For several months now I have been developing an automated strategy to trade the EURO/USD. I've bactested and further optimized the strategy over both long and short time frames. About a month ago, I began running the strategy in realtime. Yesturday, the strategy placed one trade and lost $700 total and did not trade again for the rest of the day because it reached the strategies daily loss limit. However, last night I did a quick backtest of the same day and it showed 2 trades with a total loss of $100. The two trades were not even in the same place as what was actually placed. To say the least, my confidence has been shaken in the NT backtesting tool. Am I wrong to think this way? If the NT backtesting tool is not a good bactesting engine, then what tool should I rely on? Is their another product that works with NT that would provide more accurate results? Thx.
You did not specify any details about your strategy so I am making very general comments about NT backtesting. NT replay data only stored OHLC information so if you are using Calc on Bar Close = False or if you are using anything other than time-based bars then backtesting will not give you an accurate picture. Forward testing realtime in SIM will be more helpful to your cause. There are several threads on futures.io (formerly BMT) forum that talk about NT backtesting pitfalls so I suggest you spend some time searching through the forum. The most common proposed solution is to use a second bar series in your strategy to take trades off (like a 1-bar range)
Your question entails many facets and, without having any knowledge of the strategy you refer to, any answer is speculative. However, the main culprits could be the data series or the inescapable rule which forces backtesting to use COBC. Order type can also be a factor in that the backtester will give you optimum fills that can't be easily replicated in live trading. So, in essence, the backtesting result is like going to trading heaven without dying.
I'm not aware of more accurate backtesting engines for the retail market.
Thanks for your reply. I'm sorry I'm not able to provide more detail about the strategy. I was actually looking for an abstract answer. I have spent a sizable amount of time reading through the backtesting threads. I use the BetterRenko bars to trade on mainly because of the phantom bar issue associated with Renko. I use CalculateOnbarClose = TRUE. I'm not exactly sure I understand what you mean by a second bar series. Do you mean multi time frame. Thanks for your help.
Thanks for your reply. I know backtesting is "not always that accurate" and I know sim trading is far more accurate but you can't beat the speed that backtesting offers. The only reason I asked about other products was that I am familar with NeuroShell and wasn't sure if something like that or Matlab might produce more accurate results. As you suggest, it's probably just the inherent flaw of bactesting and not the product itself. Thx.
As others have said, stay away from exotic bar types (Kase, Renko, 3LB, etc), implement a 1-range second dataseries and execute all orders on that dataseries, and then the backtest will be as accurate as it can be.
A backtest is only a small piece of the puzzle, it just gets you on the road. To drive down the freeway you need to forward test for a while in a live market.
Due to time constraints, please do not PM me if your question can be resolved or answered on the forum.
Need help? 1) Stop changing things. No new indicators, charts, or methods. Be consistent with what is in front of you first. 2) Start a journal and post to it daily with the trades you made to show your strengths and weaknesses. 3) Set goals for yourself to reach daily. Make them about how you trade, not how much money you make. 4) Accept responsibility for your actions. Stop looking elsewhere to explain away poor performance. 5) Where to start as a trader? Watch this webinar and read this thread for hundreds of questions and answers. 6) Help using the forum? Watch this video to learn general tips on using the site.
If you want to support our community, become an Elite Member.
The following user says Thank You to Big Mike for this post:
Thanks Mike. I've been using a 5 tick Better Renko bar and as far as backtesting is concerned, it is a very profitable system. Average trade is $205 with a probability of 54%. That said, as much as I try, the bank fails to cash the checks the backtesting tool writes. The 1 tick range bar is an interesting idea. However, I don't understand what you mean by "second dataseries". Are you suggesting just scraping the use of the Better Renko bar all together or use it in conjunction with the 1 tick range bar. Sorry for my thick head. I'm sure my wife will simpathize with you.
If I were to use just a 1 tick range bar, that leads me to believe that the targets will be much smaller and the number of trades daily would be much higher. Not that I have a problem with that but I am just thinking through the idea. Thx.
Thanks for your reply. I'm most interested with your last note on IB. While I agree that the data sampling issue is not great, FX cash data is inherently flawed anyway. Is it that you don't like IB at all or just the data they provide. I tried Kenetick out for a month just to see how good it was and it is great data but I wasn't sure it would really make a difference in what I was doing. What I like the most about it was that they provided 4 months of historical tick data. I usually place between one and two trades a day so its not like I'm doing high frequency trading. So, I'm just not sold on the value it would bring me personally. I guess if I was trading a more centralized market at a higher frequency rate, it would probably mean more to me. Thx.