Middle analysis - Taking home the Pips

free & uncensored discussion arena for TheRumpledOne

Moderator: moderators

pitboss
rank: <50 posts
rank: <50 posts
Posts: 31
Joined: Sat Apr 12, 2008 3:21 am
Reputation: 0
Gender: Male

Postby pitboss » Wed Aug 20, 2008 3:14 am

RicG started paper trading on 5-18-08 and uses no [0] differencial difference between the open and the middle - only qualification is open less than middle go long and open greater than middle go short - close position at end of day.

Using this criteria with my tp/sl factors, the results would be:
Total number periods 272
Total number qualifying positions 268
Total winning positions 139
Winning percentage 51.9%
Total losing postions 129
Losing percentage 48.1%
Net Advantage 4.69%

Had RicG used a .003 difference between the opening price and the middle, as I advocate, the results would have been:
Total number periods 272
Total number qualifying positions 117
Total winning positions 62
Winning percentage 53.0%
Total losing postions 55
Losing percentage 47.0%
Net Advantage 9.22%

Whether usage of betting no more than the advantage was or was not applied would make no difference in deriving the new advantage at the end of a new set of trials. It only effects the probability of losing one's bankroll and going busted.

TP and SL:
Analysis has shown that the amount up is equal to the amount down over an extended period of time - basically it is a wash with no definitive biasis either way which makes sense as this is a zero sum game. All my sl and tp do is to exit a position within a 90 percentile (1.65 * stdev) of 100% possibilities and eliminates those gains/losses that are on the far out extremes.

Using set number of pips and applying universally:
Doesn't work! 100 pips to GBP @ 1.9000 = .53% whereas 100 pips to CHF @ 1.1000 = .91% (huge difference). The key is to use percentages relative to open, bid, ask, mid, tp, sl and etc.

Number of trials relative to expectation:
The greater the number of trials the closer the statistical results come to the true probability of the event occuring. 272 trials is still a very short trial period.

Look at this way. The probability of choosing heads, flipping a coin and the coin coming up heads is 50%. If the coin was flipped twice there is only a 50% chance of the probability being exactly correct of 1 head and 1 tail. There is a 50% chance that the extreme will occur either H H or T T.

However, if the coin was flipped 10 times there is a greater probability of there being as many heads as tails with the amount of difference between their numbers each being very small. For example the probability of getting 10 heads is 1 in 1,024 - the extremes become further and further out from the center.

Please add www.kreslik.com to your ad blocker white list.
Thank you for your support.

khalid
rank: 50+ posts
rank: 50+ posts
Posts: 110
Joined: Wed Mar 14, 2007 2:15 pm
Reputation: 0
Location: Lahore
Gender: Male

Postby khalid » Wed Aug 20, 2008 4:06 am

Pitboss,

Thank you for taking the time and effort to restate the system.

So the WRONG is not in the system but in its application; it was clear something was amiss.

This is, indeed, a complete system.

NO NEED for ANY filter or additional stops.

Please accept my apologies for not reading the the thread from the start before making my suggestions.

I was so shocked at the speed and the size of the (paper) losses.

Khalid

User avatar
RicG
rank: 150+ posts
rank: 150+ posts
Posts: 274
Joined: Tue Jul 24, 2007 11:44 pm
Reputation: 8
Location: Nashville, TN
Gender: Male

Postby RicG » Wed Aug 20, 2008 9:29 am

Hi Pitboss,

Good to "see"you again. The result I posted on August 17th, does include the differential difference between the open and the middle. After you pointed out my initial error I went back to the 5-18-08 period and updated my data. Additionally when you posted later that you started using a SL and TP, I started applying those as well. However, my stats are using a set number of pips and as your post points out that "doesn't work". I'll start implementing that immediately.

The past two weeks the system did experience an extreme losing streak and that's the result of all four major pairs being in very strong trends, a concern I expressed back in May. Have you ever attempted to identify these trends in their early stages and stop trading until they exhaust themselves or do you just trade through them?

One more question, does your spreadsheet that calculates SL, TP, and position size need to be updated with new data to optimize the system? And if so, how often do you do that?

Thanks for all your help,

RicG


pitboss wrote:RicG started paper trading on 5-18-08 and uses no [0] differencial difference between the open and the middle - only qualification is open less than middle go long and open greater than middle go short - close position at end of day.

Using this criteria with my tp/sl factors, the results would be:
Total number periods 272
Total number qualifying positions 268
Total winning positions 139
Winning percentage 51.9%
Total losing postions 129
Losing percentage 48.1%
Net Advantage 4.69%

Had RicG used a .003 difference between the opening price and the middle, as I advocate, the results would have been:
Total number periods 272
Total number qualifying positions 117
Total winning positions 62
Winning percentage 53.0%
Total losing postions 55
Losing percentage 47.0%
Net Advantage 9.22%

Whether usage of betting no more than the advantage was or was not applied would make no difference in deriving the new advantage at the end of a new set of trials. It only effects the probability of losing one's bankroll and going busted.

TP and SL:
Analysis has shown that the amount up is equal to the amount down over an extended period of time - basically it is a wash with no definitive biasis either way which makes sense as this is a zero sum game. All my sl and tp do is to exit a position within a 90 percentile (1.65 * stdev) of 100% possibilities and eliminates those gains/losses that are on the far out extremes.

Using set number of pips and applying universally:
Doesn't work! 100 pips to GBP @ 1.9000 = .53% whereas 100 pips to CHF @ 1.1000 = .91% (huge difference). The key is to use percentages relative to open, bid, ask, mid, tp, sl and etc.

Number of trials relative to expectation:
The greater the number of trials the closer the statistical results come to the true probability of the event occuring. 272 trials is still a very short trial period.

Look at this way. The probability of choosing heads, flipping a coin and the coin coming up heads is 50%. If the coin was flipped twice there is only a 50% chance of the probability being exactly correct of 1 head and 1 tail. There is a 50% chance that the extreme will occur either H H or T T.

However, if the coin was flipped 10 times there is a greater probability of there being as many heads as tails with the amount of difference between their numbers each being very small. For example the probability of getting 10 heads is 1 in 1,024 - the extremes become further and further out from the center.

dennism1965
rank: <50 posts
rank: <50 posts
Posts: 3
Joined: Thu Sep 20, 2007 10:24 pm
Reputation: 0
Gender: None specified

MQ4 for Middle Analysis

Postby dennism1965 » Wed Aug 20, 2008 8:13 pm

Did someone code this using .003. Otherwise I could Neo or MT4.

pitboss
rank: <50 posts
rank: <50 posts
Posts: 31
Joined: Sat Apr 12, 2008 3:21 am
Reputation: 0
Gender: Male

Postby pitboss » Wed Aug 20, 2008 9:07 pm

Updated forex positioning excel sheet. Incorporates number of lots relative to bankroll, advantage, dollar conversion and etc. It uses 1.65 standard deviations for stop loss and 1.96 for take profit.

As far as 'up dating' or adjusting the factors, with so much trial data I have found that any 'changes' are so very-very tiny, thousands of a percentage point, that it just doesn't make sense to mess with as nothing is gained.

Regarding dealing with trends such as recently experienced, I just continue trading through them (yeah, I ate the crap too). There is no way to 'predict' a trend whether short term or extended with any measurable degree of accuracy. If there were, I would certainly be trading in the direction of the trend, get rich and lay around the beach all day.

If I flipped a coin and it landed on heads 5 consecutive times is this then a trend that heads will come up the next flip or the next 5 (the trend follower)? Or is a tail 'due' to come up the next flip or the next 5 (the contrarian)?

Actually, none of these are true. The next flip is exactly a 50-50 proposition. What occured previously has no bearing what-so-ever concerning the probability of the event's occurance.

There are hundreds of gurus and pundunts hawking how to make a bundle by beating the markets and/or forex. They'll have magnificent, elaborate charts and secret alghorithums to 'predict' what will happen in a fluid, random market place. The fact is I could have a monkey draw a dart at a board marked with Go Long or Go Short and over a period of time my monkey 'predictor' will match the 'experts' accuracy of predicting the future.
Attachments
forex_buy_position.xls
(240.5 KiB) Downloaded 228 times

Please add www.kreslik.com to your ad blocker white list.
Thank you for your support.

jhtumblin
rank: 50+ posts
rank: 50+ posts
Posts: 113
Joined: Tue Jun 13, 2006 3:42 am
Reputation: 0
Location: Atlanta, GA
Gender: None specified

Postby jhtumblin » Thu Aug 28, 2008 8:33 am

There are hundreds of gurus and pundunts hawking how to make a bundle by beating the markets and/or forex. They'll have magnificent, elaborate charts and secret alghorithums to 'predict' what will happen in a fluid, random market place. The fact is I could have a monkey draw a dart at a board marked with Go Long or Go Short and over a period of time my monkey 'predictor' will match the 'experts' accuracy of predicting the future.


This is my most favorite statement you have made yet. I'm glad we see eye to eye on it.

User avatar
TheRumpledOne
rank: 10000+ posts
rank: 10000+ posts
Posts: 15544
Joined: Sun May 14, 2006 9:31 pm
Reputation: 3035
Location: Oregon
Real name: Avery T. Horton, Jr.
Gender: None specified
Contact:

Postby TheRumpledOne » Thu Aug 28, 2008 8:55 pm

I was born in the Year of the Monkey, does that count?
IT'S NOT WHAT YOU TRADE, IT'S HOW YOU TRADE IT!

Please do NOT PM me with trading or coding questions, post them in a thread.

Please add www.kreslik.com to your ad blocker white list.
Thank you for your support.


Return to “TheRumpledOne”