The jury is still out on this long-term. I was able to get my hands on 10-years of historical 4-hour data for the 7 major CCY pairs, and I finished implementing the Monthly RSI(14) filter. The results came back really bad over this 10-year period, but I need to review this data, as it may not be the best. More to come when I have definitive results. My initial back-test (w/RSI Monthly filtering) came back really good over 5 years of 4-hour data for 3 of the major CCY pairs, so I need to find where the disconnect is.
I like this post a lot because you took the system and added to it based on your own experience and preference. Itās actually these details that creates a difference between someone who made a return of 10% and another who made a 120% for the same period.
To answer an earlier question, I havenāt traded at all last week. But March yielded 1640.9 pips for me. February 2078 pips and January only 37 (only one position closed in January.)
So my avg during the first quarter of 2015 was 1251 pips a month.
[QUOTE=ājw1981;691982ā]I started looking at my log since I began following this method since 17/03 in one of my accounts. Results are decent. Week 12 (16 - 20 March) Total Pips: 1636.7 Excluding Gold from the equation it was -134.3 Pips on Fx Week 13 (23 - 27 March) Total Pips: 641.6 Note: Did not trade on Friday, so thatās the tally for 4 days Observations: 1. Money management is of utmost importance with this system. Even a slip for a moment into getting greedy is enough to rip your account apart. 2. Trade only the first signal and add to positions only if you are already in the trade. Use proper MM rules while adding to position 3. Recovery trades happen, but it takes their own sweet time (refer to point 1 about MM in this regards) 4. I enter a trade on the closing price of the Candle that signaled the trade. Price usually hovers around this level for 2 - 5 candles before moving. 5. I strictly buy only when Stochs cross up from 20 and sell when Stochs cross down from 80. 6. Points 4 & 5 have been especially helpful in the recovery aspect 7. Last but not the least, I split my trades to book partial profits at 100%, 127.2%, 161.8% and let the final trade run its length Hope this helps others who are just getting started with this method.[/QUOTE]
Thanks for sharing the logs.
Could I know which pairs did you trade as well ?
I would like to see those results as well 
Just so Iām clear the RSI monthly was calculated using the 4 hour candles? Because that would be my explanation and I still think using a dashboard with the indicator will yield better results.
I already did the trades for EURUSD with monthly RSI so I would like to see the results of those back tests.
Thanks a lot for what you do Doughayman.
Regarding point 5: Does that mean you trade like PhilipPirrip or did you add an additional rule e.g. that the main line or the signal line or both have to come out again of the overbought/oversold level before you open a trade.
I am asking this because sometimes they move around quite a long time in oversold/overbought area so maybe it would be good to wait until the move out again of the 20 or 80 area.
Well, the good news (if itās good news) is that my 10-year data is crap. Lots of missing data all over the place, and 4-hour boundaries drift from 0,4,8,12,16,20 to 1,5,9,13,17,21 all over the place. The data cannot be trusted, so this means that the simulation I did for 10 years isnāt valid either. Scrap it. BTW, this 10-year data was from https://www.dukascopy.com/free/candelabrum/.
Yeah, my RSI Monthly (14) was ultimately calculated using 4-hour bars, which is absolutely what we want, since this is how they are going to be calculated in real-time (even with a Dashboard), so that should be the basis for back-testing too, IMO. I use the 4-bar Monthly Close for 13 prior months, and the current 4-bar close for open month. I checked it with an RSI Calculator that I built, and one that I found on the Internet. I believe it to be solid at this point.
I want to post results for the 5-year back-test that I did with 3 CCY pairs, where the data is good. Unfortunately, I stand corrected - the results were not that good with RSI threshold values of 33.33 and 66.67 (2 losing years, and overall not profitable, I think 41 trades; I think 53% winners). However, when I adjusted them to 30/70, the results were much better (only 1 losing year, but overall profitable, 32 trades; 65% winners), and when I upped it to 25/75, even better (no losing years, overall profitable, 21 trades; 67% winners)ā¦ It takes a while to sim this stuff, so let me know which one you want me to post. Data is for EURUSD, GBPUSD, and USDJPY, for 5 years, but no trades for 1st 14 months obviously, since RSI Monthly(14) has to be calculated first.
HELP: I need āgoodā 10-year, 4-hour bar data for the major 7 CCY pairs. This still needs to be validated, and if it passes, we can work on putting all the logic in a Dashboard. Donāt want to put the cart before the horse yet, but thatās just me.
Well these are very interesting findings you made. The use of 30-70 is quite common and could be used. But it has to be said this is the first time I hear about 75-25. Itās interesting nonetheless.
Iād prefer posting the 33.3 for now. However, if possible, Iād like you to keep score of the two other thresholds. That way if we do confirm that 75-25 is best then you have made a technical analysis discovery 
Just one not on the RSI I disagree. As I said earlier this would not be how the RSI was calculated in real time because we will only trade after a monthly RSI close above or below the threshold, and not a 4-hr close.
The second thing is as far as I remember we were only considering every x no. of 4 hour bars as a month. The real-time way of doing it would have calculated the open and close for the exacts months. Over time I expect them to show a significant difference.
This is actually why I want to see the 5-year test results for EURUSD because I have already done the back testing manually for this and we can compare. I think Iāve done it for GBPUSD as well but Iāll have to go back and see the post where I shared the results.
OK, then thatās a mistake on my part. I will need to revise the code to handle only the 14 last Monthly Closes on the 4-bar, and to IGNORE the current month. Let me revise this, and then Iāll post the 33.33/66.67 results on those 3 CCY pairs. Easy fix given the current implementation.
You probably missed one of my earlier posts during the week - I was considering doing every āxā bars as a month because it was easier to implement. But, I got a brainstorm on how to implement on the actual Monthly Closes. I just have to make the above change to make sure weāre in synch. Thanks for pointing that out.
OK Philip, easy change to facilitate using the last āClosedā 14-month RSIās for filtering. Here are the results. Please keep in mind that my data is only inclusive until Mid-March, so last couple weeks of data werenāt used. Monthly RSI Thresholds of 33.33 and 66.67 were used, per your request. Note: No trading 1st 14+ months of date interval, since we needed to obtain Monthly Closing values for Monthly RSI filtering. Again, for those of you following along at home, my tool, WealthLab, doesnāt support FX natively, so Iām assuming 10% of a $250,000 initial portfolio is applied to every trade.
IMPORTANT NOTES:
(1)This change to only use the last 14 Closing Month values, significantly increased the performance of the system !!! Yay!
(2) I also ran the test using 30/70, and the Net Profit was slightly less than the results below, and the Winning % was slightly less too. This certainly helps confirm your 33.33/66.67 hypothesis (although I still want to test with more data).
(3) I also ran the test using 25/75, and the Net Profit was about the same as the results below, but better than the 30/70. Of note here, 23 trades were taken and 18 were profitable (78.26%), so this had the best Win/Loss ratio of the 3 tests. Still, before we anoint this as a potential RSI filtering values method, I really want to see more data, across the major 7 CCY pairs.
Overall Performance:
Raw Return Graph:
Trades Taken:
There is one thing I wanted to point out Doughhayman.
I noticed that you usually focus on losing years in your analysis. While I agree with you that its important, I just wanted to say it is not as important as I feel you think it is. What I would be concerned with at the moment is % of winners and the win to lose ratio.
The reason for this is coming from a stockmarket background (and I think so do you) a losing year is important. If I am net loser on a year for a stockmarket Iād be devastated because it will actually have a long effect on my portfolio for the future (Iām more of a value investor.)
However in forex a losing year in EURUSD only means you lost net total on one stock, but we do not know yet how the portfolio has done. I could lose 1,000 pips for example on EURUSD but go on to make 15000 pips on AUDUSD, so Iām a net winner. Of course the opposite is true.
Itās only once we do the test on the majors (EURUSD, GBPUSD, USDJPY, EURGBP, USDCAD, AUDUSD, NZDUSD) and combine the results of a particular year that we can refer to that year as a loser.
Of course I understand we will also trade other pairs (my biggest win this year was on GBPCAD) and we may not take all the trade of one pair, but at the moment we are trying to value the performance of the system.
Thanks for that. The one thing I liked is that long and short system was higher than the rest. If you want to try the 30-70 and 75-25 be my guest.
I will look to translate the profits to pip terms, so that I can clarify to those watching how the performance would have translated in forex terms.
Great job Doughayman, I hope someone has those CCYs.
Philip, just so you know, any comments I made on a āLosing yearā was meant to be against Entire portfolio of trades across all instruments, so weāre in total agreement in that regard.
Philip, please re-read my Results post a couple of posts earlier. I made changes to account for 30/70 and 25/75 RSI filters. It looks like 25/75 has some merit, although this is still too small a data sample size, IMO. Gotta get more data to flush this out.
Thanks for clarifying that. I also agree with you on this being still a small sample size. I hope participants in the thread can direct you to useful CCYs.
In forex terms, these trades would have amounted to 2078 pips. This is an average of 90 pips expected on one trade. The average winner is 177 pips while the average loser is -109. This makes the ratio of winners to losers 1.6 (also known as risk to reward ratio.)
If you were to trade this using Kelly (more on the different money management formulas available to a forex trader after Doug is done with the testing) You will take an initial balance of $10,000 to $35,642. Your equity would have gone as high as $48,271 and you would have never been in drawdown (never less than 10,000.)
[B][I]Its important to note that as Doug provides more testing the value of the final balance will go down because the Kelly criterion will go down. This is just to give a feel for how the system would have done in forex terms until further testing, SO DONāT TAKE THESE RESULTS TOO SERIOUSLY JUST YET.[/I][/B]
I promised Iāll provide you with more details on money management. They will be the best you will ever read on the subject because I will discuss how to maximize gains in addition to the common āprotect your capital you know about.ā Only then will you get to choose a money management formula that suits your personality. I will also share Larry Williams most sought-after money management formula that made him take a 10,000 account to over 1 million in just one year and apply it to our results. This will only be done after we finishing the testing. Iām hoping for at least 100 back-tested trades but the optimal amount would be starting from 300 trades.
So start helping Doug with the CCYs so we can move with this thread forward. Exciting times ahead.
Well stated, Philip and thanks for all your valuable info. I am a total novice when it comes to FX (Iām an equities trader currently), so I need to bone up on FX, and money mgmt as it pertains to FXā¦Iām very excited about these prospects that you offer.
Additionally, once we can validate everything (and weāve already come a long way, after many refining iterations), we can each build our own Dashboards to assist with pattern recognition, and to facilitate trading. I have a little bit of a grandiose Dashboard plan in this regard, that will include the following:
-
Real-time (4-hour bar) alerts that provide each of the various 3-prong Entry points for a CCY pair;
-
Real-time alerts that provide each of the various 3-prong Exit points for a CCY pair;
-
Real-time alerts for stopped out positions;
-
Recommended Trades along with associated Fib Extension values;
-
Logging of all of the above data to a database for historical access;
-
Pushing all of the above pertinent data to a web page (refreshed every 4 hours), with use of color-coded indicators to
facilitate a quick view into pertinent info. I own many domains (part of another biz I am involved with), but I actually host
one of my domains in my home office, so pushing info onto that would be achievable;
Realistically, itās gonna take a couple of months to achieve all of the above, but patience, diligence, and perseverance conquers all, and Iām quite psyched about the long-term prospects.
So, as Philip stated, great things ahead. Letās get some data (the more the merrier). [B] Ideally, 10 years worth of 4-hour bar data for the major 7 CCY pairs would be awesome as a start. More CCY pairs, maybe those that you may be particularly interested in trading, would be great as well.[/B]
Hi Philip,
May I know why you didnāt enter any trade last week? Because based on your system I understand, buying signal appear for those six pairs I bought.its that all faults signal?(because all trades are now in the losing end, and one of the pairs EMa had crossed to the opposite), should close those pairs when the OS go to OB and crossed back?
This chart should help answer your question visually.
There are 2 entries. The first one (Long) was a failed trade, but price recovered (Even giving a few pips in profit) and then a short trade triggered. The vertical lines + down arrow indicates the price at which I enter.
Have you tried this from forextester?
http://www.forextester.com/data/datasources?utm_expid=10288340-77.M19gVxG6R3Czl5ePPSuSjg.0&utm_referrer=http%3A%2F%2Fwww.forextester.com%2FHistoryServiceDescription
Hi Philip,
First and foremost , my wholehearted THANKS to You ,for sharing such a system which is very simple in its nature and understanding and brought out by you, to help others.I am newbie in Forex markets. Eventhough I have more than 25years experience in TA, I have not come across a system like this .āHATS OFFā to you.
It is my goodluck i have come across this strategy of yours. just started on my demo account, hope to join you guys soon, on real accounts.
No I just didnāt enter because I was not at my computer when the signal gave way.
If you tell me the the pairs you entered I can tell you what Iād do.