Comment by Peter jcp on May 24, 2013 at 9:51pm

Interesting studies but please remember averages are so misleading - especially when applied to the average movement on a 4 hr candle. As an alternative example the average wage / salary in the UK is approx £27k - spread over 30+ million workers - However the real fact are 70% of workers earn a lot less than the average and the 5% who earn over £100k pa plus push it up - and you end up with a false mean.

The industry want traders to concentrate on 4 hr candles - simply because they are not the efficient and you will end up with average results  - ie Investor type pa - rather then letting all know the proper method to increase their capital safely and also quickly. 

I am now pleased it is like that - but I was not when I spent my first 2 yrs doing what the industry wanted me to.- thank goodness I persevered and found out the hard way - but it was worth it in the end ;-))

Regards 

Peter

Comment by Romano on May 24, 2013 at 10:00pm

Hello Ali, first I would like to say that I am glad to see someone from your country here. Now to your studies:

The reason why it doesnt work is because FX market is not equal to stocks. There is no "reverse to the mean"(RTM) guaranteed. Example is USD, since `70 it lost more than 90% value. This is both because being reserve currency and also continuing monetary base expansion. With stocks, you factor company earnings and other things and so you can calculate deviation and thus RTM, but currencies are different - even if they do reverse sometimes.

There was a very good article on snbchf.com about it, site is written by independent group of swiss economists and materials there are great. I remember in one of them being this explained.

So, RTM will not work well in currency market IMO and there are also other factors, like manipulation, squeezing etc...

regards

Comment by Ali Karbalaee on May 25, 2013 at 7:11pm

@ Peter ...thank you man for attention and sharing ur idea

U r right mean or average might be misleading sometimes. thats why financial researches normally compare different approaches to make sure about the findings...although here I just had a brain storm to see if a passive buy sell strategy can be fruitful

regard the time frame, the data series that I have gathered is quite long. including four majors pairs in hourly, 4 hour and daily time frames in 10 years. To apply some tests i have to run the data series in software packages like excel or gretl which takes bit time for my computer to process all the steps. so some times I just go for 4 hour to not spend too much time waiting for the process to be done .

Do you think there would be huge difference in currency pairs behavior in hourly and 4 hour time frames?

cheers

Comment by Ali Karbalaee on May 25, 2013 at 7:30pm

@ Romano ...thank you for leaving your comment.

thank you for introducing that article, should read it! just found it ... FX Theory: The Real Mean Reversion, right? will study it. but even before reading it i should say that I totally agree with the idea that mean reversion is not guaranteed in currency pairs. the tests i have done for my master thesis shows that among 4 major currencies just USDCHF and USDJPY may have this characteristic (http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2246076)...even in stocks its not guaranteed , its may appear and disappear in different market cycles

ciao

Comment by Romano on May 25, 2013 at 10:05pm

There is another trader here called Brian Twomey, he`s also a professor of university and some months ago he applied similar style based on deviation among other things. You may contact him for further discussion(or read his posts). You both seem to trying to figure out mean reversion. 

Best regards

Comment by Brian Twomey on May 26, 2013 at 3:26pm

 Ali, This was a nice attempt but finding a balanced curve on the 4 hr says nothing to exactly where prices are trading on an overall basis. And then taking the absolute value of an average gives you the same number in a different form. What good is that. If the average itself fails you on the 4 Hr, any transformations will be equally disastrous since both are the same numbers. Balanced curves and true averages are found within historical prices not a chart as short as a 4 hr.

 And then I would've eliminated USD/JPY in my study because its a pair that moves as it feels free to move which may not be market related as is the present case from the 81 low in Oct to 103 today. I would've included maybe USD/CAD since its a market related document that responds immediately to statistics where USD/JPY does not on all occasions. Further since EUR/JPY has such a low correlation (.20) to USD/CHF, possibly that would've been a pair for consideration within the matrix so then you have 2 equal pairs, 2 opposites and a cross. Your answers are not found in averages alone rather they are found in Errors. But not all error terms work equally. This point I found to be the greatest misnomer among the gazillions of academic papers I've read over many, many years going back to the 80's to this modern day. Time series papers always seem to miss their mark because the error terms are off therefore your parameter estimates will be far off. Because we are dealing with currency prices using the smallest numbers wrapped in a decimal point, you must be very careful what and how you apply any statistic since not all work equally and some reveal absolutely nothing, CV in itself is one such example. And last, before any study, you must first have your research questions firmly evaluated to know which statistic to apply. Hope I helped you get going Ali

Comment by Ali Karbalaee on May 26, 2013 at 8:13pm

@Romano ....tnx so much. actually i knew him since couple of months ago but when i moved back home l lost the access to my acc and during last few months i was just following silently:D
this post also belongs to sep 2012 that just posted again after deleting my lost acc . luckily Brian has already put his comment.however i should reply his interesting comment tomorrow.

cheers

Comment by Peter jcp on May 26, 2013 at 9:47pm

Hi Ali - re is there a huge difference in currency pairs behavior  between the 1 hr and 4 hr time frames ? Well my answer would be yes and I will explain why.

Most traders trade to make money - ie that is the main objective - and ideally you do not want to risk a lot of your capital - but you also want fast large growth - ie a retail trader even full time with a $50k live account does not want to end up with just 35% per annum - but instead really would prefer 200+ % per annum - I think most would agree with me.

Taking this line further - Retail trading is a different ball game to Bank or commercial trading - ie Retailers cannot lead they have to follow and the fact that they are not using 100's of millions of capital - they cannot afford to live on just investor returns.

So they need to trade differently.

Personally for me a 1 hr time frame is OK - and is far superior to a 4 hr time frame.

Only this week in another post we mentioned that the traders who look say daily + charts and never under 4 hrs - miss all the action - ie they see a 4 hr candle of say 80 pips in total.

If they were to drill down to an intraday time frame of 5 or 15 minutes - they would then see that during that 4 hrs of movement - we had a 45 pip rise in the first hr - then a 30 pip drop in hr 2 and then in hr 3  - a  50 pip rise and then in the final hr price moved up just 15 pips - before slightly dropping just 6 pips or so.

Therefore the 4 hr candle shows a 80 pip rise - but in reality - 4 main moves totaling 45 + 30 + 50  + 15  = ie 140 pips - ie nearly double . With the 1 hr moves you will not need so large a stops and you will bag more pips - ( only when you are good though - and that's half the problem - 90% of traders never get there)

So more pips - ie more money - smaller stops - less risk - win ratios - still higher that the investor mobs who think 50% is really good - rubbish its bad and really the same as tossing a coin;-))

I hope you get my point - and if all your studies are for the commercial world - fine - but if they are for the retail traders World - start again - with totally different parameters and objectives.

Regards 

Peter

Comment by raj patel on May 27, 2013 at 8:58am

Hi Ali, step back and understand how these time frames are made. Weather it is 1 hour or 4 hours or 6 and 8 or 10 they all represent opinion of price. The entire basis of opinion is a black hole, so dynamic in its nature that even time cannot capture an explanation let alone price.The past is full off researchers like your self  who have but failed to provide a conclusive answer.that does not mean you cannot try. Cast your net into the unknown and you may see progress. This path you are leading will leave you disappointed.

Good Luck

Comment by Ali Karbalaee on May 27, 2013 at 6:16pm

@ Brian , tnx for ur feedback

Maybe it should be added that I wanted to show how passive strategies may bring considerable loss even if trader could be both seller and buyer at the same time. some years ago around the time NFP news was releasing the idea to get two sell and buy positions at same time was popular among traders community that I was working with. the believe was to target two different side of the market and executing two TP orders in approachable zones. Maybe it could be working for a specific period and special minutes but in general I had the idea to make a similar test in long horizon to show these tricky methods can be dangerous as well. 

the reason that I went for 4 majors and 4H is just about the data series that I had . u know better than me that such huge data series may take so much time for data cleaning and fixing the errors (10 years 4 majors including hourly to daily). thats i why I mostly use the data series that I have prepared and already spent so much time on it. meanwhile availability of the other pairs through the passed time is another issue that I normally have. 

Also regard the mean, I took it out from absolute values of returns  to have an estimation about potentiality of the every market move. Then I counted half of it to be in the safe side (by the idea of tricky trader , if u dont mind I call him/her like that). also by taking average of return instead of absolute values I just could have shorter range of every market move.

in case of CV, dont u think it can be a good approach (plus to some others) to compare risk to reward even in currency portfolios? 

Ciao  

 

Comment

• ‹ Previous
• 1
• 2
• Next ›
• Page