Statistical Probability Trading With Price Action | Page 26 | Forex Factory

Options

Go to
Permalinks
Print Thread

Related Instruments

GBP/USD

Similar Threads

GbpUsd Statistical Probability With Price Action 109 replies

EurJpy Statistical Probability Trading With Price Action 46 replies

GbpJpy Statistical Probability With Price Action 22 replies

Statistical Validation on trading 24 replies

Attachments: Statistical Probability Trading With Price Action
Exit Attachments

Statistical Probability Trading With Price Action

  • #501
  • Aug 12, 2007 11:39am Aug 12, 2007 11:39am
UsdJpy DAR
Attached Image (click to enlarge)
Click to Enlarge Name: 32 UsdJpy Daily Range Average.jpg Size: 17 KB
  • #502
  • Aug 12, 2007 11:41am Aug 12, 2007 11:41am
UsdJpy DRM the pink line is the DAR for 2007 and the blue line is the RM for the past 20 days in relation to the 2007 DAR
Attached Image (click to enlarge)
Click to Enlarge Name: 33 UsdJpy Daily Range Chart.jpg Size: 26 KB
  • #503
  • Aug 12, 2007 11:42am Aug 12, 2007 11:42am
UsdJpy WAR
Attached Image (click to enlarge)
Click to Enlarge Name: 34 UsdJpyWeekly Range.jpg Size: 20 KB
  • #504
  • Aug 12, 2007 11:44am Aug 12, 2007 11:44am
The pink line is the 2007 WAR and the blue line is the WRM of motion for all the weeks of the year.
Attached Image (click to enlarge)
Click to Enlarge Name: 35 UsdJpyWeekly Range Chart.jpg Size: 21 KB
  • #505
  • Aug 12, 2007 12:09pm Aug 12, 2007 12:09pm
Another look at WAR for 2007 and comparing it with the WRM for the year. The weekly averaged changes from week to week. It's based on adding each weeks RM together and dividing it my the number of weeks together. For there have been 32 trading weeks this year. So all the previous weeks individual RM was added together and divided by 32. The pink line shows how the 2007 WAR peaked after 11 weeks and has gradually declined.

GbpUsd Posts 505-509 pg 34
GbpJpy Posts 510-514 pg 35
EurJpy Posts 518-522 pg 35
UsdJpy Posst 524-529 pg 35
Attached Image (click to enlarge)
Click to Enlarge Name: 35 UsdJpyWeekly Range Chart.jpg Size: 26 KB
  • #506
  • Aug 13, 2007 5:50am Aug 13, 2007 5:50am
FWIW, I am long at 2.0125, the bounce off of 100 day .5 fib. TP 2.0185, SL 2.0094. Pippo
Vincit qui se vincit.
  • #507
  • Aug 13, 2007 6:55am Aug 13, 2007 6:55am
Have closed this trade at 2.0144. Did not like price action at 2.015, The 20 day and 5 day low, which had been broken and now appears to be acting as resistance.
Vincit qui se vincit.
  • #508
  • Aug 13, 2007 2:38pm Aug 13, 2007 2:38pm
For those of you that haven't been following the thread,
I was inspired by goodthing's work and decided to do my own stats for GBP USD using 8 years of historical data.

Here are some results (finally) based on weekly highs and lows:

In the past 8 years consisting of 410, 5 day trading weeks
GBP USD has retraced to

61.8% : 254 times, for a 62% probability retracement to this level WITH respect to Bullish/Bearish movement during the same week or the following week

50% : 292 times, for a 71% probability

38.2% : 335 times, for an 82% probability

23.6 : 378 times, for a 92% probability

For only 32 weeks out of 410 weeks has GBP USD not retraced to any fib level during the same or following week and simply continued trending towards an extension for an 8% probability OR just traded in a very narrow range not going up or down..

I will have more interesting things to know about the last 8 years soon.

Some notes regarding probability:

Probability ONLY means that an event is likely to occur, it does not necessarily mean that if there is a 92% probability of something happening then then it's sure as hell gonna happen as opposed to an event with 8% probability, but that is the most likely event based on past data.

GOODTHINGS: Thank you for the eye-opening experience in this thread.

P.S I'd initially tried to apply logic for retracement before the week's end in Excel but it was waaaay too much headbanging and seems irrelevant to trading realistically, but if someone is an Excel guru reading this I'd appreciate some help to do it anyway.
  • #509
  • Aug 13, 2007 3:01pm Aug 13, 2007 3:01pm
Quoting bobblong
For those of you that haven't been following the thread,
I was inspired by goodthing's work and decided to do my own stats for GBP USD using 8 years of historical data.

Here are some results (finally) based on weekly highs and lows:

In the past 8 years consisting of 410, 5 day trading weeks
GBP USD has retraced to

61.8% : 254 times, for a 62% probability retracement to this level WITH respect to Bullish/Bearish movement during the same week or the following week

50% : 292 times, for a 71% probability

38.2% : 335 times, for an 82% probability

23.6 : 378 times, for a 92% probability

For only 32 weeks out of 410 weeks has GBP USD not retraced to any fib level during the same or following week and simply continued trending towards an extension for an 8% probability OR just traded in a very narrow range not going up or down..

I will have more interesting things to know about the last 8 years soon.

Some notes regarding probability:

Probability ONLY means that an event is likely to occur, it does not necessarily mean that if there is a 92% probability of something happening then then it's sure as hell gonna happen as opposed to an event with 8% probability, but that is the most likely event based on past data.

GOODTHINGS: Thank you for the eye-opening experience in this thread.

P.S I'd initially tried to apply logic for retracement before the week's end in Excel but it was waaaay too much headbanging and seems irrelevant to trading realistically, but if someone is an Excel guru reading this I'd appreciate some help to do it anyway.
That is quick work on the 8 years looking at the attachment below it seems consistant over the year. They say the Forex movement is based on human emotion and human emotion has been consistently erratic with some what of a pattern. This thread may help to develop some type of system. Go to this link then post 523 then to the link in the post.

http://www.forexfactory.com/showthre...=27624&page=35

It sounds like Jr97 knows excel real well.
  • #510
  • Aug 13, 2007 3:09pm Aug 13, 2007 3:09pm
Daily support at the 89 ema.
Attached Image
  • #511
  • Aug 13, 2007 3:11pm Aug 13, 2007 3:11pm
Weekly support at the 21 ema. Looking at this chart on the weekly and the previous post for the daily 89 there is support then a bounce. On the daily it has broken the 89 ema a little then bounced. The weekly 21 ema has also been broken a little then bounce. 2.0045 is about where the daily 89 ema is .
Attached Image
  • #512
  • Edited 7:44pm Aug 13, 2007 7:24pm | Edited 7:44pm
I have no doubt a credible strategy may be developed based on historical retracements, after all people keep telling us/me that trading is ALL about high probability.. so what could be better than actual statistical probabilities which you've pointed the results are related to a pattern of human emotion which seems to be fairly consistent.

In order to develop a strategy the questions asked from the data must be more refined.. like IF Movement = Bearish AND Daily XX EMA Slope = Bearish THEN IF Ret. (weekly)= 61.8%-+10pips AND Price FAILS Beyond THEN Sell .. Hell a lot of thinking must go into this and also much price action discretion must be applied
Bemac is doing a great job but is trading very low timeframes for me. For me, TP's would be based on dynamic fibs from an MA applied to median/closing price.

Who is this Jr97? I don't see him anywhere
  • #513
  • Aug 13, 2007 9:42pm Aug 13, 2007 9:42pm
I'm working on a PHP/SQL page which you can upload exported charts history and it will tell you the retracements and whatnot for any time period you select, average daily, weekly, monthly range, how many times the range hit 100, 120, 150, etc etc.. just been really busy with work and learning MQL. Will post it in here when I am finished.
  • #514
  • Aug 13, 2007 9:45pm Aug 13, 2007 9:45pm
Quoting bobblong
I have no doubt a credible strategy may be developed based on historical retracements, after all people keep telling us/me that trading is ALL about high probability.. so what could be better than actual statistical probabilities which you've pointed the results are related to a pattern of human emotion which seems to be fairly consistent.

In order to develop a strategy the questions asked from the data must be more refined.. like IF Movement = Bearish AND Daily XX EMA Slope = Bearish THEN IF Ret. (weekly)= 61.8%-+10pips AND Price FAILS Beyond THEN Sell .. Hell a lot of thinking must go into this and also much price action discretion must be applied
Bemac is doing a great job but is trading very low timeframes for me. For me, TP's would be based on dynamic fibs from an MA applied to median/closing price.

Who is this Jr97? I don't see him anywhere

Am developing an EA for this right now.. got it working just testing for bugs, with the little free time I do have lately, basically once it reaches your minimum specified range (ei: 100 for daily, or 200 daily on pairs like gbp/jpy) (or even 400 for a weekly range), once it hits the 23 fib, it is likely to continue to 38 and maybe 50, 61.. so it buys/sells at 23 and t/p is set at 38.. will also update when i have finished this.

Only problem is actually picking the daily or weekly tops or bottoms for retracements, but I've found even if you don't pick the top or bottom, theres a good chance of retracement. my ea gets these smaller retracements (say if the range is at 120, and it takes the trade, then the range goes up to 190, it takes a trade then as well, if it retraces again)
  • #515
  • Aug 14, 2007 1:10am Aug 14, 2007 1:10am
Quoting ScottH
I'm working on a PHP/SQL page which you can upload exported charts history and it will tell you the retracements and whatnot for any time period you select, average daily, weekly, monthly range, how many times the range hit 100, 120, 150, etc etc.. just been really busy with work and learning MQL. Will post it in here when I am finished.
Be looking forward to it
  • #516
  • Aug 14, 2007 1:19am Aug 14, 2007 1:19am
Quoting bobblong
For those of you that haven't been following the thread,
I was inspired by goodthing's work and decided to do my own stats for GBP USD using 8 years of historical data.

Here are some results (finally) based on weekly highs and lows:

In the past 8 years consisting of 410, 5 day trading weeks
GBP USD has retraced to

61.8% : 254 times, for a 62% probability retracement to this level WITH respect to Bullish/Bearish movement during the same week or the following week

50% : 292 times, for a 71% probability

38.2% : 335 times, for an 82% probability

23.6 : 378 times, for a 92% probability

For only 32 weeks out of 410 weeks has GBP USD not retraced to any fib level during the same or following week and simply continued trending towards an extension for an 8% probability OR just traded in a very narrow range not going up or down..

I will have more interesting things to know about the last 8 years soon.

Some notes regarding probability:

Probability ONLY means that an event is likely to occur, it does not necessarily mean that if there is a 92% probability of something happening then then it's sure as hell gonna happen as opposed to an event with 8% probability, but that is the most likely event based on past data.

GOODTHINGS: Thank you for the eye-opening experience in this thread.

P.S I'd initially tried to apply logic for retracement before the week's end in Excel but it was waaaay too much headbanging and seems irrelevant to trading realistically, but if someone is an Excel guru reading this I'd appreciate some help to do it anyway.
Did you happen to make note of the individual years. For instance how does 2005 or 2004 compare to the overall average. Do you have a break out of year to year.
  • #517
  • Edited 5:19am Aug 14, 2007 4:51am | Edited 5:19am
Scott: Sounds interesting would love to take a peek. Though I am suprised that you're having trouble picking weekly highs and lows.. here's the Excel fomula for picking the W-High relevant to the first week in the sheet

=IF(AND(A2>=A3,A2>=A4,A2>=A5,A2>=A6),A2,
IF(AND(A3>=A2,A3>=A4,A3>=A5,A3>=A6),A3,
IF(AND(A4>=A3,A4>=A2,A4>=A5,A4>=A6),A4,
IF(AND(A5>=A4,A5>=A3,A5>=A2,A5>=A6),A5,
IF(AND(A6>=A5,A6>=A4,A6>=A3,A6>=A2),A6)))))

When feeding in daily data, the problem was that there's no time signature .. so values had to be compared, on occasion the high and low of the week occurs on the same day and must be accounted for (IF HIGH & LOW = SAME DAY > THEN IF > CLOSE LOWER THAN OPEN, MOVEMENT = BEARISH ELSE BULLISH) Total ROM for the week was used for the calculation from the top & bottom, but that is a given.
Attached Image (click to enlarge)
Click to Enlarge Name: mysheet.jpg Size: 43 KB
  • #518
  • Edited 6:55am Aug 14, 2007 6:13am | Edited 6:55am
goodthings:

Here are the results broken down, year by year.
Some slippage of a day or two occurs between some years so a week in year XXXX may include 1 or two days from the following or previous year, but not always.

All based on 5 day trading weeks

Final part of 1999 (10.2 Weeks):

61.8: 59%
50.0: 69%
38.2: 78%
23.6: 88%

2000: (52 Weeks)

61.8: 54%
50.0: 70%
38.2: 87%
23.6: 94%

2001: (52 Weeks)

61.8: 71%
50.0: 77%
38.2: 87%
23.6: 96%

2002: (52 Weeks)

61.8: 54%
50.0: 65%
38.2: 79%
23.6: 88%

2003: (52 Weeks)

61.8: 56%
50.0: 64%
38.2: 71%
23.6: 85%

2004: (53(slippage) Weeks)

61.8: 68%
50.0: 80%
38.2: 85%
23.6: 91%

2005: (52 Weeks)

61.8: 58%
50.0: 62%
38.2: 77%
23.6: 87%

2006: (52 Weeks)

61.8: 67%
50.0: 77%
38.2: 81%
23.6: 92%

2007 (34 Weeks upto 31/7/07)

61.8: 59%
50.0: 68%
38.2: 80%
23.6: 91%

I'd have to say the results are very consistent with only 10%~ variance between years and in more recent times hardly any variance.
Would be interesting to have results for at least 30-40 years
  • #519
  • Aug 14, 2007 7:48am Aug 14, 2007 7:48am
Quoting bobblong
When feeding in daily data, the problem was that there's no time signature ..
In the future, you may wish to use to use this VBA macro to separate the Year, Month and Day from the 'text' date that's imported from MQL4. Three blank colums need to be inserted to the right of the Date colum before running it.

Public Sub DateToNumbers()
Dim FullDate, Yr, MD, M, D
Dim counter As Integer
For counter = 2 To 100 'These numbers assume one colum heading and 100 rows of data

FullDate = Worksheets("GBPUSD60").Cells(counter, 1) 'Name of worksheet is current worksheet
' Example 2007.01.02
Yr = Left(FullDate, 4) 'returns 2007
MD = Right(FullDate, 5) 'returns 01.02
M = Left(MD, 2) 'returns 01
D = Right(FullDate, 2) 'returns 02
Worksheets("GBPUSD60").Cells(counter, 2) = Yr
Worksheets("GBPUSD60").Cells(counter, 3) = M
Worksheets("GBPUSD60").Cells(counter, 4) = D
Next counter

End Sub

If you wish I can send you a worksheet with the macro in it, but I know how sensative some people are about macro viruses
  • #520
  • Aug 14, 2007 8:45am Aug 14, 2007 8:45am
Bobblong: I just mean it's impossible to know what the high/low will be. Of course if the weekly average is 300 pips, and it starts to retrace at 315 (or even 280 for that matter), then that is obviously a good point to place a trade, however it always *could* just be a minor retracement, and go in the original direction to say 370 pips before retracing again, which would be the real weekly range.
2 traders viewing now