simple, tradable, awesome statistics

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

Moderator: moderators

deltaskelta
rank: 50+ posts
rank: 50+ posts
Posts: 56
Joined: Tue Jul 03, 2012 10:08 pm
Reputation: 4
Gender: Male
Contact:

Postby deltaskelta » Thu Jul 19, 2012 3:59 pm

Oh man, I feel great. From the discussion at BP I have made a formula to express which TF is king.

The discussion is here :

http://forums.babypips.com/free-forex-t ... post375017

if you anyone wants to read what has been said so far, (page 104)

Relativity, I call the formula KING FACTOR. It goes like this:

(range mean / range stdev) / (range mean / time (minutes))

and it goes something like this:

Image





Its pretty obvious which TF is king so far, I just have to add W1 and MN1 in there as well to see what happens

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

Relativity
rank: 150+ posts
rank: 150+ posts
Posts: 364
Joined: Mon Nov 15, 2010 4:19 pm
Reputation: 0
Gender: None specified

Postby Relativity » Thu Jul 19, 2012 4:24 pm

I didn't do something like this. But the concept of 'at which point price / time movement will being to have an acceptable level of degrade'. I merely used this :

Time Factor for e.g M5 is 5, since the timeframe above it is M1x5
This one is simple.

Pip Factor for e.g M5 is 32.9, since 3.6 pips / 6 pips / 5 time factor * 100
This is to compare how well this TF pip wise really performed when compared to time factor. Aka speed.

Results for both balanced (only counting candles within candles, MTF wise) and flat (don't care about MTF).

Results look almost the same. H1 always 'performs well at least 70% of time'. The more interesting thing is D1 and W1. There will be a dip at D1, then at W1, one can see an obvious (attempt) towards 'recovery'.

I don't have enough data for balanced results for MN1 TF, but enough for flat results. One can see MN1 didn't attempt to 'continue the recovery'.

WHAT THE HELL IS THAT?! =>
Attachments
untitled1.PNG
Flat
untitled1.PNG (36.76 KiB) Viewed 4963 times
untitled2.PNG
Balanced
untitled2.PNG (33.71 KiB) Viewed 4963 times

deltaskelta
rank: 50+ posts
rank: 50+ posts
Posts: 56
Joined: Tue Jul 03, 2012 10:08 pm
Reputation: 4
Gender: Male
Contact:

Postby deltaskelta » Thu Jul 19, 2012 6:07 pm

EDIT: NVM everything I said before, if you read this post, I deleted the whole thing, because I finally got it after hours of reverse engineering your graph.
Last edited by deltaskelta on Thu Jul 19, 2012 10:07 pm, edited 1 time in total.

User avatar
PebbleTrader
rank: 1000+ posts
rank: 1000+ posts
Posts: 1633
Joined: Fri Nov 12, 2010 2:15 am
Reputation: 15
Gender: Male

Postby PebbleTrader » Thu Jul 19, 2012 9:29 pm

If we believe that price is the same no matter what time frame you are looking at...then there is no such thing as a superior time frame to trade. Price is moving the same amount.

Timeframes only show a condensed version of what actually took place. M1, M5, M15, M30, H1, D1, W1, etc., is an indicator that show only a small portion of the different price levels that actually occurred most often the high, low, open and close in the form of a shape (bar or candlestick).

I could create my own custom time frame and it wouldn't matter one bit.

The only reason we use different time frames is so we can see more or less data.

As for time, it is not as useful as some would believe. It's just there to create a sense of order in the market where without time, complete chaos would exist.

It is not so much the passage of time that really matters, but the subsequent intervals of price numbers that follow.

In the end, it doesn't matter to me how long it takes for a big fat check to arrive just as long as it's in a reasonable amount of time. 8)

For the most part, time should just be ignored

:)
Life is just a journey

deltaskelta
rank: 50+ posts
rank: 50+ posts
Posts: 56
Joined: Tue Jul 03, 2012 10:08 pm
Reputation: 4
Gender: Male
Contact:

Postby deltaskelta » Thu Jul 19, 2012 10:39 pm

PebbleTrader wrote:If we believe that price is the same no matter what time frame you are looking at...then there is no such thing as a superior time frame to trade. Price is moving the same amount.

Timeframes only show a condensed version of what actually took place. M1, M5, M15, M30, H1, D1, W1, etc., is an indicator that show only a small portion of the different price levels that actually occurred most often the high, low, open and close in the form of a shape (bar or candlestick).

I could create my own custom time frame and it wouldn't matter one bit.

The only reason we use different time frames is so we can see more or less data.

As for time, it is not as useful as some would believe. It's just there to create a sense of order in the market where without time, complete chaos would exist.

It is not so much the passage of time that really matters, but the subsequent intervals of price numbers that follow.

In the end, it doesn't matter to me how long it takes for a big fat check to arrive just as long as it's in a reasonable amount of time. 8)

For the most part, time should just be ignored

:)


I know what you are saying, and in theory I believe it is correct, but in reality people do give weight to time in order to limit the chaos.

So time must be looked at as a way to judge what your opponents are doing and provides a clear path to act accordingly. If there is a lot more people and money acting on a higher TF, then that TF will have more weight attached to it.

In short, the characteristics of price makes it "behave" the same on all TF, but what we are trying to do is find the TF with the highest magnitude of influence over the rest.

Just my humble interpretation and opinion on the matter.

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

deltaskelta
rank: 50+ posts
rank: 50+ posts
Posts: 56
Joined: Tue Jul 03, 2012 10:08 pm
Reputation: 4
Gender: Male
Contact:

Postby deltaskelta » Thu Jul 19, 2012 10:59 pm

Relativity wrote:I didn't do something like this. But the concept of 'at which point price / time movement will being to have an acceptable level of degrade'. I merely used this :

Time Factor for e.g M5 is 5, since the timeframe above it is M1x5
This one is simple.

Pip Factor for e.g M5 is 32.9, since 3.6 pips / 6 pips / 5 time factor * 100
This is to compare how well this TF pip wise really performed when compared to time factor. Aka speed.

Results for both balanced (only counting candles within candles, MTF wise) and flat (don't care about MTF).

Results look almost the same. H1 always 'performs well at least 70% of time'. The more interesting thing is D1 and W1. There will be a dip at D1, then at W1, one can see an obvious (attempt) towards 'recovery'.

I don't have enough data for balanced results for MN1 TF, but enough for flat results. One can see MN1 didn't attempt to 'continue the recovery'.

WHAT THE HELL IS THAT?! =>


Relativity, I had one observation of your charts now that I understand exactly what is going on in them.

The calculation for pip factor is = (higher TF range/lowerTF range)/(time factor)

but when it gets plotted into the purple curve, it is being multiplied by (time factor) again, thus canceling out the time factor, which is important in the calculation. So you are left with (higher tf range/lower tf range) I do not know if you did this on purpose or on accident, but it seems to me that the highest pip factor would be the king TF.

Is there something I am missing here?

deltaskelta
rank: 50+ posts
rank: 50+ posts
Posts: 56
Joined: Tue Jul 03, 2012 10:08 pm
Reputation: 4
Gender: Male
Contact:

Postby deltaskelta » Fri Jul 20, 2012 12:14 am

ok, so aside from my last question regarding what relativity posted, I think I have found some mistakes in my calculations from the KING FACTOR stat.

-The mistake was in the fact that the range/time part of the ratio did not account for the difference in magnitude of the two different TF ranges. I was calulating pips/minute, relativity's way (pip factor) calculates (% of range increase/% of time increase) which is done by doing [(higher TF range/lower TF range)/TF multiple]

-So now the King Factor is this

(pip factor)/(range/stdev)

or

[(higher TF range/lower TF range)/TF multiple]/[stdev/range]


-What this factor gives us is a rating of how "great" (pip factor) the TF is in terms of its coverage of pips
-and it plots that against how unpredictable (stdev/range) the TF is. a lower number here means it is more predictable AKA more normal distribution, this would also put a lower number in the denominator of the KING FACTOR calculation, making the king factor calculation larger when the TF is more predictable.

whew, this was a lot of work, how useful is it? I HAVE NO IDEA, I may have just wasted my time, but only time will tell.

BTW, I googled mean to stdev ratio and I guess it is called a coefficient of variance

http://en.wikipedia.org/wiki/Coefficient_of_variation

Image

Relativity
rank: 150+ posts
rank: 150+ posts
Posts: 364
Joined: Mon Nov 15, 2010 4:19 pm
Reputation: 0
Gender: None specified

Postby Relativity » Fri Jul 20, 2012 3:34 am

deltaskelta wrote:
PebbleTrader wrote:If we believe that price is the same no matter what time frame you are looking at...then there is no such thing as a superior time frame to trade. Price is moving the same amount.

Timeframes only show a condensed version of what actually took place. M1, M5, M15, M30, H1, D1, W1, etc., is an indicator that show only a small portion of the different price levels that actually occurred most often the high, low, open and close in the form of a shape (bar or candlestick).

I could create my own custom time frame and it wouldn't matter one bit.

The only reason we use different time frames is so we can see more or less data.

As for time, it is not as useful as some would believe. It's just there to create a sense of order in the market where without time, complete chaos would exist.

It is not so much the passage of time that really matters, but the subsequent intervals of price numbers that follow.

In the end, it doesn't matter to me how long it takes for a big fat check to arrive just as long as it's in a reasonable amount of time. 8)

For the most part, time should just be ignored

:)


I know what you are saying, and in theory I believe it is correct, but in reality people do give weight to time in order to limit the chaos.

So time must be looked at as a way to judge what your opponents are doing and provides a clear path to act accordingly. If there is a lot more people and money acting on a higher TF, then that TF will have more weight attached to it.

In short, the characteristics of price makes it "behave" the same on all TF, but what we are trying to do is find the TF with the highest magnitude of influence over the rest.

Just my humble interpretation and opinion on the matter.


Yes, its compressed time. I would have to have tick charts, but :

-We live in a real world where FX price is moved by majority; international transcations.
-There are only 120 trading hours a week (yes, there are valid debates about price moving and trading still going over weekend, but thats another day)
-Not everyone has access to good data / charting software. Part of my personal agenda is to bring out the best of MT4/MT5, as an edge to retail traders.

My research has shown a H1 candle in a certain hour is of different makeup of a H1 candle in another hour. It does give some level of weightage, since it suggests 'certain players that are expected are just not there OR has changed'. This to me, is helpful information, if I am gunning for more than 20 pips + looking for other players in the market that also want to push price in the same direction.

Otherwise, I will just 'take whatever I can' within 20 pips.

Losers in the market :
don't know how to hold onto good trades long enough
take small profits too early
die too quickly
are often tempted to lower TFs
tend to gather up over time as late buy orders in highs
tend to gather up over time as late sell orders in lows

These are related to time and can be read off the charts, at certain hours of the day / week + when at certain price levels + at certain timeframes.

I want to know where are they, thats all. So that I can take their money! So far, its working very well and is consistent with what I am already been doing (subconsciously) anyway. Just mapping these things out consciously.

Relativity
rank: 150+ posts
rank: 150+ posts
Posts: 364
Joined: Mon Nov 15, 2010 4:19 pm
Reputation: 0
Gender: None specified

Postby Relativity » Fri Jul 20, 2012 4:05 am

deltaskelta wrote:
Relativity wrote:I didn't do something like this. But the concept of 'at which point price / time movement will being to have an acceptable level of degrade'. I merely used this :

Time Factor for e.g M5 is 5, since the timeframe above it is M1x5
This one is simple.

Pip Factor for e.g M5 is 32.9, since 3.6 pips / 6 pips / 5 time factor * 100
This is to compare how well this TF pip wise really performed when compared to time factor. Aka speed.

Results for both balanced (only counting candles within candles, MTF wise) and flat (don't care about MTF).

Results look almost the same. H1 always 'performs well at least 70% of time'. The more interesting thing is D1 and W1. There will be a dip at D1, then at W1, one can see an obvious (attempt) towards 'recovery'.

I don't have enough data for balanced results for MN1 TF, but enough for flat results. One can see MN1 didn't attempt to 'continue the recovery'.

WHAT THE HELL IS THAT?! =>


Relativity, I had one observation of your charts now that I understand exactly what is going on in them.

The calculation for pip factor is = (higher TF range/lowerTF range)/(time factor)

but when it gets plotted into the purple curve, it is being multiplied by (time factor) again, thus canceling out the time factor, which is important in the calculation. So you are left with (higher tf range/lower tf range) I do not know if you did this on purpose or on accident, but it seems to me that the highest pip factor would be the king TF.

Is there something I am missing here?


Yes my mistake regarding the formula. Too many formulas = confusing. Thats why I code them once and for all and don't turn back lol.

The calculation for pip factor is = (higher range/lowerTF TF range)/(time factor)*100

(6 pips / 3.6 pips) / (5 time factor * 100)

No, the highest pip factor doesn't always mean its king. I am attempting to compare its performance in relation to time factor. Thats why I added it in to see how much % it really performs to its related time factor.

What surprised me is that D1 performed worse than W1 overall. And MN1 performed much worse.

This tells me 4 things :
-If I want to pick bottoms/tops, I am better off trying to pick them off D1 or MN1 extremes. If I end up being wrong, I can't end up that bad.
-Why D1 extremes? Maybe there are significant players exiting the market by the end of day?
-Why MN1 extremes? Internation companies tend to pay their workers by the month? There is more public panic when price hits quarterly/annual extremes? All the better; more prey into the market!
-If I want to ride trends or look for condition of current trend, I am better off looking at W1 bodies.

Seem to go along with the way I trade all right. If I want to study this further, I have to investgate freq dist + variance of winners wicks / losers wicks / bodies. ALL in one go. Been wanting to do that for a long time thou. 1/4 1/2 1/4 is generally correct, but I still see something that left untouched in this area.

User avatar
Jalarupa
rank: 1000+ posts
rank: 1000+ posts
Posts: 1300
Joined: Tue Feb 23, 2010 5:50 pm
Reputation: 324
Location: South Africa
Gender: None specified

Postby Jalarupa » Fri Jul 20, 2012 4:46 am

"My research has shown a H1 candle in a certain hour is of different makeup of a H1 candle in another hour. It does give some level of weightage, since it suggests 'certain players that are expected are just not there OR has changed'. This to me, is helpful information, if I am gunning for more than 20 pips + looking for other players in the market that also want to push price in the same direction."

Sure, its called the session opens ;-)
:smt065

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


Return to “statistics”