Math Help (Probability) & My Research Results | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

ECN && STP && Scalping && Hedging - Broker Review - Oct 2009 24 replies

Math Help: Probability 30 replies

Setup frequency and probability - Math problem 8 replies

help me do a math: what is Margin Call? 0 replies

Math help, Can't figure out if MT4 or I am dumber 2 replies

Attachments: Math Help (Probability) & My Research Results
Exit Attachments

Math Help (Probability) & My Research Results

  • #1
  • First Post: Sep 4, 2011 9:12am Sep 4, 2011 9:12am
Hello I will be happy to share some results but I also need some math help.


Basically there is a good % of the time where the market will take the shortest path.. in other words the market is more likely to do the least effort as opposed to the most effort.

What I need help in calculating is probabilities based on the smallest number...




for example if I was to take three number 200, 400, and 150


in this case

150 = 20%
200 = 26.66%
400 = 53.33%


The issue with forex is that the lowest number is actually the highest %.. the most probable to happen.. how can I calculate this equation such that the lower set of numbers is the highest %




If you do not understand what I am saying just focus on the bold and underline statement for now.. I will be happy to explain this in more depth.. but this is a true phenomenon in the market even tho it will piss most people off....
  • #2
  • Sep 4, 2011 9:24am Sep 4, 2011 9:24am
Invert the numbers by dividing 1 through the numbers, after that do the same math.

1/150 = 0,00666666667 = 47%
1/200 = 0,005 = 35.29%
1/400 = 0,0025 = 17.6%

(numbers more or less)

Hope that helps,
Cheers
  • #3
  • Sep 4, 2011 9:27am Sep 4, 2011 9:27am
Quoting esquire
Invert the numbers by dividing 1 through the numbers, after that do the same math.

1/150 = 0,00666666667 = 47%
1/200 = 0,005 = 35.29%
1/400 = 0,0025 = 17.6%

(numbers more or less)

Hope that helps,
Cheers
Is it really that simple? I asked this question in math forums for the last month on and off and they kept saying it couldn't be calculated due to some reason

other people answered way off topic..

so you can always invert a number by using 1.. makes since.. I kept trying to invert the probabilities
  • #4
  • Sep 4, 2011 10:13am Sep 4, 2011 10:13am
Quoting newyear498
Is it really that simple? I asked this question in math forums for the last month on and off and they kept saying it couldn't be calculated due to some reason

other people answered way off topic..

so you can always invert a number by using 1.. makes since.. I kept trying to invert the probabilities
There may be an issue here with rounding errors.. is there anyway I can avoid this.. I think I need my data to be extremely accurate with the way it rounds


I notice things are off by a %'s so either I am wrong or there is a rounding error

Thanks
  • #5
  • Sep 4, 2011 6:24pm Sep 4, 2011 6:24pm
Check this short calculation i made with some random numbers (attachment), all adds up to one (so it should be correct). Also see how the underlying differences are about the same (on the left, 0.57 is more or less double of 0.30, on the right 0.0092 is more or less the double of 0.0049), supporting the idea that it sort of does the trick.

Perhaps it's not the correct way to do it (im actually pretty sure of that), but I think it's a very close approximation. The right way to do it will probably be a complex algoritm, if it's possible at all.

Anyway, just try it on some of your data, see if you like the outcomes and decide whether to use this or not. If this deosn't suite you, then i'm not able to help any further i'm afraid

Cheers
Attached File(s)
File Type: xls Map1.xls   19 KB | 704 downloads
  • #6
  • Sep 4, 2011 6:41pm Sep 4, 2011 6:41pm
Quoting newyear498
Hello I will be happy to share some results but I also need some math help.


Basically there is a good % of the time where the market will take the shortest path.. in other words the market is more likely to do the least effort as opposed to the most effort.

What I need help in calculating is probabilities based on the smallest number...




for example if I was to take three number 200, 400, and 150


in this case

150 = 20%
200 = 26.66%
400 = 53.33%


The issue with forex is that...

How do you come to that conclusion? you can not take a number and assign a % to it, that's not math!
  • #7
  • Sep 4, 2011 7:11pm Sep 4, 2011 7:11pm
Quoting AmeriTrini
How do you come to that conclusion? you can not take a number and assign a % to it, that's not math!
150+200+400 = 750

750 is the total data

thus 150/750 = 20%

etc
  • #8
  • Sep 4, 2011 8:37pm Sep 4, 2011 8:37pm
Quoting newyear498
150+200+400 = 750

750 is the total data

thus 150/750 = 20%

etc
I'm not sure, but I think AmeriTrini may not have been questioning how you got the percentages. Clearly, 150 is 20% of 750 - and so on. I think he/she may have meant why you assigned 750 as 100%.
  • #9
  • Sep 4, 2011 8:52pm Sep 4, 2011 8:52pm
Quoting 2+2=4ex
I'm not sure, but I think AmeriTrini may not have been questioning how you got the percentages. Clearly, 150 is 20% of 750 - and so on. I think he/she may have meant why you assigned 750 as 100%.
750 = sum of all observations therefore 100% when converted to %
"To hold, you must first open your hand. Let go." - Lao Tzu
  • #10
  • Sep 4, 2011 8:58pm Sep 4, 2011 8:58pm
my take on esquire's numbers:

ps: newyear, why not rounding using common sense (if number is closest to x then number is x.)
Attached File(s)
File Type: xls newyear498.xls   15 KB | 413 downloads
"To hold, you must first open your hand. Let go." - Lao Tzu
  • #11
  • Edited 9:26pm Sep 4, 2011 8:59pm | Edited 9:26pm
Quoting 2+2=4ex
I'm not sure, but I think AmeriTrini may not have been questioning how you got the percentages. Clearly, 150 is 20% of 750 - and so on. I think he/she may have meant why you assigned 750 as 100%.
Basically I have run backtests and recorded market data that behaves near random

So I already know the figures I am looking for.. and thus far esquires method is almost perfectly correlated to the data I am looking at



There is a huge gap of me being shitty at math and trying to convey my message.. that is the issue here


Let me lay it out in a non forex way.. basically if I have 750 people.. there are three towns with the sizes 400,200, and 150.

150 represents 20% of the total population..

I know the smallest towns in my case are surprisingly the producers of the most pollution.. thus I need to figure out how much pollution each group is making. (and I assume all data is normally distributed)




Anyways.. I could just be going off explaining bad again but theres only so many ways I can reword it.




In forex terms.. if you have three situations.. one where price fluctuates 400 pips.. other two are 200, and 150

well you are more probable that you are going to fluctuate 150 pips than you would fluctuate 400


Case in point is I have the data infront of me from several time periods and such.. using method esquire presented is extremely correlated and thus I have some interesting finds between random distributions IMHO




not that I am a nay say'er that the markets are random.. I'm just saying they are extremely close to in specified ways and using such data can be used as an edge.

The degree of randomness screws you over when your trying to profit in a directional bias, yet when trying to profit from a random market the non random aspects screw you over... thats some additional finding.. I am probably at a point now where most people would have no idea what I'm talking about since most are hard and strict in supporting or opposing random market theory.
  • #12
  • Sep 4, 2011 9:05pm Sep 4, 2011 9:05pm
Quoting vox dei
750 = sum of all observations therefore 100% when converted to %
Understood. Never was a time I didn't understand that. And I don't care to know what the 750 is. I was only elaborating on what AmeriTrini might have meant. NewYear replied to him by telling him 150 is 20% of 750, but I thought that was plainly obvious and so I was just stating that maybe he was questioning how he came up with 750 as a total.
  • #13
  • Sep 4, 2011 9:07pm Sep 4, 2011 9:07pm
Quoting 2+2=4ex
Understood. Never was a time I didn't understand that. And I don't care to know what the 750 is. I was only elaborating on what AmeriTrini might have meant. NewYear replied to him by telling him 150 is 20% of 750, but I thought that was plainly obvious and so I was just stating that maybe he was questioning how he came up with 750 as a total.
Allright then. I should have quoted AmeriTrini instead.
"To hold, you must first open your hand. Let go." - Lao Tzu
  • #14
  • Sep 4, 2011 9:08pm Sep 4, 2011 9:08pm
Quoting vox dei
my take on esquire's numbers:

ps: newyear, why not rounding using common sense (if number is closest to x then number is x.)
Thanks.. yeah rounding is not a huge issue.. my initial guesses were wrong.. esquires data actually is more accurate and as a result found out I was not using a negative expectancy concept..

I originally thought some of my data was reading 50% but now I understand it was more like 49.30%.. which is what the market data said (but I assumed market data can only be so close by law of large numbers etc.) so I assumed it was really more like 50%..

long story short since its 49.30% my theory was negative due to spread only and can be worked around..


Thanks again
  • #15
  • Sep 4, 2011 9:14pm Sep 4, 2011 9:14pm
Ok take a box snap shot of the market.

We have something that looks like this

100
75
50
25
0
-25
-50
-75
-100


0 = your start point.. in a random world you would assume that you have a 50/50 chance of price making it to 100 or -100



well here is why my little numbers are more probable.. listen..

100 pips represents going directly up to +100

200 pips represents dropping to -50.. then going up to +100



So we have a 100 pip journey and a 200 pip journey.. the data shows the smallest journey is the most probable one... thus this is why I need small number to be represent the larger probability of happening... esquires method seems to work.. if you have a better way (or if I esquire is wrong for whatever reason) then feel free to suggest a proper method. (Again I'm basically working under assumptions of random distribution)



THANK YOU!!!
  • #16
  • Sep 4, 2011 9:16pm Sep 4, 2011 9:16pm
Quoting newyear498
Basically I have run backtests and recorded market data that behaves near random

So I already know the figures I am looking for.. and thus far esquires method is almost perfectly correlated to the data I am looking at



There is a huge gap of me being shitty at math and trying to convey my message.. that is the issue here


Let me lay it out in a non forex way.. basically if I have 750 people.. there are three towns with the sizes 400,200, and 150.

150 represents 20% of the total population..

I know the smallest towns in my case are...
I completely understand where you are going. As I stated in my previous post, I had no ambitions of intruding on your research or numbers, but if I had to guess I figured you were working on correlations (in pips). I think you said it best in your last paragraph. The markets sometimes show flashes of non-randomness, yet you cannot rely on it because it is random enough to still mess you up. It also seems to me that you may be looking at too small a window. This also depends on what pair you are talking about of course. Trends are the edge. And they are the edge because they don't just occur for no reason. When you figure that part out, then it just becomes a matter of when to enter and when to exit.
  • #17
  • Sep 4, 2011 9:26pm Sep 4, 2011 9:26pm
Quoting newyear498
Ok take a box snap shot of the market.

We have something that looks like this

100
75
50
25
0
-25
-50
-75
-100


0 = your start point.. in a random world you would assume that you have a 50/50 chance of price making it to 100 or -100



well here is why my little numbers are more probable.. listen..

100 pips represents going directly up to +100

200 pips represents dropping to -50.. then going up to +100



[b]So we have a 100 pip journey and a 200 pip journey.. the data...
Maybe you're on to something. I'm no math wiz myself. However to me, this strategy (or methodology) looks like it is inviting the black swan in with a wide open door. Just seems too risky at a glance. Again, I'm no math expert so I can't say whether this has some sort of statistical edge.

Remember, you also have to figure in time and costs. Why take a strategy that works 60% of the time, but takes 3 years to double account when you can trade a strategy that works 20% of the time but doubles account in 1 year?
  • #18
  • Sep 4, 2011 9:30pm Sep 4, 2011 9:30pm
Quoting 2+2=4ex
Maybe you're on to something. I'm no math wiz myself. However to me, this strategy (or methodology) looks like it is inviting the black swan in with a wide open door. Just seems too risky at a glance. Again, I'm no math expert so I can't say whether this has some sort of statistical edge.

Remember, you also have to figure in time and costs. Why take a strategy that works 60% of the time, but takes 3 years to double account when you can trade a strategy that works 20% of the time but doubles account in 1 year?
Certainly!!!

How the risk to reward is skewed is a completely different angle to my question. I seem to understand the R:R however.. the probability is what I was confused about.

Good point none the less
  • #19
  • Sep 4, 2011 9:31pm Sep 4, 2011 9:31pm
Quoting 2+2=4ex
Trends are the edge. And they are the edge because they don't just occur for no reason. When you figure that part out, then it just becomes a matter of when to enter and when to exit.
Just want to elaborate on this a little. A trend is not the only price movement someone can take advantage of. There are other price behaviours that normally carry higher probability. They also don't occur for just no reason.
  • #20
  • Sep 4, 2011 9:42pm Sep 4, 2011 9:42pm
Quoting 2+2=4ex
Just want to elaborate on this a little. A trend is not the only price movement someone can take advantage of. There are other price behaviours that normally carry higher probability. They also don't occur for just no reason.
absolutely.. in my eyes I view randomness in the sense that trades influencing the long term and non random movement of the market are not in sync with one another.. thus you have these traders placing orders at various times resulting in a random layer (which is essentially a normal distribution of order flow) building the non random.

So despite the fact the markets are not random.. you have a very close to normal distribution of order flow.. probably the best well put way I can word my views, research, and theories.

However as you mentioned about black swans.. a signature of a non normal distribution and a non random market.. this is a trippy circle I am looking around in
0 traders viewing now