Each time frame has it’s own data set, so even though i uploaded the 1 minute file(2001-2013) to my 1 minute history, my 4 hour history still only goes from 2009-2013, as you can see from the screenshot. It seems like each time frame requires it’s own history file.
I grabbed all of the 4 hour data and started characterizing it, Avg Volatility, Std Dev. Volatility, Z Score, Normal Dist, Z Score from Rolling Avg., etc. I haven’t had a chance to look at all of it, but I did notice a few things. On some of my optimization tests, my program did extremely well except for a chunk of time in mid 2011 - 2012. The first thing I noticed is the garbage data in that time range. I am going to assume it did poorly in that time frame because of the bad sections of data. This is encouraging but my program essentially optimized bad data, and still did well. The other thing I noticed is the volatility decreases into 2012. I am going to optimize for each range of volatility 7-8, 6-7, 5-6, 4-5, etc. that way my program changes to it’s most optimized settings for the given volatility at that time, and may have it completely shut off if volatility drops too low(below 2.5?). I may try to create a second EA that takes advantage low volatility periods, and have it turn on when volatility drops out of the range of this EA. This is from my assumption that changing volatility affects the success of an EA more than any other market dynamic.
Not looking to hijack, but do you believe the M1 data from that source is accurate? I pulled m15 data from another source and found it to have problems - chunks of dates missing. The data from the testing site had some small anomolies (especially closer to 2001), but overall seemed complete I have pulled all pairs of M1 data from the source referenced in this thread and am recompiling it into the following timeframes for all pairs: M5, M10, M15, M30, H1, H4, & D1. Let me know if you have interest…sorry again folks…carry on 
edit: the D1 data will be normalized from GMT 21:00 or GMT 22:00 depending on NYC Daylight savings time…although other open/close times could be easily done
Yes that is correct… make sure your MT4 terminal is offline or disconnected from the broker when you tinker with history files.
Then wipe all the originals upload m1, load m1 chart and run the “period_converter” script which is included with MT4.
MT4 terminal that you use for backtesting should always be offline and never signed in.
Krugman, if you haven’t already, you might find my thread helpful for developing your strategy. Best of luck!
Yes, now I remember, I did have to do these steps…
This is so nice of you to put all this together.
I’m saving your thread and if anyone ever need any help regarding back testing I will send him/her to your forum.
Thanks for sharing!
Thanks Clark! I have read through your thread a little bit, what a gem! I think my goal for this weekend will be going through your thread and getting everything setup per your instructions.
I finished updating the EA which now changes it’s setting based on how the market conditions. The picture attached has it on a more aggressive 5% risk per trade MM strategy.
This is kind of funny, I didn’t realize that I was running tests on the EURAUD(I thought I was on the EURUSD). My EA did extremely well, perhaps even better than on the EURUSD. The test period is from 09/3 to 12/1, it took about 2.5 times the trades my EURUSD test did in the same time period, and with a smaller draw down. Maybe it was meant for the EURUAD??
hi, krugman:
have been following your posts (both here and on FSO’s thread) with great interests. Just curious: since I am trading manually with price action and also looking for an EA based on ‘price action’, is it possible for me to get a demo version of your EA, and if it is profitable, I will consider buying it from you? THX
[QUOTE=“zhangshuo;562756”]hi, krugman: have been following your posts (both here and on FSO’s thread) with great interests. Just curious: since I am trading manually with price action and also looking for an EA based on ‘price action’, is it possible for me to get a demo version of your EA, and if it is profitable, I will consider buying it from you? THX[/QUOTE]
Hey Zhangshuo, thank you for stopping by. I noticed you saw my posts on FSO thread, have you had a chance to check out my thread “Price action that matters”. It is a great price action thread.
I believe in a well rounded basket of trading methods. That may mean trading price action, using a successful EA or rule based system, and other trading methods. This can help boost a traders profit and help your losses because when one method is losing the other could be winning. That is why I have been passionate in both price action and trying to create a successful EA. I will consider giving out a demo version. I have finished up both the EURUSD version and the EURAUD version of the EA, and I plan on starting them up tomorrow on a demo and posting the results to the public on myfxbook. I will keep posting updates and improvements here I this thread. Thank you for the interest in it.
hi, krugman, thx for your timely reply. Just like you, I also believe in trading price action and using EA. That is why I am interested in your EA. Definitely will be stopping by your other thread. Please keep us posted. THX. And if your EA is indeed good with value, I will consider buying it if you ever decide to sell it.
[QUOTE=“zhangshuo;562781”] hi, krugman, thx for your timely reply. Just like you, I also believe in trading price action and using EA. That is why I am interested in your EA. Definitely will be stopping by your other thread. Please keep us posted. THX. And if your EA is indeed good with value, I will consider buying it if you ever decide to sell it.[/QUOTE]
I look forward to seeing you in the PA thread. I know it works well on almost 5 years of back testing on 2 different pairs. If I can see it working well on live testing that is the proof I’ll need to know it actually does work, and would most likely offer it for sale and let others benefit from it.
I am going to be testing it on the USDJPY and and GBPJPY also, I plan on this EA being able to run on 4 or more pairs at the same time.
Hello Krugman. Im a newbie learning PA following both yours and Jonathon’s thread. I feel very fortunate having people like you sharing their knowledge for free. Thank you
Can you give us a link to your MyFxBook profile so we can follow your indicator’s progress? It sounds very promissing
Sorry for my english - it’s not my native.
Regards,
René
[QUOTE=“moretti12;563523”]Hello Krugman. Im a newbie learning PA following both yours and Jonathon’s thread. I feel very fortunate having people like you sharing their knowledge for free. Thank you Can you give us a link to your MyFxBook profile so we can follow your indicator’s progress? It sounds very promissing Sorry for my english - it’s not my native. Regards, René[/QUOTE]
Hey moretti, I am glad you have been enjoying and benefitting from my thread! It has really been developing into a great resource for price action. I will get a myfxbook link up here and put it in the first post for everyone to access easily. I am still trying to figure out how everything works, so bear with me for now. I have never used mt4 for trading or myfxbook before so I am still learning how everything works. Keep checking in for updates. I hope to see you in the price action that matters thread!
The EURAUD cross pair has been in a slightly more trending regime over the last several months. I expect that your trading model is based on trend continuation or a breakout with the trend rather than a contrarian style. As a result it would be expected that EURAUD realizes a higher return compared to the more mean-reverting EURUSD. I think you’ll find that during the backtesting period you posted, if one were to run any style of trend-following based model one would see similar returns (~1.55-2.25 profit factor).
Hindsight is always 20/20 of course. 
Top = Symbols
Side = Scope (timeframe)
Red = mean-reverting/ranging regime
Green = trending regime
This is personally something I’ve developed to see the type of market we’re in very quickly. You can kind of tell that the EURUSD in the long term has been more mean-reverting (monthly and weekly scope) whereas EURAUD is just slightly more tending (differs in the weekly). Feel free to use the image I provided to test other supposedly suitable symbols or even test symbols that are supposed to not be suitable for your strategy. You might find a pattern, or maybe just an interesting coincidence. 
Also, I find it very interesting that for the most part, all the symbols are “red” on the monthly scope since mid '07-08. It’s no wonder mean-reverting systems have been on the come back in the last several years…
I “optimized” this with the data I had going from 2009 to 2012. I was able to get 1 minute data and use a period convertor to go all the way back to 2005. I ran it from 2005 to 2009 and it produced the same results. This is the outcome I am looking for essentially right? That my EA ran just as good in data that it was never optimized for as in data that it was fitting for. This either means that 2005-2009 were identical to 2009-2012, or my EA truly works in a way that it fits the changing markets. What do you think? It seems like if I optimized my data for 2005-2009 and then just ran it from 2009 to present and it continued to produce the same high quality results, that is very close to a forward test, because essentially the EA was never told the settings that would produce the highest wins from 09-13?
Honestly, I couldn’t tell you without actually taking a closer look at your model.
But, it seems to pass a certain robustness level, which could be due to the fact that you are using dynamic SL/TP rules which, from what I can recall, was based upon recent volatility.