Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
Can I test a stretegy using MY datafeed But ANOTHER broker demo account?
I actually have Webank as broker and datafeed. I'm having some BIG difference between the backtesting results of my strategy and the same strategy applied to a chart with real time data.
For example I'm leaving multichart open with the Mini Dax chart 30 tick with my signal, for some hours in realtime (Without autotrading activated). then going to view strategy performance report I can see for example -400 euro nel loss and a list of operations.
Then if I simply Re-compile the signal in the powerlanguage editor then all the performance are Re-Calculated (because the data that it was using in real time are NOW hystorical recorded data) and then going to view strategy performance report they are completely different and I see + 600 or earnings (it's just a number, point is that I see very good performance)
So that's discouraging... I can't trust those backtesting result until I discover what is the cause of this discrepance.
I was thinking to use a DEMO BROKER account with VIRTUAL money to test my strategy in real time using my Datafeed.
Is that possible??
Can I use my DATA FEED and buy/sell on ANOTHER virtual free demo account?
Thanks in advance,
Regards,
David
Can you help answer these questions from other members on NexusFi?
Your issue sounds like a common one - the strategy backtest engine calculates different trades than what occurs in real time. This could be caused by a few things, data being just one of them.
It could be, for example, your profit and stop levels, where the strategy backtest might think your profit target is hit first, but in real time your stop gets hits first. It could also be your entry rules - it is impossible for me to say.
In effect, you might be unintentionally "tricking" the backtest engine to give better results.
Try simplifying your strategy, use only market orders, then slowly add components (stops, targets, etc) until you see the discrepancy disappear/reappear.
For years, I only used market orders in Tradestation, until I felt comfortable with live vs backtest performance of strategies.
Without knowing more, I would trust the real time results more than the backtest.
First of all I have IOG true because I need my conditions to be calculated on tick basis (maybe in the future I could change it for test).
At the moment I'm already using "buy 1 contract next bar at market" as you suggested.
And for exit position I use "sell 1 contract Next bar at Open <---------- could be this the problem?
Then I have a SetStopLoss(some$$$) and that's it!
I left multichart open since this morning at 8:15 with 6 different charts with different time frame and I'll take a note this evening at 19:30 of the performance of the signal applied to each charts.
Then I'll recompile the signal so the strategy will be calculated over hystorical data (because the real time data of today are going to be hystorical data in the evening) to see the difference in NET PROFIT and number of operations. I'm sure that the different will be huge.
After some thoughts my questions are:
1) Based on your experience, a test made in realtime (and not using just hystorical data) how much could differ from reality? I think it should be pretty the same.
2) Am I correct by saying that " the more higher the chart time frame is, then closer the results will be compared with real autotrading".... Example: by using 30 tick or 1 minute bars it would be diffucult to simulate exact entry and exit prices (expecially with non-regular chart types like heikin ashi) .... but with 3 , 7 , 9 minutes bar the entry and exit levels should be easier to calculate..
That's interesting... I definitely have to try the strategy with IOG disabled and see the difference!
thanks again!
can I just ask to reply to this question? (I'm curious)
1) Based on your experience, a test made in realtime with chart and signal WITHOUT AUTOTRADING (and not using just hystorical data) how much could differ from reality? I think it should be pretty the same...
This is really hard to digest. Even a realtime simulation could be so different from real... so the only way to test your system is to GO LIVE... ( at least for system like mine)
After the real simulation of yesterday I have to completely discard 90tick and 130 tick charts... even 1 minute charts gave me horrible results...(of course with hystorical backtesting all of them showed magical results eh eh )..
but 3, 7 and 9 minutes real time simulation charts close with positive earning the day.... 7 minutes seems to work better on my strategy... 10/13 operations and a gain of 317 euro on minidax...
Then of course I've tried to REALOAD the backtest at the end of the day and it gave me 800 more euros of gain...
For my kind of strategy I think that backtesting on hystorical data is totally a waste of time...
The key is to create a strategy that performs the same way live as in backtest. It can be done.
Try 1 minute and higher bars. Use only market orders to buy/sell. No IBOG. This will work. Then, once you get comfortable, you can start doing more complicated order types.