EURUSD session activity

[Liquid validity]

Been crunching in excel, I thought the U.S. session had seemed
more lively than the European.

Dataset 01.01.13 - present.

Edit - post updated with more charts

 

[Shakone]

12:30 looks larger than expected. Is that because of U.S. clock adjustments at different times to us?

 

[Liquid validity]

12:30 looks larger than expected. Is that because of U.S. clock adjustments at different times to us?

No all timestamps are GMT (bar start) including DST adjustments.
Eurozone traders readying for U.S. open.
Due to sample size, will probably be affected by a few
releases in that bracket - ECB min. bid rate for instance (not checked releases, just example).

**********
EDIT
All attached files link:
http://www.trade2win.com/boards/misc.php?do=showattachments&t=178822

 

[Purple Brain]

What is the timebase on the x-axis - BST, GMT or EST?

Is it statistically valid as there seems to be little more than 5 pips in it when you adjust for that outlier bar?

I've done quite a bit of this type of time based analysis looking for reversals rather than volatility and I've yet to find any statistically significant time based effects. Certainly the middle east start up each day has begun to produce relatively frequent reversals as obviously does the Europe/London wedge, but little else I've been able to ascertain.

 

[Liquid validity]

What is the timebase on the x-axis - BST, GMT or EST?

Is it statistically valid as there seems to be little more than 5 pips in it when you adjust for that outlier bar?

I've done quite a bit of this type of time based analysis looking for reversals rather than volatility and I've yet to find any statistically significant time based effects. Certainly the middle east start up each day has begun to produce relatively frequent reversals as obviously does the Europe/London wedge, but little else I've been able to ascertain.

GMT - DST adjustments included in timestamps.
No its largely just a snapshot of past behaviour this year.
True to question the statistical validity due to sample size.
Each time bar has a sample size of 174 - not enough to be valid really.
Just curiosity value mostly.

 

[Eurex]

Been crunching in excel, I thought the U.S. session had seemed
more lively than the European.

Dataset 01.01.13 - present.

I trade Euro Futures 6E and it gets lot trendier when NY opens around 1300 and volume is good till Europe close at 1630 GMT.

 

[random12345]

GMT - DST adjustments included in timestamps.
No its largely just a snapshot of past behaviour this year.
True to question the statistical validity due to sample size.
Each time bar has a sample size of 174 - not enough to be valid really.
Just curiosity value mostly.

Interesting, the average range is more consistent than I would have guessed, but the US session has been breaking most European ranges formed with notable consistency during August and then flying for 30-50 pts. So though the US gross movement may be basically the same, it tends to be more single directional in nature, which for me is more profitable.

 

[Liquid validity]

Interesting, the average range is more consistent than I would have guessed, but the US session has been breaking most European ranges formed with notable consistency during August and then flying for 30-50 pts. So though the US gross movement may be basically the same, it tends to be more single directional in nature, which for me is more profitable.

Agree.
I am going to expand on this, working on each year back to 2004.
Maybe include half and full hour brackets as well, and correlate
open / close of same period - O/C range broadly similar will point to
directional.

Ideally would prefer it to be tick rather than time series for obvious reasons.
Don't have enough tick data that far back, so time series it is, along with
the arbitrary element that brings.

 

[random12345]

Agree.
I am going to expand on this, working on each year back to 2004.
Maybe include half and full hour brackets as well, and correlate
open / close of same period - O/C range broadly similar will point to
directional.

Ideally would prefer it to be tick rather than time series for obvious reasons.
Don't have enough tick data that far back, so time series it is, along with
the arbitrary element that brings.

Nice work LV.

Certainly it used to be a more active pair, but it should be interesting to see whether it still exhibited similar behaviour, albeit over a larger range of points.

 

[Liquid validity]

Nice work LV.

Certainly it used to be a more active pair, but it should be interesting to see whether it still exhibited similar behaviour, albeit over a larger range of points.

Yeah I'll pop them up when I'm done, better yet, I can post up the excel
routine so you can run on anything you want yourself.
Wait a minute, I'll put up instructions in next post.
Its manual excel, cba to script it.
Pretty quick though, using sumif and data array.

 

[Liquid validity]

DUKAS TIMESTAMPS = START OF BAR -- GMT 00 -- LONDON
VERIFIED WITH LIVE MBT CHART COMPARISON

Dowload dataset in m15 / H1 / DAILY ask format for 1 year at a time:
http://www.dukascopy.com/swiss/english/marketwatch/historical/?ibref=2061

Open and save as excel workbook straight away as
EURUSD TF YYYY TIME OF DAY
Original downloaded file is.csv

TIMESTAMP FORMATTING
copy timestamp column to next cell

DATE FORMATTING
1st column find and replace _0* with nothing (removes time leaving date, _underscore=space)
1st column find and replace _1* with nothing (removes time leaving date, _underscore=space)
1st column find and replace _2* with nothing (removes time leaving date, _underscore=space)
***** SAVE *****

Date column left can't be date formatted
Format date column as text
Find replace . with /
***** SAVE *****

TIME FORMATTING
2nd column (copied original time & date column)
Find replace *2013 (YYYY)with nothing (removes date leaving time)
Do the same for every year (each year will be separate .csv anyway due to excel file limitations)
2nd column again, find replace .000 with nothing
***** SAVE *****

ADD DAY OF WEEK
Insert new column after date column
Select B2 and paste this formula into formula bar::
=TEXT((A2), "ddd")
Copy that cell
Paste into last cell of column
Back to B2 - select cell
CTRL+SHIFT+DOWNARROW
CTRL+D to auto fill whole column
***** SAVE *****

REMOVE SAT & SUN
Find and delete all sundays & saturdays
Home tab > conditional format > highlight cells > text containing > sun (red)
Home tab > conditional format > highlight cells > text containing > sat (red)
DATA tab > sort by day of week column - col B > sort by cell colour
My data has headers = tick box
Delete rows with red for sat / sun
***** SAVE *****

DELETE VOLUME COLUMN


===========================================
Once one sheet is complete:
Clone first completed sheet to create 2012-2004 sheets
then delete all price data leaving formulas and chart

Copy paste formatted data into clone of first sheet
select columns A-G, not entire sheet
===========================================


Once one sheet is complete the steps below can be skipped
===========================================

ADD High / Low RANGE FOR EACH OHLC DATAPOINT
Select H2 and paste this formula into formula bar:
=(E2-F2)*10000 -- gives points range for that OHLC datapoint
Copy that cell
Paste into last cell of column
Back to B2 - select cell
CTRL+SHIFT+DOWNARROW
CTRL+D to auto fill whole column
***** SAVE *****

ADD Open / Close RANGE FOR EACH OHLC DATAPOINT
Select I2 and paste this formula into formula bar:
=ABS(D2-G2)*10000 -- gives absolute points range for that OHLC datapoint
Copy that cell
Paste into last cell of column
Back to B2 - select cell
CTRL+SHIFT+DOWNARROW
CTRL+D to auto fill whole column
***** SAVE *****


**** TIME OF DAY ANALYSIS ****
Sort by column C - time - sort by values A-Z

TIME WINDOWS
Remove times you don't need:
Delete sheet rows = midnight to 06:30 inclusive
Delete sheet rows = 20:00 to midnight inclusive
Copy column C - Time to column K
Data tab - remove duplicates in column K
***** SAVE *****

TOTAL FOR EACH TIME BRACKET
Use coumn L for total of each time bracket
Enter this formula into cell L2
Key for columns:
C = Time
H = Hi/Lo range
K = Individual time brackets - 1 cell per value

Select L2 and paste this formula into formula bar:
=SUMIF(C2:C30000,K2,H2:H30000)/COUNTIF(C2:C30000,K2)

Copy L2 cell and paste into cell below last cell in column (row 54)
Select L2 cell, then select whole column:
CTRL+SHIFT+DOWNARROW
Fill all cells in column with formula:
CTRL+D
***** SAVE *****

Create bar chart by selecting both columns K&L as dataset
===========================================

 

[Liquid validity]

Those are the instructions for the data source I used, its OK for this type of stuff.
Anything that requires better resolution obviously needs quality tick data.

Most of it will apply to other data formats, just the date and time
formatting routine will be different.
No skin off my nose with this anyway.
How people act on it, if at all, will vary anyway.
Anyone who does use it won't have much impact anyway.

 

[wackypete2]

I can tell you from direct trading experience the range of movements in fx during the US session has been dismal for the past 2 years. What LV's chart doesn't show is High/Low range for the session. If the average candle is 10 pips but they are all 10 pips up, 10 pips down then it's untradeable. The UK session tends to trend much better.

I trade Euro Futures 6E and it gets lot trendier when NY opens around 1300 and volume is good till Europe close at 1630 GMT.

I concur. After 11:30am NY time fx dies a horrible death on most days.

Peter

 

[Liquid validity]

I can tell you from direct trading experience the range of movements in fx during the US session has been dismal for the past 2 years. What LV's chart doesn't show is High/Low range for the session. If the average candle is 10 pips but they are all 10 pips up, 10 pips down then it's untradeable. The UK session tends to trend much better.

Fair point and true, I want to expand the time brackets for that as well.
Putting it into context, 2013 has given me most of the gains from
exits close to rollover, whilst entries were from Euro session.
So in that sense, my view of the U.S. session won't be the same as yours.

 

[random12345]

I can tell you from direct trading experience the range of movements in fx during the US session has been dismal for the past 2 years. What LV's chart doesn't show is High/Low range for the session. If the average candle is 10 pips but they are all 10 pips up, 10 pips down then it's untradeable. The UK session tends to trend much better.



I concur. After 11:30am NY time fx dies a horrible death on most days.

Peter

The ranging of the latter US session can however, be so predictable as not to be a problem. The ranging of the European session is for me, more difficult to accommodate when it occurs. Often I will 'make bank' when New York opens and breaks the high or low of the Frankfurt + London range in those circumstances. This has been especially true during summer months.

 

[random12345]

Though if you depend solely on the US session to put steak in the fridge then yeah... that might not be enjoyable.

 

[Purple Brain]

I can tell you from direct trading experience the range of movements in fx during the US session has been dismal for the past 2 years. What LV's chart doesn't show is High/Low range for the session. If the average candle is 10 pips but they are all 10 pips up, 10 pips down then it's untradeable. The UK session tends to trend much better.



I concur. After 11:30am NY time fx dies a horrible death on most days.

Peter

Is this because the bulk of FX trading which has been traditionally centered primarily around the London session has moved even more strongly into that time arc? If so, why is even less FX volume traded in the US these days than before? Over-regulation (aka Dodd-Frank)?

 

[random12345]

Is this because the bulk of FX trading which has been traditionally centered primarily around the London session has moved even more strongly into that time arc? If so, why is even less FX volume traded in the US these days than before? Over-regulation (aka Dodd-Frank)?

Volatility drops are debated again and again and normally when they peak with some kind of consensus for the reason, the vix or whatever explodes North.

I'm a firm believer that any kind of predictability, volatile or not, is beneficial and though boring, it has been predictable lately.

For me, I haven't seen a day outside of predicted gross movement since 31/7 on the U/J, which was FOMC misinterpretation. I lost money that day.

 

[Liquid validity]

EURUSD H1, plotted the open close as well this time.
Relationship pretty similar with m15.

 

[Liquid validity]

I'm doing this from 2013 back to 2004.
I'll chuck the lot up if there is interest.
Any requests, shout out now - EURUSD only for time being though
as thats my primary interest