have had to withstand certain drawdowns between the signals wich I couldn't accept mentally due to my personal risk tolerance.

    Probably a problem, many systemtraders encounter from time to time.

  6. Bsulli


    I have subscribed to his daily signalwatch for several years. Very worthwhile for a free email service. I talk with them several times a year and as for any referencable users they refuse to give out any names of users that I can talk to. I agree about the user forums and most of them dying off after starting. I'm curious as to why the ARMS II cost the extra big bucks to have access to. Are they selling the second tier goods in shrink wrap and saving the better product for the loyalist only? I've read several of their articles on the ARMS II stuff and that seems the better of the two products, but it bother me to have to use one product to get the other.

    The few comments I've read here on ET have been consistent on the lagging of signals for futures trading use in realtime. Hopefully some other user here on ET that uses it in RT with futures might comment.

    Thats want I holding out for anyway.

    Thanks again Privateer.

  7. Bsulli


    I've also trialed his intraday messaging service and that can be helpful as well at time. I noticed on that service he mostly calls the signals using a one minute time frame and even those at times are posted late.

    Anyone else used the intraday signal service with good results?

  8. ARMS II

    Well, as far as I know from several email conversations with Nirv staff in the past, ARMS II has been an ongoing development project - outside of the "official" product line.

    By joining "the club" members contribute to the financing of this development-program. And are of course entitled to use the released ARMS II versions in their day to day trading.

    It's not that Nirvana maintaned two classes of customers, because in the beginning of the ARMS models, the outcome was totally unclear and nobody knew at that time, how much better /( or worse ) this model will be.

    I'm not up to date with my info on ARMS II and I'm also not eager to shell out those big bucks for testing it.
    I mean - there's no refund other than hopefully better trading results and - of course, being a member of the Club - whatever this means.

    Well, maybe one day I might decide to test. If it's as good as it sounds, it might pay for itself in very short period of time ( one or two good ARMS II signals would be enough to pay for ARMS II, ha,ha ).

    Good luck on quest.