stumbled upon this thread by random… nice and concise entry method.
hey doug you can use my FT2 dataset (4hr up to monthly) - 27 crosses, BUT they require me to post 5times before allowing me to post a link. I tried to upload the zip as attachment to no avail - perhaps size too big… so if you want the data here you go: bit (dot) do (slash) 2UUL
hope it’s clear there are just text files inside. the CSV format is compatible with FT2
just remove all your history from FT2 before importing the CSVs. It doesn’t matter at all you won’t have lower TFs if you only enter/exit on bar closes. If you need lower TFs shoot me a pm.
H4 goes back to 2004, higher TFs way earlier (1986 for majors).
[B]All the data has 5pm EST / U.S. DST timestamps.[/B] (=5 daily candles weekly, day/4hr starts at midnight). The data is accurate and correct - actually the last decade has been compiled from ticks collected from various sources. There’s not a single candle missing. I can vouch for the data because it’s my own work.
It’s the latest compilation I uploaded to a friend some weeks ago so the last month or two are missing but shouldn’t matter much - you can get good recent data from any source.
Thank you, jw1981. This is 1-minute bar interval, but I can write a program to convert them to 4-hour bars. I’m going to look at the other poster’s data first, which is 4-hour, but I can use this as a backup. Thank you again for posting this.
I think you are over-complicating it. It’s very simple to just code up this strategy (it’s pretty mechanical), and it’s done (if you want alerts, it can be added to the strategy). Wouldn’t take more than 2 hours to do that. I’ve already made note of this thread and have plans on coding it up in the near future, once I finish my current projects. With permission from Philip, I will release the code as well as all test results for any pair/timeframe/parameter sets.
I agree that it’s mechanical, and welcome your parallel effort. I will be happy to share my code as well; I actually did with an early release in this thread. I have a lot of comments and functions for repetitive tasks, and with all the nuances that have been discussed along the way, it’s about 500 lines of Wealth Lab code, but I think I’m very much in synch with Philip’s algorithm and nuances. About 20% of the code is to calculate Monthly RSI from 4-hour bars, which was due to a technical deficiency in WL.
And BTW, my Dashboard presentation idea is just using my technology, and agreed, need to generate the Alerts from the code (very simple). Adding to a historical DB, and pushing it out to a website additional work. 2 months for me, since I have other business endeavors.
You have my permission. I think the problem for Doug was finding the data rather than the coding.
But let me ask you how would coding C++ then work on Meta trader? and how would you test the results after that?
Hello everyone, I am working out a few bugs right now but I have an EA made. It autodraws fib , places trades , etc. I will be releasing it here when done. @PhillipPirrip, is there anything else that should be added to my EA before release that hasn’t been said by you? I have everything basically done but any input of any other indicators and such that you’d like added , it’s appreciated to know while I haven’t posted it yet.
Not sure what trading environment you are using, and if you have access to historical data. My primary broker is IB, and unfortunately, there is no back-testing facility, and no historical data. WL is something I am very familiar with development-wise, but it requires importing of data. I’m not familiar with FX trading systems that coming with inherent back-testing environments and historical data, since I am not an FX trader (yet). If you have one, then please, give it a shot (if only 2 hours required). If not, then I’m really not sure how you are going to build a facility that processes trades, generates results, etc., in such a short span of time.
Well, courtesy Engine’s provided 4-hour data over 11 years for the major 7 CCY pairs, I was able to run a simulation using the current code base. Again, my version of WL doesn’t natively support FX, so all trades are denominated in $'s (and are traded like shares) based on 10% portfolio allocation per trade, with a starting account of $ 250,000. Monthly RSI(14) threshold values are 33.33/66.67 respectively. The results overall are not good. I inspected the data and it looks good from the sense that there is no missing data, no bizarre looking #'s, etc. - the data sets definitely looked complete.
[B]Overall Performance[/B]:
[B]Raw Return Graph[/B]:
[B]Trades[/B]: (had to save as a .TXT file - really it is a CSV, since this forum won’t accept Excel files)
[B]Action Items[/B]:
I need people to look at some of these trades and see if anything is amiss. Again, this is the same exact code base that was used in last simulation (3 CCY pairs; 2009-2015), which turned up real nice. Keep in mind, that data ends in Jan, 2015. I obviously used different data here, so that is a delta.
Engine, can you comment on the source of this data, and why you feel it’s reliable (not doubting that it isn’t), just trying to get some clues.
Thanks, I may just whip up the code tonight if I have time. Data isn’t a problem for me, I’ve been collecting ticks for the past 7-8 years for 28 FX pairs.
I’ll code a version that everyone can use, not just myself.
Mostly EBS and Currenex tick data with some really small patches from Integral (2010+)… I personally skimmed through most of it for days, maybe weeks (1min resolution) looking for gaps, evident glitches and bad ticks. I’m building custom weighted indices off that data and my system relies on the accuracy heavily so I know it’s ok.
I can upload any pair in 1min resolution - CSV or HST (MT4) + an MT4 indicator counting missing bars; if you have some experience, you will recognize the quality of data with plain eyes.
Yeah this is quite strange. Unfortunately I can’t view the file. What you can do is make sure that the other 26 trades we tested earlier are in this batch as they are. If they are not then there is something wrong.
Again, courtesy Engine’s provided 4-hour data over 11 years for the major 7 CCY pairs, I was able to run a simulation using the current code base. Again, my version of WL doesn’t natively support FX, so all trades are denominated in $'s (and are traded like shares) based on 10% portfolio allocation per trade, with a starting account of $ 250,000. Monthly RSI(14) threshold values were set for this test at 25/75 respectively (prior post was at 33.33/66.67) The results overall are still not good, but improved over the 33.33/66.67 filtering simulation. # trades was reduced from 200 to 83, over this designated period.
[B]Overall Performance[/B]:
[B]Raw Return Grap[/B]
[B]Trades[/B]: (saved as a .TXT file; it’s in CSV format)
I just spot-checked EURUSD and all trades from 2011 to 1/2015 were in both simulations; just that the # of shares were different, since trade size is calculated as 10% of current portfolio worth. This is with different data sets being compared.
Again, same exact code base, just different data sets. But this spot check seems to say that at least the 2 data sets are consistent (whether they are right or wrong is a different question).
Also, the file had to be named with a .txt extension, due to forum limitations. You should be able to open it in Excel. Possibly save the file to your computer, rename the file to have a .CSV file extension, and then try opening it in Excel.
there are just text files inside. the CSV format is compatible with FT2