Several years ago, I became interested in technical analysis and automated trading and I developed a few systems with Metastock. However, the technology did not exist to automate my strategies, and I have a full-time job, so I couldn't trade them. Recently, I've become interested again and have found an overwhelming number of options for data, system testing, real-time trade signal generation and even fully automated trading. In addition, I've found this forum (a wealth of information- unthinkable ten years ago!). I am looking for a platform with the following features: 1. A source of historical data for backtesting. 2. A "system tester" that will allow me to backtest my strategies. This "system tester" need not be overly sophisticated, nor should it require me to learn a computer programming language. It should allow me to use EOD and intraday data. (In other words, something along the line of Metastock.) 3. A means to do real-time, fully-automated trading of my system at a broker that has reasonable commissions, decent customer service and quality execution. In the beginning, I will be trading QQQ and SPY, though the ability expand to a wide variety of futures/stocks/options/foreign markets is a mild plus. 4. Decent margin rates at said broker. My research so far has led me to TradeStation, as it has all of the items above bundled together. However, two issues remain unresolved: A. The TradeStation platform is $99 per month, which is quite expensive for my account balance. B. It seems that IB is much more popular with this forum's users. Therefore, I would appreciate comments from anyone who has knowledge of a product or products that can do all of the above items for less than $99 per month, and I would appreciate comments from anyone who can tell me more about how the following items can be done with IB (as it is not integrated like TS) and how much their IB option costs.
1. Tradestation 2. esignal 3. Wealth Lab I would avoid Metastock. Having been a user of MS since inception in 1985, it does not have the functionality that a "trader" wants for testing, automation, etc and they are slow to make changes. Esignal is quickly becoming a one-stop-shop software with ties to many brokers...very nice. Wealth Lab looks to be on the cutting edge for most all of what you want. And of course Tradestation has been around for years and continues to evolve. If you want to use IB as a broker, I would go with #2 or #3. Just remember you need data, so esignal will have it all and work with IB. The programing of the technicals for both is much more complex that MS, so there will be a learning period to get going. Best, Bob
Does ESig have enough historical data for backtesting? Their weakness has been lack of intraday historical data, but I'm not up to date on them.
Here are some more facts about WLD3: 1. We have added drag and drop programming to our new version. That makes it easier to get started into the language and our users can add new pre-defined rules to the ChartScript Wizard. So just as WLD3 comes with over preprogrammed systems it will come with plenty of new rules soon (currently app. 100). BTW, this is the feature that I love the most! 2. WLD3 allows you to hook up to different data vendors at the same time. So if one data vendor is down you stil have the other one in place. A feature that neither TS nor eSignal can offer and that I feel is very important. 3. WLD3 also allows you to import historical data from any source and test any period of time. 4. WLD3 is currently hooked up with IB but we will add new brokers with time. So again you have the freedom of choice and you might want to have several accounts just to b e on the safe site? 5. Dont forget the only WLD3 allows you to do a kind of system backtesting that no other software (TS or Esignal) does. Regards. Volker
...E-Signal at present only provides 60 trading days of one minute data. I find that this is more than enough to do intraday system development, especialy as I don't want so much data that I am optimizing the system for outdated volatility. However, in an exchange with another intradaytrader poster here (regrettably, I forget whom), I was cautioned that two years of data is better. Something to think about is that a not terribly complex strategy takes 8 minutes to run on 60 days of one minute data on my machine (not a screamer). I may make 50 runs to test and optimize a system, so that is a significant time investment. Consider also that you may have to periodically reoptimize for volatility. Automation is the current problem with E-Signal to IB. It can be done with third party DynaOrder (complex to my eye) until E-Signal fully automates (allegedly next year). BTW, I am absolutely pre-orgasmic about E-Signal's back testing laungage. If you can conceive of it, it can be coded. Best regards. - Mike
Mike. Just take my advice 60 days is a joke for backtesting!!! Also I was under the impression that eSignal was connected with IB naturally? since I am not familiar with the eSginal programming languageand their backtesting capabilities, how to you backtest 30 different markets and use a money-management rule that depents on your overall avaialble account capital/value? How do you do a Monte Carlo Simulation with the results you have? Here is another advice, be careful when using the optimization tool. Take a good look at the 3D graph before you decide to trade a system! Regards. Volker
6 months + of intraday data for esignal will be available by the end of Oct. `03. This will build to 2 years as time passes.
...thank you kindly for the advice. I have certainly had experiences which support your contention about 60 days being a joke, but I also have experiences which contradict it. Do you believe that 60 days is inadequate to develop a one minute intraday system? What about the oft-repeated saw that 20 samples is approaching statistical significance (I personally don't believe that)? BTW, I won't trade on just a 60 days test, I wait until subsequest backtesting, plust watching the system in real time, has confirmed the original test. At the risk of sounding stupid relative to your obviously deep experience, I am not seeking a robust system which works at all times across many markets, like the black box vendors do. I want a system which works at one particular time of day on one particular instrument with one particular account size with my personal risk tolerance (overfitted, haha!). The money management rule is not a classical one, but is a combination of selecting the number of contracts for the account size considering the drawdown with the backtested stop. I am not using an optimization tool per se, but rather multiple backtests with parametric variations in E-Signal. Painful, but it only costs time, of which I have plenty. There is always something else to do while the CPU crunches. Having done optimizations before in a physical system context, I understand the risk of finding a secondary (or worse) "optimum". I confess that I do not understand the applicability of Monte Carlo techniques (I understand the principle from othe contexts) to system development. Could you please enlighten me? Thanks again. - Mike
Mike. I did not want to sound like I am lecturing you, I just wanted to advice you that trading a system is not so easy. My general saying is that if you are happy and succesful then you should be happy. I just dont want you to turn around one day and say "trading systems can never be profitable". As I can see from your last post you don't really trade system but you do have a lot of discretion in you trading and therefor I understand that you believe that you don't need all the tools that one would need for good backtesting and analysing the results. No I am obviously not representing eSignal. Volker