Time diversification and trading currency pairs, USDJPY case

Trading currencies via online trading platforms has given unique opportunity to retail Forex traders to execute their orders with real time rates so fast and accurate. Even by applying programmed strategies it might be done faster than manual actions to click on sell/buy (soft) buttons. Technology has a key role to make a fast -actually very fast- environment for FX traders.

In this environment opening and closing positions quickly is a popular method which is known in extreme cases with minimum holding period as Scalping.

This little study is trying to understand if the holding period of a position may have any effect on the risk and return.  To dig about this an hourly data series of USDJPY rates from January 2002 to December 2011 (10 years) is prepared. Then USDJPY’s risk and return in 8 different portfolios are studied, while time horizons are including Hourly, 5 Hour, 10 Hour, 24 Hour (Daily), 72 Hour (3Day), 120 Hour (Weekly), 528 Hour (Monthly) and 1584 Hour (3 Month).

The idea for each portfolio is identical just holding period in each portfolio is different. In example; for Hourly portfolio USDJPY is sold then is bought in an hour (closing the position) and new position is taken till next hour…this process has ran for all 10 years (overlapping).

It should be mentioned that all positions are SELL USDJPY. It’s like a trader is getting in and out of the positions (Short positions) hourly (or 5H, 10H, 1D, 3D, 1W, 1M, 3M) for whole 10 years constantly (Overlapping).


To understand how much would be gained for an imaginary trader at the end of 10 years; cumulative results for each portfolio are reflected on below table.

 As it’s reflected in above table, longer holding period is bringing higher return. Actually the imaginary trader could make more than 8.5 million pips by keeping open Sell positions for 3 months (working days are considered). While keeping USDJPY sell position for an hour could bring just 5 thousand pips!

But it is not all! What would happen for the risk and also risk to return ratio? Does it really worth to keep USDJPY positions for longer period of time?!


First , its needed to see how much return each portfolio has made in average. Due to this target, avergae profit/loss of each portfolio is considered during the ten years. As it’s expected longer horizion of time have had higher profit. In average, by increasing holding period profit is increasing step by step from hourly to 3 months portfolios.


To obatin the risk of the portfolios in this case, standard deviation of profits/losses might be known as the risk.

Similar to the return, by increasing the holding period risk is going higher and higher. The riskiest portfolio belongs to the portfolio with longest holding period (3 months) while, minimum observed risk belongs to the portfolio with shortest holding period (Hourly). This fact brings the considerartion about risk to reward ratio. Does it worth to accept higher risk for more possible profit?


Although by longger time horizon risk and return both increasing in USDJPY case but acutally the return is increasing faster. By compraing risk to return ratios it can be seen that from hourly to 3 months portfolios the reward of each unit of risk is rising.  

In other words, if a trader considers just higher profit for trading USDJPY, longger holding period might be a better option. But if the priority is based on the risk aversion, shorter holding period can be recomanded. In General, longger holding periods not only bring higher possible profits but also the risk to reward ratios indicating that longer horizion may worth more than short term!

2013 Update

Similar study is done for USDJPY during first two months of 2013 (January and Feburary), with hourly to 5 days holding periods. But this time buy position is considered because USDJPY was rising on that time.  Results are reflected in belwo table which are similar with findings about 2002 to 2012.

Views: 510

Comment by talisman on May 29, 2013 at 10:01pm

it is very interesting, not 100 percent sure i understand totally,  but in my my mind you cant always make straight line comparisons.

can you explain better how opportunities plays a role in your analysis?  peter is correct in always saying that the maximum pips to be made is by trading shorter time frames because opportunities to enter increase.  capitalizing on those opportunities becomes arguably more difficult exponentially due to increased variables to allow for ( imo ), but thats a whole other discussion. thx, really enjoy the effort and energy put into your posts.

Comment by Romano on May 29, 2013 at 11:12pm

Hello Ali, I must agree with Peter that this study is flawed but for different reasons.

You cant know for sure what will price do, for example it could whipsaw for many months on even large TF and then results would be much worse. Your data are basically linear, note that 1h-4h-daily etc... number of pips raise almost linearly - thats because UY anyway most of the time went down and u have today that information but u cant know future price action.


If anything, study prove that shorter TF exhibit better RR which is far more important than number of pips made. But even that is flawed because it depend on strategy what your RR will be.

Today for example I "scalped" on 4h TF and made around 130pips on 2 pairs, but my SL even for such TFs was no more than 20 pips! In fact even that was quite margin I left, price anyway didnt go against me more than 6 pips(and I expected that)! Thats because even on such TF, with right setup my strategy allow me to execute such low risk high reward trade.

But here is an a answer to your RR study: market is fractal

best regards

Comment by Ali Karbalaee on May 31, 2013 at 1:17pm

@ Peter , it should be considered that the cumulative result is 8 mln pips , not the average! its like selling USDJPY in the described order and then counting how much pip is left at the end of ten year period. Its so simple and and can be applied if there is any doubt about!

also regard holding the position for long time like 3 months, u should understand that its a study and the points is about comparing short time vs long term in this aspect. Even in retail fx it would be possible to keep ur position for long time when ur acc is active and u r constantly taking the new positions.

Im putting two graph which may help.

in above picture re is representing the return. In example, USDJPY_1H is just price and USDJPY_re_1H , is hourly return during the ten years.also USDJPY_120H is representing 5 day return , step by step.

now the cumulative returns in below chart:

consider, we are getting the position, then holding the position for 3 months and then closing it. but since opening the first one in first hour , the next positions are getting opened every hour. 

** Y axis is the return and its negative because USDJPY has been failing



Comment by Ali Karbalaee on May 31, 2013 at 7:45pm

@ Talisman , I hope I've got ur point right (if not pls tell me ) but if u mean how we recognize to open the positions here: 

The assumption is USDJPY has sold every hour. the point is to understand if long horizon may bring more profit or short horizon and which is riskier. So it is like start trading the currency pair with "0.1 lot or 10,000 K volume" in 2002 and ending up at 2012. the way we sell is easy , we just sell every hour but the difference between portfolios is about the holding period. for H1 portfolio , the holding periods are just one hour, but for 24 hour (Daily) portfolios, the holding periods are 24 hours and for monthly portfolios the holding periods is as long as a month and ... which means the positions would be closed at the end of holding periods (by taking the reverse position).

whats is described above is a way to study about a fact not a way to trade. otherwise who could say USDJPY is decreasing for 10 years like that to just take sell positions for whole 10 years ... 

Even it should be added that positions can be Buy, then in case of USDJPY we would consider short or long horizon which may have less loss.

What I have posted here is a part of a work that I have done on 4 major pairs which includes some more tests like pitfall risk, value at risk, maximum draw down and ...during 2002-12 .

Enjoy the weekend



Comment by Ali Karbalaee on May 31, 2013 at 8:03pm

@ Romano

I agree with u generally. seems there is a misunderstanding,  this blog post shouldnt be over estimated. As I already wrote for Talisman its a simple way to study a fact. Now I should add that a way to study a fact in general. Thats why I used specific words like "Can" , "may" and etc to describe the findings in the text. As u already mentioned , depends to the strategy -or actually , the methodology to open the position, statistical characteristics of the currency pair and methodology to close the positions- longer or shorter horizon for the holding period might be advised.

Whatever I post here , is just a study in a specific approaches that I describe and it doesn't mean that its the only fact existing there. Results may differ by changing the parameter s and elements.

Have a great weekend.   

Comment by Ali Karbalaee on May 31, 2013 at 8:27pm

@ Peter, dont worry about, Im here. If there is any mistake I would be responsible for. As u see my acc is with my full name with direct connection to my linkedin acc, so in few secs even my phone number can be found to claim about what I post here....lol 

and if you would like to know who has done it, its simple there is smt with the name of Google u can easily search the keywords and find how many studies have been done about this subject. To help u I can advise using "scholar.google.com" as well ;) 

Comment by Ali Karbalaee on June 7, 2013 at 3:10pm

@Peter, tnx for ur time and explaining your doubt in detail.

It not double counted. It might be needed to explain that In 3-month portfolio we are selling every one hour but we would close the positions after 3months. In this way we are having a new position every hour which would be closed three month later after opening.

The total return (8 mln pip) is like :

cumulative 10 year Return (Total)=Return1 + Return2 + Retun3 + Retun4 + …

And each Return in 3month portfolio is calculated like below:

Return1= H1585 (3month) - H1

Return2=H1586 - H2

Return3=H1587 - H3

H4=H1598 - H4

It’s the reason that we call it overlapping. However non-overlapping consideration might be good but by studying portfolios in non overlapped method we would miss so many spaces during the holding periods while we cannot be sure when our imaginary trader could start trading exactly. Moreover our data series length would be shorter on that way (less reliability).

It should be repeated that it’s an approach to make sure statistically and logically the results are reliable (in general). Otherwise in practice selling/buying a currency pair like this is not doable, needed so much money, a really good broker to execute all orders at the right time for ten years and so many other things. Don’t look at as a trading strategy. It just demonstrates that holding currency positions for longer period of time may have higher return (loss) and also higher risk. Although risk to reward ratio is decreasing in time horizon.

Let me know your idea about.

Srr for late reply, have had busy days


Comment by Ali Karbalaee on June 7, 2013 at 3:11pm

* Ops :

H4 = H1588 - H4


You need to be a member of Forex Social Network to add comments!


© 2018   Created by FXStreet.   Powered by

Badges  |  Report an Issue  |  Terms of Service


Live Video