Most true.
I don't agree with the rest if what you said however. Just because you master trend trading doesn't mean other styles (such as tape reading/orderflow, mean reversion algorithms, etc.) can't work. You need to get over that typical mindset.
Market Statistics & Probabilities (requests welcome)
- #82
- Edited Jan 8, 2011 9:24am Jan 7, 2011 6:21pm | Edited Jan 8, 2011 9:24am
- Joined Nov 2010 | Status: Pips... or GTFO! | 1,023 Posts
Heres my 2 cents
nothing is "random" because that would mean the market ranged randomly.. there is a purpose to ranging so its not random.. an agreement on a price level or a mix of uncertainty.. whatever the reason its not random
When I use the term random I am usually using the term wrongly
Help me out.. whats the word to describe so many variables happening so quickly that it is extremely hard to analyze all the data before the time is up..
thats what I mistakenly (and probably many of us) refer to as random.. in reality its never random.. because since its the investors money moving the market it could only be random if investors were making trades randomly
I also use the term random to refer to any pattern I pick out to trade.. again its not the best term to describe it.. but its like picking any random pattern from the chart.. so I say.. ok.. inside bars are my randomly picked event (but again randoms wrong word since I chose inside bars for whatever reason.. they look like consolidation etc..)
so now I will remove the word random from the rest of this post.. but feel free to suggest the right word latter..
so looking at my inside bars there are just to many events that could take place around the world in a given second that can really provide any huge advantage my inside bar will react a certain way
so its this huge collection of information we must combat with.. imagine the world of star wars and your a day trader.. you never know when the death stars going to blow up a planet.. its 300 million light years away from you.. how can you process that information with an inside bar
so its this information overload most of us come up with other terms for.. and not just physical ones.. but also social human reaction to the event.. how many times have amazing earnings come out for a company and the next day price drops..
so i feel everything is happening for a reason and is technically not random.. but these things that happen for a reason don't always happen with the same signals like an inside bar.. but they could happen 60% of the time on inside bars...
and if all the conditions and data leads to 50% probabilities.. that does not mean the markets are "random" it just means that the events cannot be anticipated by any conditions we studied...
beyond just bar charts there are also the fundamentalist investors who hope certain conditions will lead to higher probability events..
company xyz found the cure to the common cold and warren buffet starts pouring his money into the xyz.. but theres to much information overload.. he places the trade just as terrorist blow up the company since it opposes there religious beliefs
the condition can't always overcome information overload but it might 99% of the time.. every action from searching for a cure to the common cold.. a reason to invest.. and a reason to blow up a company..
but I doubt this is much of a insight to many.. its pretty common stuff.. so I think myself and many are probably just misusing words if anything.. hope this was not a waste of time
nothing is "random" because that would mean the market ranged randomly.. there is a purpose to ranging so its not random.. an agreement on a price level or a mix of uncertainty.. whatever the reason its not random
When I use the term random I am usually using the term wrongly
Help me out.. whats the word to describe so many variables happening so quickly that it is extremely hard to analyze all the data before the time is up..
thats what I mistakenly (and probably many of us) refer to as random.. in reality its never random.. because since its the investors money moving the market it could only be random if investors were making trades randomly
I also use the term random to refer to any pattern I pick out to trade.. again its not the best term to describe it.. but its like picking any random pattern from the chart.. so I say.. ok.. inside bars are my randomly picked event (but again randoms wrong word since I chose inside bars for whatever reason.. they look like consolidation etc..)
so now I will remove the word random from the rest of this post.. but feel free to suggest the right word latter..
so looking at my inside bars there are just to many events that could take place around the world in a given second that can really provide any huge advantage my inside bar will react a certain way
so its this huge collection of information we must combat with.. imagine the world of star wars and your a day trader.. you never know when the death stars going to blow up a planet.. its 300 million light years away from you.. how can you process that information with an inside bar
so its this information overload most of us come up with other terms for.. and not just physical ones.. but also social human reaction to the event.. how many times have amazing earnings come out for a company and the next day price drops..
so i feel everything is happening for a reason and is technically not random.. but these things that happen for a reason don't always happen with the same signals like an inside bar.. but they could happen 60% of the time on inside bars...
and if all the conditions and data leads to 50% probabilities.. that does not mean the markets are "random" it just means that the events cannot be anticipated by any conditions we studied...
beyond just bar charts there are also the fundamentalist investors who hope certain conditions will lead to higher probability events..
company xyz found the cure to the common cold and warren buffet starts pouring his money into the xyz.. but theres to much information overload.. he places the trade just as terrorist blow up the company since it opposes there religious beliefs
the condition can't always overcome information overload but it might 99% of the time.. every action from searching for a cure to the common cold.. a reason to invest.. and a reason to blow up a company..
but I doubt this is much of a insight to many.. its pretty common stuff.. so I think myself and many are probably just misusing words if anything.. hope this was not a waste of time
- #83
- Edited 8:27pm Jan 7, 2011 7:11pm | Edited 8:27pm
- Joined Sep 2008 | Status: hardcore chartist | 2,173 Posts
- #84
- Jan 7, 2011 8:22pm Jan 7, 2011 8:22pm
- Joined Nov 2010 | Status: Pips... or GTFO! | 1,023 Posts
Ok,
I like the idea of researching the Gap's so you want to see how often gap gets closed from over the weekend correct?
another thing I am thinking about is "psychology numbers" since I never really traded off them what should I test.. 00,25,50 ... or 20,30 etc.. we can see how often price breaks through 80 and 20 it goes to 00
both of the two cases I listed above will be tested with its opposite difference.. so gap is 50 pips distance.. does it close gap or go up 50 pips
from 80 to 00 will it go to 00 first or the 60
I like the idea of researching the Gap's so you want to see how often gap gets closed from over the weekend correct?
another thing I am thinking about is "psychology numbers" since I never really traded off them what should I test.. 00,25,50 ... or 20,30 etc.. we can see how often price breaks through 80 and 20 it goes to 00
both of the two cases I listed above will be tested with its opposite difference.. so gap is 50 pips distance.. does it close gap or go up 50 pips
from 80 to 00 will it go to 00 first or the 60
- #85
- Edited 9:25am Jan 8, 2011 9:11am | Edited 9:25am
- Joined Nov 2010 | Status: Pips... or GTFO! | 1,023 Posts
- #88
- Jan 9, 2011 7:47pm Jan 9, 2011 7:47pm
- Joined Nov 2010 | Status: Pips... or GTFO! | 1,023 Posts
It is not.
btw SD = Std Deviation.
and what you meant by SD?
I just put SD to hint that is what my SD would refer to in my data below because I was lazy and writing fast.. but guess that backfired since look at all the characters I need to explain my reasoning.
Chart of Standard Distribution.. I didn't make it.. just found it a second ago and looks like someone already figure this out way before me lol
- #89
- Jan 10, 2011 1:36am Jan 10, 2011 1:36am
- | Membership Revoked | Joined Jun 2010 | 1,246 Posts
Not sure why you disagree? Please explain, thanks
and what you meant by SD?
I just put SD to hint that is what my SD would refer to in my data below because I was lazy and writing fast.. but guess that backfired since look at all the characters I need to explain my reasoning. http://cdn.forexfactory.com/images/s.../yim/happy.gif
Chart of Standard Distribution.. I didn't make it.. just found it a second ago and looks like someone already figure this out way before me lol
But the markets are not Gaussian, they're worse. lol
Not sure why you disagree? Please explain, thanks
and what you meant by SD?
I just put SD to hint that is what my SD would refer to in my data below because I was lazy and writing fast.. but guess that backfired since look at all the characters I need to explain my reasoning.
Chart of Standard Distribution.. I didn't make it.. just found it a second ago and looks like someone already figure this out way before me lol
Judging from how you use the word SD in your post, I guess you are confuse with standard normal distribution and standard deviation
- #91
- Jan 10, 2011 10:51am Jan 10, 2011 10:51am
- Joined Nov 2010 | Status: Pips... or GTFO! | 1,023 Posts
That's the good ol' Normal distribution plot (also called Gaussian distribution). It's the distribution you get when (for example) flipping coins. Your probability percentages are correct within a Gaussian context.
But the markets are not Gaussian, they're worse. lol
You may want to count the % bars that close within 20-bars Bowling Girl Band in last 10000 bars.
Judging from how you use the word SD in your post, I guess you are confuse with standard normal distribution and standard deviation
Thanks for correcting me.. again I think I'm a quant in training here lol its a good thing I posted that then because I would have never learned
Thank you!
- #92
- Jan 10, 2011 8:36pm Jan 10, 2011 8:36pm
- Joined Nov 2010 | Status: Pips... or GTFO! | 1,023 Posts
- #93
- Jan 10, 2011 9:39pm Jan 10, 2011 9:39pm
- Joined Nov 2010 | Status: Pips... or GTFO! | 1,023 Posts
- #94
- Jan 10, 2011 10:40pm Jan 10, 2011 10:40pm
- Joined Nov 2010 | Status: Pips... or GTFO! | 1,023 Posts
I'm just working on the basic rule of thumb that in general the longer the consolidation period (time) the bigger the movement away from that price range will be (movement) and trying to find a relationship there between time and how far it moves on average. The volume involved would also play a role but perhaps just keep it simple for now.
Ahh target! Thats another problem cause how do you judge distance? When do you stop measuring how far. Perhaps when the momentum dries up and it reverses or consolidates again I'm not sure!
Anyway the vouch...
haha.. well actually I agree that consolidation is surly stored energy.. I would say its like sleeping and you wake up your ready to kick ass until you get tired again
for me you can tell that 50% probabilities is what I am interested in personally
like you said about money management.. its all about figuring out how to cut loses and add to winners..
I am still a little shabby on the exact aspects you want tested.. I suppose you are interested in the average breakout range in comparison to the consolidation range?
I am pretty amazing with algorithms and problem solving as far as logical reasoning goes (not to sound conceded but I amaze myself a lot) its a weird experience to say the least lol
so I'm sure I could find a way to define this criteria..
- #95
- Jan 10, 2011 10:44pm Jan 10, 2011 10:44pm
- Joined Nov 2010 | Status: Pips... or GTFO! | 1,023 Posts
for the record I am doing some personal study this week
I am sure I will get back to doing some studies and tests such as the
gap fading
psychology levels (maybe, this presents some annoyances)
and breakout levels..
you can keep adding to the list if you want
but yup I'll get to it I just got some other things to do
I am sure I will get back to doing some studies and tests such as the
gap fading
psychology levels (maybe, this presents some annoyances)
and breakout levels..
you can keep adding to the list if you want
but yup I'll get to it I just got some other things to do
- #96
- Edited 1:29am Jan 11, 2011 12:34am | Edited 1:29am
Ok,
I like the idea of researching the Gap's so you want to see how often gap gets closed from over the weekend correct?
another thing I am thinking about is "psychology numbers" since I never really traded off them what should I test.. 00,25,50 ... or 20,30 etc.. we can see how often price breaks through 80 and 20 it goes to 00
both of the two cases I listed above will be tested with its opposite difference.. so gap is 50 pips distance.. does it close gap or go up 50 pips
from 80 to 00 will it go to 00 first or the 60
Thing is, I woke up this morning remembering some interesting thing from a thread I started a while back.
Let me share this single post with you by 7bit.
http://www.forexfactory.com/showpost...1&postcount=12
Below are the two pictures from the post.
The first chart shows closes. The x-axis has all the last two digits of a four digit broker obviously and the y-axis is the amount of times it closed at those levels. Around the 00 and 50 psych levels, price doesn't close often there. From what I've seen, it'll either range a bit around there but never close on the line or just zoom the hell past it.
The second chart shows the direction after closing at certain levels. Looks almost 50/50 and I don't give much weight to this chart because I can't put it to useful application but I'm appreciative for it nonetheless and know it has use.
Of course 50000 M1 candles is no where near 10 years worth of data. It's like 35 days.
Maybe you can turn it into 10 years, newyear
But what I really wanted to see if you can test is this:
First let me make two quick statements.
It's similar to the PDF you posted. You'll recall that the study tested every candle for 5 years (2000 - 2004) and watched 24 candles after each tested candle, they watched to see if a certain trigger was met after said candle closed. The triggers were 10, 15, 20, 25, and 50 after the candle close and once hit, they measured the probability of it going a certain distance with a 20 pip stop everytime.
Second, I will quote a quote.
Also, if I'm permitted to quote from his book (Beat the Odds in Forex Trading, copyright Igor Toschakov, Wiley & Sons, 2006; pages 114-5):
As far as I know, lots of traders usually place stops considering money management reasons only. ........ Technical levels such as trend lines, supports, or resistances (frequently being close at a short distance behind the stops) are not taken into consideration. Such a practice results in stops that are triggered at the most improper moment. Too frequently, a prospective and potentially profitable position is liquidated and a loss is taken. To avoid such situations, here is another rule I always follow: Place no stops based on reasons of money management only. The binding of a stop should always be done to a specific technical level.”
Bolded by hanover.
I want the stops to be equal to the triggers. This way if price is triggered, it will be stopped out by the round number and not by our money management strategy alone. Y'know... getting your pipsworth for your stop
So those two ideas being known, This is how I'd like this test to be designed. Count only candles that open near 00 or 50. Let's say it has to open within 5 pips of each? This can be changed if you think it'll yield better results or the entire criteria for which candles are counted can be changed to what you see fit.
The PDF only leaves the trigger on for 24 candles. I think this test should leave the trigger on indefinitely.
If we were to use ones within 5 pips of the 00 and 50, then place triggers (pending orders) x pips from the 00 or 50.
The value of x can by anything just the the PDF. Personally, I would like to see x equal 10pips and if you're not too busy, 20 would be interesting to see as well.
So once triggered 10 pips away from the 00 or 50, a stop will be placed 10 pips from the trigger. So in this case, right on the 00 or 50.
(In the 20 pip trigger case, it'd be a 20 pip stop)
And then possibly it can be tested like the PDF? Like show a probability for net +20 which was really +10 pips (Dunno why they couldn't of just said +10). So there would be one for +10pips after the trigger, +20, all the way up to... well whatever inconveniences least
And it would be tested on multiple timeframes like the PDF. For various opens and the such.
Of course you can change the entire design of the testing system. I was just trying to do a tiny bit of legwork for you because this particular statistic intrigues me the most.
Please let me know what you think and I hope maybe there will be some interesting numbers behind this.
http://www.forexfactory.com/attachme...4&d=1290553562
http://www.forexfactory.com/attachme...5&d=1290553562
Data by 7bit.
- #99
- Jan 11, 2011 9:42am Jan 11, 2011 9:42am
- Joined Nov 2010 | Status: Pips... or GTFO! | 1,023 Posts
You really are the man!
Thing is, I woke up this morning remembering some interesting thing from a thread I started a while back.
Let me share this single post with you by 7bit.
http://www.forexfactory.com/showpost...1&postcount=12
Below are the two pictures from the post.
The first chart shows closes....
Thanks for your kind words
I think the test I have in mind actually will do this.. and I realized I'm an idiot and can do this phyc test easy I just had an error in thinking when it came to my algorithm
STOPLOSS
in regards to the stop loss.. I actually have done some thought and I think it would make sense if you really wanted to optimize your stoploss to use the optimal level of volume
ofcourse forex doesn't have true volume so I don't wanna get started with this bullshit argument.. but we all know we can generate our own measure of volume
but also besides really needing volume if you measure the time the currency is spending in one price range you can determine just like market profile.. using the point of control in a currency I personally feel is the optimal stop loss
the idea is that the currency will have a point of control that it will always want to return to.. but of course throughout the day this point of control will change
and you can measure your support and resistant points this way.. in fact its really the same thing if not better than drawing a line in my opinion
so in a way when we as speculating retail traders are placing trades we are trying to predict where the new point of control will form.. and are battling this previous point of control that is trying to pull price back to where it was
so its kind of like a catch 20 in a way.. we have a current point of control price.. but we know that a new point of control has to form.. so we need to predict this new point of control.. but price is trying to return to the current point of control (understand why trading is so hard now?)
but the optimal stop loss will be the current point of control since if we are wrong in our prediction price will surly be returning to the current point of control
if you are not sure what I'm talking about.. well I'm talking about Market Profile and concepts from the book "Mind over Markets"
essentially to determine the Point of Control you would look at a period of say 24 units of 30M bars
for each price point inside the 30M bar it gets a value of 1.. the price levels with greatest value is the point of control.. so here in this example each letter represents a 30M bar that was touching that price level..
1285.00 M
1284.75 HM
1284.50 DEGHM
1284.25 DEGHM
1284.00 DEGHM
1283.75 DEFGHIMN
1283.50 DEFGHIMN
1283.25 DEFGHIMN
1283.00 DEFGHIJKMN
1282.75 DFGHIJKMN
1282.50 DFGJKLMN
1282.25 DJKLMN
1282.00 DJKLMN
1281.75 DJKLMN
1281.50 DLMN
1281.25 DLMN
1281.00 LMN
1280.75 LMN
1280.50 MN
so the point of control is 1283.00 in this example
well you mention something about Stop Loss and I just went off on my opinion of what would be the optimal stop loss.. hope you like
- #100
- Jan 11, 2011 9:57am Jan 11, 2011 9:57am
- Joined Nov 2010 | Status: Pips... or GTFO! | 1,023 Posts
Great thread Mr newyear.
With regards to the above, with these stats we should be able to make a profitable system from this?
the problem with these stats and the fallacy of open and closing prices (which I hate every fucking concept I hear that involves "only if price closes above this point" and stuff like that
open and close price is the coincidence of where price was at when all zero's aligned.. its the equivalent to where you were standing when the twin towers got hit by the plane
so what happens when opening price is a few ticks from the high of previous bar?
I ran the test again.. this is the result when you use the previous bars middle point rather than the newbars open price (which is same as last bars close price)
so the middle point of previous bar..
it makes new high.. chance it returns to previous bars mean = 38.92%
so now its made a new high and return to the mean point.. chance it makes it back to the high = 30.29%
and if you want to know the percent of times over a 10 year period that price made a new high.. return to mean of previous bar.. and then made it back to the high.. 10.22%
so as you can see.. the original test with open price was usually near the high when it succeeded.. and thus it often presented poor risk to reward
hope this helps