FX Statistics | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Performance statistics 4 replies

Trader Statistics and Broker Statistics 0 replies

Salesmen, Statistics and Survivors 0 replies

Forex case studies and statistics.Where? 2 replies

Is there any news Statistics? 8 replies

Attachments: FX Statistics
Exit Attachments

FX Statistics

  • Poll

  • Does using statistical and probability studies provide an edge in trading ?
  •  
  •  
  •  
  •  
  •  
  • #1
  • First Post: Aug 30, 2009 1:04am Aug 30, 2009 1:04am
I've got an idea.......

Anyone who remembers I Love Lucy, the 1950's TV sitcom recalls the madness
that would follow when Lucy would utter those four little words.

Well, here's my idea:

I would like to begin a thread that should focus on Forex Statistics
and how they may be applied to one's trading in everyday use.

Now, I've searched through the archives here at Forex Factory and discovered
a number of threads started on this very subject with the best of intentions.
They included some remarkable contributions by some well respected members, both past and present.

Many of them led nowhere.

Other established threads delved into the subject of statistical trading and probability studies,
later to fizzle out, or just fade away into the archives, likely to remain unseen for ages.

Until now.

I don't wish to make this a repository "dump" for endless streams of numbers and averages,
nor a high level discussion on neural nets, binary analysis, or regression theory.
(WTF did he just say ?)

The idea is simple. Statistics and probabilities have a viable use within the realm of market speculation.
Know the odds. Know if they are favorable for a given situation. Know when to say no.

Also, unlike most active threads currently in circulation here at FF, I would propose
one thing that we don't see as often. Remember, there was mention of other threads,
many of which contain the results of thousands of man-hours of research already performed.

I would encourage the RE-POSTING of those efforts,
with proper credit given of course, here in this thread.

In that way, we will not make waste of the valuable time of many who have already done their homework.
A second chance, if you will, to find value in what was once of value to others and in my opinion,
still remains valuable. Hope you all like stats. I'll expand further as we go along.



Let's get started....
  • #2
  • Edited 1:56am Aug 30, 2009 1:24am | Edited 1:56am
Okay, a few more guidelines and suggestions for content.

All pairs, doesn't matter to me. We'll try to keep topic close to Forex though.
However, gold and oil (they trade as currencies too) and index related stats are welcome.

Talk about the Dow, if you please. There are over 100 years of stats there.
Some of us trade the e-minis and nowadays there's an undeniable link between
the forex and equity asset classes (yes, forex is considered an asset class these days).

Spreads too. Spread statistics are of great historic value as well.

ATR's definitely ! You probably have indicators that will quote the average true range
for your favorite pair on a popular time frame. Maybe that's a stat you watch closely.
However, the ATR of a lesser traded pair, or on a less popular TF might spark an idea.

Images would be excellent and strongly encouraged, but remember, it's about stats
and statistics are usually a big bunch of boring numbers. It doesn't mean it has to be boring
and I'll bet we could make it interesting, but be forewarned....it can be that way sometimes.
After all, a chart is just a bunch of price and time stats, graphically represented on a screen or page.

You might say we're all statistical traders in that respect.

Any good humor is also welcome. Some of you know how I throw a party.
  • #3
  • Edited 1:57am Aug 30, 2009 1:35am | Edited 1:57am
Odds and probability studies.

I want to be careful here.

I like to be inclusive and get everybody with something to contribute involved.
I could write a book filled with what I DON'T KNOW. Many here could be my co-authors.
There are some among us here who have vast knowledge of the study of probability theory
and could wax far too eloquently beyond my level of understanding and I salute your talents.

I would ask if you can to please contribute anything that might add to our discussion here,
while keeping in mind that we have limits and we all learn, absorb and apply what we know
at different levels of comprehension. Just because I don't understand it yet doesn't mean
that someone else won't find value in it. So, when in doubt, just post it out there.

It's just rocket science.
  • #4
  • Aug 30, 2009 2:15am Aug 30, 2009 2:15am
Patterns

Another probability study is that of patterns, or more to the point, the likelihood or probability
that a given "pattern" is as such and that said pattern will either successfully complete or fail.

You know about head and shoulders ? Do you really ? How often are H&S patterns
actually continuation patterns, as opposed to a significant top for a given time frame in a particular pair ?

Statistically ? Well, that's a good question. Narrow down the pair and the TF
and you'll have a stat that could assist you in your next trade.
Notice I haven't cited the particular statistical odds for a H&S pattern here,
because these are the sorts of things I'd wish to see interest in.

How about the stats of 3 bar reversals? What are the odds of bar #4 reversing course ?
Wouldn't it be cool to know ? Maybe we'll touch on that one too.
  • #5
  • Aug 30, 2009 2:27am Aug 30, 2009 2:27am
Retracement of daily ranges. This is one of the MOST RELIABLE stats I follow.

How often do the majors, by and large, retrace at least 23.6% of their daily range
either by day's end, or by sometime in the following session ? If you don't know this one,
it will surprise the pants off of some of you out there. How about the 38.2% ?

I want recalc some of my data and share this one with you personally.

It's a good one. It wil take a few hours to compile the work in the form I'd wish to present it.

I wish to give credit to member 'goodthings', who has been on a posting hiatus from FF
for the last few months (since February '09), but who did an incredible job back in 2007
with his thread entitled, "Statistical Probability Trading With Price Action".

His thread, for historical reference may be accessed here:
http://www.forexfactory.com/showthread.php?t=27624

That data is outdated, but, I would wish to resurrect some of his efforts
and pay homage to the work goodthings performed for us back then so that
some of us could perhaps benefit today from the groundwork he already labored over.

In other words, take what was given to us and build on it.
  • #6
  • Aug 30, 2009 2:44am Aug 30, 2009 2:44am
One more thing...

I will need to retire to my study and begin to compile what I have so that I might
share it with you all in a form that will hopefully be found useful, at least by some.

This takes time.

The process can be arduous. You have an idea. You explore the idea. You mine the data.
You check and confirm the data, You construct a display mechanism for the data.
Even then, to keep current, you must update the data every n intervals.

I would also kindly ask of the audience that if the idea of statistical analysis appeals to you,

please participate.

This can be a lot of work and when an idea comes to the surface, it sometimes requires a team
of traders to sort out the data and results, even in our modern age of high-end computer technology.

So, like many things in life, we often get what we give. As always, I'll do what I can.

Well, that's the end of the introductions.

Let the Stats begin ! :nerd:
  • #7
  • Aug 30, 2009 4:06am Aug 30, 2009 4:06am
EUR/USD

Average Daily ranges for 2009
Month by month

January:......261
February:.....206
March:.........217
April:...........181
May:...........172
June:..........194
July:...........140
*August:.....134

Average Daily range
*Overall for 2009: 188

* as of 08/28/20009


Okay, what this obviously shows is that the daily traded range
on the Euro pair has decreased significantly since the start of the year
and as expected, has fallen off sharply in the past couple of months
due to the summer time trading lull. We are well below the average.
  • #8
  • Aug 30, 2009 4:16am Aug 30, 2009 4:16am
Would pick this option:

YES, but I do not use statistics and/or probability studies in my trading for other reasons.
forget about tomorrow, just steal away into the night
  • #9
  • Aug 30, 2009 4:21am Aug 30, 2009 4:21am
And the AMR is only 49% for the current month as we are near its end. Would be lovely to see a good extension tommorow into months end
  • #10
  • Aug 30, 2009 4:35am Aug 30, 2009 4:35am
In the 171 days traded thus far in 2009, Sundays counted as Mondays,
it's a real even split on the up vs down days. It usually is unless a huge
undercurrent is trending the market strongly over a long period of time.

The count is 85 days positive, 86 days negative. Law of averages there.

Of all these days, using 5:00 PM Est as a separation point,
the average move from day to day is about 92 pips.

That is a constant whether a bullish day (93.48 pips)or a bearish day (90.47 pips).

It can be said that the average statistical net change from day to day
is equal to about half of the average daily traded range for the euro.

This number has also decreased lately compared with the volatility seen
at the beginning of the year, but only slightly, as the euro was averaging
over 100 pips net change per day compared to it's current ~92.
  • #11
  • Aug 30, 2009 4:43am Aug 30, 2009 4:43am
Quoting belekas
And the AMR is only 49% for the current month as we are near its end.
Would be lovely to see a good extension tommorow into months end

Yup. Monthy and Weekly ranges are severely lagging their historic averages as well.
That also backs up the notion that as soon as market participation levels return,
we've got a move in the works....more like fireworks I would suspect.
  • #12
  • Aug 30, 2009 4:47am Aug 30, 2009 4:47am
Quoting spekitox
Would pick this option:

YES, but I do not use statistics and/or probability studies in my trading for other reasons.

Penny for your thoughts.......

Do the use of stats conflict with your trading strategy and thereby reduce or affect your edge ?
  • #13
  • Aug 30, 2009 5:00am Aug 30, 2009 5:00am
Quoting toshi
EUR/USD

Average Daily ranges for 2009

Average Daily range
*Overall for 2009: 188

* as of 08/28/20009


Okay, what this obviously shows is that the daily traded range
on the Euro pair has decreased significantly since the start of the year
and as expected, has fallen off sharply in the past couple of months
due to the summer time trading lull. We are well below the average.
Hi toshi,

This looks like it could be a really interesting thread. Thanks for the initiative.

Could the decreasing range in E/U alluded to above be a function of the overall decreasing volume of trading in spot fx and fx futures? I read a few months ago that CME trading volume was down by about 30% in the last year. I have read elsewhere that spot fx trading volume has decreased in the last year, as well.

I have been interested in using statistical analysis in my own trading - have been playing with a spreadsheet that can be used to draw histograms of recurring price (relative to volume) in a given time period (ala Market Profile) but also allows all typical stat function analysis on the imported data. Still playing with it, trying to make it user-friendly. Maybe I will give it away here when its done.
"If The Fool persists in his Folly he will become wise." - William Blake
  • #14
  • Aug 30, 2009 5:01am Aug 30, 2009 5:01am
Quoting toshi
EUR/USD

Average Daily ranges for 2009
Month by month

January:......261
February:.....206
March:.........217
April:...........181
May:...........172
June:..........194...
it would be interesting to compare those values from previous years. Some of the volatility from early in the year could have been a result of the turmoil in the financial systems. The markets started going crazy around october of 08.
  • #15
  • Aug 30, 2009 5:14am Aug 30, 2009 5:14am
Linear time is also important in the world of statistical averages.

We all know that the candles with the greatest volume and widest ranges usually occur
during the time that London is active and also at the start of the US session.

Smallest range, choppy action is almost a given between the few hours when New York goes home
and Asia starts it's trading session. I believe Tokyo is 13 hrs ahead of NYC.

So time matters, unless your an existentialist who doesn't like money, or something.

For the laundry list, would be worthy of note to explore time of day statistics, such as:

Statistically, when does a pair hit it's high or low for the day ?
What hour or between what hours ?
Ask the question, "..If a pair has not returned to it's given high/low for the day
by say, 12:00 noon eastern, what are the odds that it will break said high or low ?"
The historical data could imply a popular kick off time for such action.

The start of the day (Sydney/Tokyo) and the London fixing appear to be
popular time windows for extremes, but what do the statistics reveal.

Talk about some Pivots...

On bullish (or bearish) days, what are the odds of hitting R1 (or S1) ? How about R2 (or S2)?
One might bet it happens fairly often, hmmm ? How often ? Numbers would speak louder.

Once ranges and volumes return to "normal" (we'll assume they will),
it could be good to know some of this.
  • #16
  • Aug 30, 2009 5:29am Aug 30, 2009 5:29am
[quote=toshi;3005352]Linear time is also important in the world of statistical averages.


So time matters, unless your an existentialist who doesn't like money, or something.

For the laundry list, would be worthy of note to explore time of day statistics, such as:

Statistically, when does a pair hit it's high or low for the day ?
What hour or between what hours ?
Ask the question, "..If a pair has not returned to it's given high/low for the day
by say, 12:00 noon eastern, what are the odds that it will break said high or low ?"

[quote]

Indeed. One that I am trying to explore is, "During what period is a given range most often & most strictly adhered to?" The answer is hinted at on the price/time histogram - but I just need to spend a little time to flush out some more detailed results from the data. Why bother? Well, I believe that there are ranging chunks of the PA that can be exploited for very high probability, low gain trades. Varies per pair, some of coures are more ranging than others, but ranging periods appear reliably across the board, typically in thin markets...

See Nanningbob's thread on the AUD/NZD. (Too bad retail spread is so high.) If you could identify particular ranging periods where price range was sufficient to reliably hit multiple 2-3 pip (for instance) winners, and you could automate that trading process....well, obviously, let the good times roll.
"If The Fool persists in his Folly he will become wise." - William Blake
  • #17
  • Aug 30, 2009 5:41am Aug 30, 2009 5:41am
Quoting magnumfreak
it would be interesting to compare those values from previous years.
Some of the volatility from early in the year could have been a result of the turmoil in the financial systems.
The markets started going crazy around october of 08.

I'm certain that's at least part of the reason, some residual effect of last year's meltdown,
coupled with tax selling both before and after the year end, spiting PNL over different years,
and asset reallocation.

Here is where Seasonal statistical analysis comes into play.

Good old Jake Bernstein was one of the better cycle trader/analysts around
back in the 80's and early 90's. Although his work mostly focused on commodity cycles,
he wrote a great deal on the cyclical trends of the British Pound in much of his earlier work.
Claimed it was more predictable from cycle to cycle. There was no euro at the time.

So, comparing Q1 of '09 against say Q1 of 2000 through 2008 would be a good place to start.
One difference however, is that the volume, ranges and market environments
were different then, so some computational scaling might be in order.

Percentages of daily ranges as opposed to raw numbers for comparison could paint a more accurate picture.

But that's a good observation. Thank you.
  • #18
  • Aug 30, 2009 5:49am Aug 30, 2009 5:49am
Quoting The Fool
Hi toshi,

This looks like it could be a really interesting thread. Thanks for the initiative.
Thank you.

Quoting The Fool
I have been interested in using statistical analysis in my own trading -
have been playing with a spreadsheet that can be used to draw histograms
of recurring price (relative to volume).....

Maybe I will give it away here when its done.
See, that's what I'm talking about. Thank you again, Fool.
That would be great. I hope some of the folks out there that
live and breathe this stuff might pop in from time to time and
add a little here and there.

Much of the work has already been done, kind of like the C-Seti groups
with this vast amount of data at their disposal, but the need for
interpreters and analyst to find E.T in all that static. It's a job.

I'll see how it goes. Thanks again, man.
  • #19
  • Aug 30, 2009 5:56am Aug 30, 2009 5:56am
Quoting The Fool

See Nanningbob's thread on the AUD/NZD. (Too bad retail spread is so high.) If you could identify particular ranging periods where price range was sufficient to reliably hit multiple 2-3 pip (for instance) winners, and you could automate that trading process....well, obviously, let the good times roll.
Know what ? The Euro/CHF has been a lot like this in the past. And the spread ain't so bad.
I've made some huge spot trades on this pair in the past, just trying to squeeze a pip or 2 out
and as it usually only moved a little, and mostly during London, I knew when I could literally rob the bank,
and when to stay in the gettaway car. Might check that one too.
  • #20
  • Aug 30, 2009 7:44am Aug 30, 2009 7:44am
I don't use stats to anticipate price movements but I admit they can be profitable. For example, statistically it's not a surprise if the pound breaks on mondays before or around the London open.

Many people trade box breakouts profitably, it's also based on stats/probabilities if I'm not mistaken.

The reason I don't use them is because I'm way too lazy to sit down, sieve thru millions of rows of price data to find some correlation (which might stop working next week).

And there's no way I'd put on a trade based on something I'm not sure of. "I only believe in statistics that I doctored myself"

Quoting toshi
Penny for your thoughts.......

Do the use of stats conflict with your trading strategy and thereby reduce or affect your edge ?
forget about tomorrow, just steal away into the night
0 traders viewing now