Statistics, Statistics and more Statistics

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

Moderator: moderators

pika
rank: 50+ posts
rank: 50+ posts
Posts: 74
Joined: Tue Sep 21, 2010 5:00 pm
Reputation: 0
Gender: None specified

Postby pika » Mon Dec 06, 2010 3:41 pm

Thanks for sharing, Relativity. You have substantiated your case with the fixed sample statistics and I will consider the implications and your conclusions seriously. Since the results do not seek to reflect the cost efficiency of trading under the different timeframes, it will be interesting to see where the optimal point is if a spread of, say, 2-3pips can be factored into the study, and if possible, uncovering the corresponding optimal trading frequencies. Once again, thanks for sharing this interestng fact with everyone here.

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 » Tue Dec 07, 2010 3:42 pm

I am not sure from here. What I am sure and currently testing out according to the analysis (and seems to work very well so far) is this :

When using a trend following system (e.g new rat seems to be a trend following system, unless I got it wrong):
Work forward through TFs looking at avg range or avg body
T/P #1 = M15 avg range/body
T/P #2 = M30 avg range/body or even H1 avg range/body depending on volatility
When using a retracement catching system (e.g. old rat) :
Work backwards through TFs looking at avg wicks
T/P #1 = M1440 avg wick
T/P #2 = M240 or M60 avg wick

S/L for both systems approach = M5 avg range + spread

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

Postby Relativity » Mon Feb 14, 2011 4:57 pm

I spotted something very regular with these statistics. I realised its very tradable! (FYI i did implement this finding into my current system, seems to work well, but will test further) Can anyone else see it?
Attachments
eurusd stats.png
eurusd stats
eurusd stats.png (38.7 KiB) Viewed 7257 times
audusd stats.png
audusd stats
audusd stats.png (39.59 KiB) Viewed 7257 times
usdjpy stats.png
usdjpy stats
usdjpy stats.png (40.11 KiB) Viewed 7259 times

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 » Mon Feb 14, 2011 11:26 pm

Old Rat vs New Rat?

It's the same Rat!!

The difference is using Rat Reversals on D1 to select which pairs to trade.
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.

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

Postby Relativity » Tue Feb 15, 2011 3:37 am

TheRumpledOne wrote:Old Rat vs New Rat?

It's the same Rat!!

The difference is using Rat Reversals on D1 to select which pairs to trade.


Yes I see that as well, but the other thing is that the body : wick ratio on the average is pretty consistent across all timeframes. Of course we want to use the D1 candle since there will be more cheese to eat.

The wick size may also vary somewhat with different pairs, like USDJPY has it at less than 20 pips on the average. The ratio is slightly different as well. But its still tradable I suppose.

SB talked about the rat zone being both mobile in position and varying in size. My current system is now using a 96 LWMA outer band on M15 TF (with deviation based on body/wick ratio, computed around 1.9 to sometimes 2.2 depending on pair) to capture this.

The band can contract or expand in size depending on volatility. It can move with price. I call this the dynamic rat zone. Works almost like boilinger bands. It keeps me away from vertical markets, which increases whipsaw, unless a trend setup is prepared in advanced. If so, the same dynamic rat zone becomes a basic trend channel, allowing breakout trades (like a semafor?).

It also tells me when horizontal markets are (IMO nice for rat trading since you can scale in and out of the zones all day without getting hurt at all) are available. Then this same zone becomes a standard rat reversal zone.

At least, this is what I've observed so far.

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

pika
rank: 50+ posts
rank: 50+ posts
Posts: 74
Joined: Tue Sep 21, 2010 5:00 pm
Reputation: 0
Gender: None specified

Postby pika » Tue Feb 15, 2011 4:41 am

Relativity wrote:Yes I see that as well, but the other thing is that the body : wick ratio on the average is pretty consistent across all timeframes. Of course we want to use the D1 candle since there will be more cheese to eat.

The wick size may also vary somewhat with different pairs, like USDJPY has it at less than 20 pips on the average. The ratio is slightly different as well. But its still tradable I suppose.

SB talked about the rat zone being both mobile in position and varying in size. My current system is now using a 96 LWMA outer band on M15 TF (with deviation based on body/wick ratio, computed around 1.9 to sometimes 2.2 depending on pair) to capture this.

The band can contract or expand in size depending on volatility. It can move with price. I call this the dynamic rat zone. Works almost like boilinger bands. It keeps me away from vertical markets, which increases whipsaw, unless a trend setup is prepared in advanced. If so, the same dynamic rat zone becomes a basic trend channel, allowing breakout trades (like a semafor?).

It also tells me when horizontal markets are (IMO nice for rat trading since you can scale in and out of the zones all day without getting hurt at all) are available. Then this same zone becomes a standard rat reversal zone.

At least, this is what I've observed so far.


Relativity,
You are right about the relative wick:body ratio. My finding is similar to your conclusion. Wicks to body ratio is 25%:50%25% on average and consistent with all pairs. I cannot fully exploit this fact as the daily range and candlestick patterns can vary. But it does serve as a guide when to take profit for rat reversal method since we are trading off the wicks.

I am interested to know how your method works in the dynamic band as you have mentioned (96 LWMA?). I'm sorry to say that the charts you've posted look complicated to me and rather cluttered with notes which I can't understand. If it's possible, would you mind sharing it in a more concise way? For me, I just set the reversal trade zone at 10% of the long range daily ATR below/above the prevailing high/low prices instead of the 20 pips range, as different pairs have different prices and daily ranges so it is more proportionate to use a ratio instead of fixed range pips.

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

Postby Relativity » Tue Feb 15, 2011 1:13 pm

We have to overcome a few problems when looking into D1 ratios in this manner.

1- Standard D1 candles being of different timezone's due to different brokers.
I do not think we can use the standard D1 unless you trade a specific market timing/open, since the standard D1 is broker timing based. I trade almost all over the place, so it doesn't make sense for me to use a small D1 candle that is just formed. That is if the broker is based on NY timing. I solved the problem partially using M15 x 96 candles and putting them into a box marked with high/low/start time/end time. (ref: SB thread)

2- View of chart can be distorted
I tried to use ratio like how you did, but I realised the point of trading candle wicks is about getting pips either the average length known (my method) or a length that occurs at the highest frequency (TRO's method). Both happened to be almost the same length (20 pips for most pairs), so I am delighted with the results. Not the length of the current candle's wick derived from the current D1 candle ratio taken from its current range.

I understand about the issue of 'opportunity', but I realised the risk is not worth it since it will distort my view of the chart. Too many calculations here and there (e.g. in this case averaging price once and then averaging it further a 2nd time) can become a huge problem. Keeping calculations simple helps alot. Also, a single news based candle can blow up the ratio into pieces (spoken from my personal experience when I tried to use ratio).

About the M15 96 LWMA :

The idea is M15 x 96 = D1 candle timing. So the MA, in order to capture the day's movement has to be 96 on a M15 chart. And I chose M15 for its Pip/Time E being a good baseline.

How did it end up being LWMA : If you read the SB thread, there is a system 'created' by Wavetop/SB/whoever. It uses 108 LWMA on H1 and captures the weekly wave movement very very well. 108 hours = approx 4.5 days, around 1 week which is actually 4.5 trading days, excluding 0.5 day for the erratic news friday; very much excellent common sense. So I experimented using LWMA to capture the daily movement, 'just to see if it works'. So far, not bad!

But this doesn't tell me the top of the wave, but the bottom/correction. So I borrowed an idea from bollinger bands and played around it for a while. Standard deviations are a very basic math calculation like averages, so I thought why not test them out? The standard bands uses 2 dev for its outer bands, which happened to be almost the same as the body/wick ratio.

So you see, a lot of my current system comes from pieces of other systems that I find useful in following and capturing the PA, especially the PA entering the rat zone and PA exiting the rat zone. This is what the LWMA Band + SB does.

I had to spend a lot of time to make sure all of them work together without conflicting each other. At the same time not distorting the real market/chart view. TRO's main idea (catching the wick) still trumps in the end of the day. He mentioned that method of entry doesn't matter, which is very true. I had to work it out myself, but I will never scrap the general idea of rat zones from my system.

Either way, I wonder how do you get your ranges? As in setting the reversal trade zone at 10% of the long range daily ATR. I am not a fan of ATR since I find it 'overly calculated'.

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 » Tue Feb 15, 2011 5:01 pm

Relativity wrote:We have to overcome a few problems when looking into D1 ratios in this manner.

1- Standard D1 candles being of different timezone's due to different brokers.
I do not think we can use the standard D1 unless you trade a specific market timing/open, since the standard D1 is broker timing based. I trade almost all over the place, so it doesn't make sense for me to use a small D1 candle that is just formed. That is if the broker is based on NY timing. I solved the problem partially using M15 x 96 candles and putting them into a box marked with high/low/start time/end time. (ref: SB thread)

2- View of chart can be distorted
I tried to use ratio like how you did, but I realised the point of trading candle wicks is about getting pips either the average length known (my method) or a length that occurs at the highest frequency (TRO's method). Both happened to be almost the same length (20 pips for most pairs), so I am delighted with the results. Not the length of the current candle's wick derived from the current D1 candle ratio taken from its current range.

I understand about the issue of 'opportunity', but I realised the risk is not worth it since it will distort my view of the chart. Too many calculations here and there (e.g. in this case averaging price once and then averaging it further a 2nd time) can become a huge problem. Keeping calculations simple helps alot. Also, a single news based candle can blow up the ratio into pieces (spoken from my personal experience when I tried to use ratio).

About the M15 96 LWMA :

The idea is M15 x 96 = D1 candle timing. So the MA, in order to capture the day's movement has to be 96 on a M15 chart. And I chose M15 for its Pip/Time E being a good baseline.

How did it end up being LWMA : If you read the SB thread, there is a system 'created' by Wavetop/SB/whoever. It uses 108 LWMA on H1 and captures the weekly wave movement very very well. 108 hours = approx 4.5 days, around 1 week which is actually 4.5 trading days, excluding 0.5 day for the erratic news friday; very much excellent common sense. So I experimented using LWMA to capture the daily movement, 'just to see if it works'. So far, not bad!

But this doesn't tell me the top of the wave, but the bottom/correction. So I borrowed an idea from bollinger bands and played around it for a while. Standard deviations are a very basic math calculation like averages, so I thought why not test them out? The standard bands uses 2 dev for its outer bands, which happened to be almost the same as the body/wick ratio.

So you see, a lot of my current system comes from pieces of other systems that I find useful in following and capturing the PA, especially the PA entering the rat zone and PA exiting the rat zone. This is what the LWMA Band + SB does.

I had to spend a lot of time to make sure all of them work together without conflicting each other. At the same time not distorting the real market/chart view. TRO's main idea (catching the wick) still trumps in the end of the day. He mentioned that method of entry doesn't matter, which is very true. I had to work it out myself, but I will never scrap the general idea of rat zones from my system.

Either way, I wonder how do you get your ranges? As in setting the reversal trade zone at 10% of the long range daily ATR. I am not a fan of ATR since I find it 'overly calculated'.


Hey Relativity,

Great work btw I hope you can crack the market dna one day...

Just an observation, I think the 108LWMA came from my speculation apon seeing a forex factory picture posted by the SB... I used that as my base case scenario because at the time it was working for me to pick strong S&R levels in a method created at the II forum site. However closer inspection made me realize that the bands offering S&R were however bollinger bands of some sort...

Also regarding the idea of catching the wick, it is also echoed strongly by MightyOne in his enter off the Daily+extreme teachings...

All and all it looks like you got the makings for a interesting system... but yeah news can upset any system very fast... I think this is why TRO doesn't like to trade before the news...

pika
rank: 50+ posts
rank: 50+ posts
Posts: 74
Joined: Tue Sep 21, 2010 5:00 pm
Reputation: 0
Gender: None specified

Postby pika » Tue Feb 15, 2011 5:40 pm

Relativity wrote:We have to overcome a few problems when looking into D1 ratios in this manner.

1- Standard D1 candles being of different timezone's due to different brokers.
I do not think we can use the standard D1 unless you trade a specific market timing/open, since the standard D1 is broker timing based. I trade almost all over the place, so it doesn't make sense for me to use a small D1 candle that is just formed. That is if the broker is based on NY timing. I solved the problem partially using M15 x 96 candles and putting them into a box marked with high/low/start time/end time. (ref: SB thread)

2- View of chart can be distorted
I tried to use ratio like how you did, but I realised the point of trading candle wicks is about getting pips either the average length known (my method) or a length that occurs at the highest frequency (TRO's method). Both happened to be almost the same length (20 pips for most pairs), so I am delighted with the results. Not the length of the current candle's wick derived from the current D1 candle ratio taken from its current range.

I understand about the issue of 'opportunity', but I realised the risk is not worth it since it will distort my view of the chart. Too many calculations here and there (e.g. in this case averaging price once and then averaging it further a 2nd time) can become a huge problem. Keeping calculations simple helps alot. Also, a single news based candle can blow up the ratio into pieces (spoken from my personal experience when I tried to use ratio).

About the M15 96 LWMA :

The idea is M15 x 96 = D1 candle timing. So the MA, in order to capture the day's movement has to be 96 on a M15 chart. And I chose M15 for its Pip/Time E being a good baseline.

How did it end up being LWMA : If you read the SB thread, there is a system 'created' by Wavetop/SB/whoever. It uses 108 LWMA on H1 and captures the weekly wave movement very very well. 108 hours = approx 4.5 days, around 1 week which is actually 4.5 trading days, excluding 0.5 day for the erratic news friday; very much excellent common sense. So I experimented using LWMA to capture the daily movement, 'just to see if it works'. So far, not bad!

But this doesn't tell me the top of the wave, but the bottom/correction. So I borrowed an idea from bollinger bands and played around it for a while. Standard deviations are a very basic math calculation like averages, so I thought why not test them out? The standard bands uses 2 dev for its outer bands, which happened to be almost the same as the body/wick ratio.

So you see, a lot of my current system comes from pieces of other systems that I find useful in following and capturing the PA, especially the PA entering the rat zone and PA exiting the rat zone. This is what the LWMA Band + SB does.

I had to spend a lot of time to make sure all of them work together without conflicting each other. At the same time not distorting the real market/chart view. TRO's main idea (catching the wick) still trumps in the end of the day. He mentioned that method of entry doesn't matter, which is very true. I had to work it out myself, but I will never scrap the general idea of rat zones from my system.

Either way, I wonder how do you get your ranges? As in setting the reversal trade zone at 10% of the long range daily ATR. I am not a fan of ATR since I find it 'overly calculated'.

Thanks Relativity, for your unreserved sharing. I will have to read your message carefully to understand it and maybe seek your advice again. You must have taken a lot of time and effort to integrate these concepts together.

How I get my ranges - I use 10% x average daily range (ATR 100 on D1). E.g. ATR100 of GBPUSD = 150pips, so the reversal trade zone will be within 15 pips from the prevailing daily high/low. I will wait for price to retest the zone with 2 or more M15 high/low bars penetrating the zone before taking a trade. Watching the price action and candlestick pattern within the zone is important. Stop loss is set to the same zone width (e.g. 15 pips for GBPUSD).

Why 10% of ATR? Rationale is if the average wick is 25% of ATR, then I can expect to profit the remainng 15% of the ATR, theoretically. Even on days when the range happen to be smaller, e.g. 60% of ATR or a 90pips range using the GBPUSD example, there is still a good chance to gain 25%x90 - 15 = 7.5 pips, provided that daily candle happen to fit with the wick:body ratio.

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 » Tue Feb 15, 2011 9:15 pm

Relativity wrote:
TheRumpledOne wrote:Old Rat vs New Rat?

It's the same Rat!!

The difference is using Rat Reversals on D1 to select which pairs to trade.


Yes I see that as well, but the other thing is that the body : wick ratio on the average is pretty consistent across all timeframes. Of course we want to use the D1 candle since there will be more cheese to eat.

The wick size may also vary somewhat with different pairs, like USDJPY has it at less than 20 pips on the average. The ratio is slightly different as well. But its still tradable I suppose.

SB talked about the rat zone being both mobile in position and varying in size. My current system is now using a 96 LWMA outer band on M15 TF (with deviation based on body/wick ratio, computed around 1.9 to sometimes 2.2 depending on pair) to capture this.

The band can contract or expand in size depending on volatility. It can move with price. I call this the dynamic rat zone. Works almost like boilinger bands. It keeps me away from vertical markets, which increases whipsaw, unless a trend setup is prepared in advanced. If so, the same dynamic rat zone becomes a basic trend channel, allowing breakout trades (like a semafor?).

It also tells me when horizontal markets are (IMO nice for rat trading since you can scale in and out of the zones all day without getting hurt at all) are available. Then this same zone becomes a standard rat reversal zone.

At least, this is what I've observed so far.


"The band can contract or expand in size depending on volatility. It can move with price. I call this the dynamic rat zone. "

Careful or you will wind up in Yale!!
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 “statistics”