MidPoint statistics - surprising results!

never design a trading system without proper prior analysis. post your market statistics here

Moderator: moderators

Ali Son
rank: 150+ posts
rank: 150+ posts
Posts: 179
Joined: Sun May 28, 2006 1:30 am
Reputation: 0
Location: Boston, MA
Gender: None specified

Postby Ali Son » Tue Jun 06, 2006 3:41 pm

I did see and like your more extensive graph postings on TS, just now, which for some reason hadn't popped up ony my screen??? (I have to moitor this to see if's it's a bug as it's happened to me 2-3 times here), though I may have just missed them this time. VERY GOOD WORK!

Still a synopsis would be in order, as you (we) figure out how to implement it in a system, per your post on TRO's forum. I'm sure you must have some thoughts on it. Not all of us are mathematically inclined. For example, it seems strange to me that the 50% zone stats show an an expected (reasonable) decline in the probability that the "heads/tails" "roll" will continue BUT suddenly shoots back up to a 62% probability that price will be on that side of the midpoint out of the blue after a steady decline to oly 48% by the 7th consecutive bar.

If I am reading/interpreting it correctly, doesn't that seem odd, based on these statistically sisgnifcant results?

Also, what is your framework for developing a system, or do you prefer to wait until you've actually posted it to hopefully explain it? Maybe if you tell us now, people might be able to offer you some input as to whether you're approaching it from the right direction and save you some unnecessary backpedaling.


-OPTIONSWINGS

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

misterclean999
rank: <50 posts
rank: <50 posts
Posts: 4
Joined: Wed May 17, 2006 10:56 am
Reputation: 0
Gender: None specified

TS

Postby misterclean999 » Tue Jun 06, 2006 7:31 pm

Where on the TS forums can it be found?

User avatar
michal.kreslik
rank: 1000+ posts
rank: 1000+ posts
Posts: 1047
Joined: Sat May 13, 2006 2:40 am
Reputation: 36
Location: Monte Carlo, Monaco
Real name: Michal Kreslik
Gender: Male

Re: TS

Postby michal.kreslik » Tue Jun 06, 2006 7:38 pm

miseterclean, Ali,

if you can't see the attached images & files, please LOG IN :)

You should check "Log me on automatically each visit“ in the login dialogue. See this article on the theme: http://kreslik.com/forums/viewtopic.php?t=37

For security reasons and to reduce the server load, the users that are not logged on don't see any attachements in the posts.

I will post the Zone strategy along with the code soon.

obx
rank: <50 posts
rank: <50 posts
Posts: 41
Joined: Fri May 26, 2006 12:35 pm
Reputation: 0
Gender: Male

Postby obx » Wed Jun 07, 2006 12:58 pm

Michal,

I second Ali Son's request for a 'plain english' review of your thinking. I really appreciate what you've done with this and would really appreciate if you would take a few minutes and take us inside your head from the beginning up to a synopsis of you conclusions. If I remember correctly it all started with tunnel ... ?

Many thanks,

obx

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:

Re: Zone divider

Postby TheRumpledOne » Thu Jun 08, 2006 7:27 am

I use previous high/low.

So I use the midpoint of the previous bar.

That's the whole point!!

heyen wrote:Hello Folks,

there is one problem to the statistic:

3 consecutive same bars (equal H,L,C) will show up as 3 consecutive positive or negativ bars, but the price itself did not change over time (only within the bars).

i recommend rerunnig using the previous close instead the midpoint as reference. H & L will only be known after the bars is closed. Also only previous H & L should be referenced to avoid "lookoing into the future".

Sorry Avery, this remained unnoticed in the StockFetcher Forum.

But after all this work i had to point that out!

Thomas

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

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 Jun 08, 2006 7:32 am

I asked Michal to run stats to confirm what I saw.

Look at the 60 minute EURUSD chart. Once the price crosses the middle, if you trade in that direction, you never lose, unless you are greedy.

You can rake 5 or more pips almost every time.

After the euro moves 3 times in the same direction, time to look for a reversal.

Look at the daily chart. The 3 times seems to hold true here too.

Easy money!

User avatar
michal.kreslik
rank: 1000+ posts
rank: 1000+ posts
Posts: 1047
Joined: Sat May 13, 2006 2:40 am
Reputation: 36
Location: Monte Carlo, Monaco
Real name: Michal Kreslik
Gender: Male

Postby michal.kreslik » Fri Jun 09, 2006 12:22 am

As promised, I have written a sample trading system to illustrate the above concept in real trading. I posted the article into the Strategy trading forum in the following thread:

http://kreslik.com/forums/viewtopic.php?p=238#238

aspTrader
rank: 50+ posts
rank: 50+ posts
Posts: 57
Joined: Fri May 19, 2006 5:05 am
Reputation: 0
Gender: None specified

Postby aspTrader » Fri Jun 09, 2006 2:47 am

Hi Michal,

This is very interesting...

I have a thought for you.... In using Time based bars, a variable you can't control for is the Price Height of each bar. However, with Constant Range Bars all the bar heights would be the same.

It seems to me that this same study would be extremely interesting to perform on CRBs.

User avatar
michal.kreslik
rank: 1000+ posts
rank: 1000+ posts
Posts: 1047
Joined: Sat May 13, 2006 2:40 am
Reputation: 36
Location: Monte Carlo, Monaco
Real name: Michal Kreslik
Gender: Male

Postby michal.kreslik » Fri Jun 09, 2006 9:29 am

aspTrader,

certainly. Not only I am curious how my strategies would behave on the CRBs but that would really be a treat to be able to optimize the bar height/number of ticks/timeframe, too.

Ahhh, I can only dream about it with the prehistoric Tradestation optimizer. But, the end is near :) After all the 6/6/2006 was a devilish date wasn't that :smt077

I am looking forward to your new NeoTicker optimizer presentation!

aspTrader
rank: 50+ posts
rank: 50+ posts
Posts: 57
Joined: Fri May 19, 2006 5:05 am
Reputation: 0
Gender: None specified

Postby aspTrader » Fri Jun 09, 2006 11:52 am

Hi Michal...

Yes... The NeoTicker optimizer has exceeded the expectations I had... and I had high expectations because, well, I believe in Lawrence Chan.

The NeoTicker Grid Optimizer has options that make it more than just a brute force optimizer... There isn't enough documentation on this yet so it's hard to say what exactly is going on... But a friend started an opt run last night that had 56 million iterations.... Let's see today how far along he is.

Regarding this zone analysis and CRBs...

I think what really is important to analyze with CRBs is not midpoint zone statistics but statistics related to Bar Overlap... The point is... When you're working with bars of variable price heights you can't expect to be too exacting in your analysis.

But when all Bars are the same price height, new kinds of analysis can be performed that are more exacting. I've done a bit on this but there is more to do. For example...

"Bar Overlap" is a term I made up to describe cases where:

High[0] > High[1] and Low[0] < High[1]

"Bar Overlap" is the difference specified in the equation:

High[1] - Low[0]

I have a BarOverlap indicator that displays BarOverlap discretely for each bar and also contains a running summation of BarOverlap values for certain lookback periods... but I haven't taken it very far. I'd be happy to work further on this kind of thing with you if/when you decide to start using the #1 trading platform on the market...

-- aspTrader

:)
Last edited by aspTrader on Fri Jun 09, 2006 12:25 pm, edited 1 time in total.

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


Return to “statistics”