Similarity: A Practical Approach | Page 24 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Similarity Trading - The ZZBB Method 215 replies

Rainbow similarity 256 replies

Help implementing zigzag similarity concept indicator 0 replies

Similarity Stuff 17 replies

Directional Similarity 0 replies

Attachments: Similarity: A Practical Approach
Exit Attachments

Similarity: A Practical Approach

  • #461
  • Jul 31, 2013 1:44pm Jul 31, 2013 1:44pm
Quoting FxGunner
{quote} Congrats Xuenn ! Aa a "practical approach" of the CI indicator : how do you handle the CI repainting ? I am always in the "learning phase". I hesitate into the two methods... CI is really easier but this is the repainting stuff that makes it not easy. Thank you.
Hi FxGunner,

Since I am new to this please take what I say with a huge spoon of salt.

I have lots of repainting going on in my CI as well.
I do not mind some repaintings actually if you trade in the trend. Most of the time, the price continues off in the main trend. (Check #1)

I used some of the methods in the old similarity thread (you can find them in Neio's Similarity CI pdf as well):

1.) Trading in the direction of the trend only. You can find similar ideas in the old thread.
Attached Image (click to enlarge)
Click to Enlarge Name: chartexample.jpg Size: 102 KB


2.) I use bollinger bands.
3.) My target is quite small (~10 pips). This may be more of a turn off.
4.) I have a stop loss set. If the stop loss is too far away, I will not take the trade (too risky for me).

I tried switching to a smaller time frame (5min) from my previous 30 min/1hr time frame to have a better picture about the repainting issue.
Perhaps you can observe the repainting using a demo account?

Anyway this is my understanding of the method.
Please correct me if there is any mistake.
  • #462
  • Jul 31, 2013 1:50pm Jul 31, 2013 1:50pm
Quoting Effex
{quote} Great post Xuenn! Congrats for your results! The CI method continues to be great profits for many The problem I had (and I think many others) was getting the settings right on the indicator. Did you use Eurusdd original formula on the indicator or did you have to adjust anything? Do you mind posting your settings for us to try out as it seems it differs a lot cross brokers as well when using CI. (in fact this was one of eurusdd main teachings as well - cross broker dissimilarities) Happy pips!
I did not use EURUSDD's original parameters for the CI.

I have two computers running the same platform of the same account- both of them are using different parameters!
The same parameters on my two different computers produced different charts.

I am ashamed to say that I actually haven't got much clue about the workings of the Cycle Identifier (what are the data used etc). Perhaps other users can shed some light on this? Especially on the price action filter and the length.
  • #463
  • Jul 31, 2013 1:53pm Jul 31, 2013 1:53pm
Quoting xuenn
{quote} Hi FxGunner, Since I am new to this please take what I say with a huge spoon of salt. I have lots of repainting going on in my CI as well. I do not mind some repaintings actually if you trade in the trend. Most of the time, the price continues off in the main trend. (Check #1) I used some of the methods in the old similarity thread (you can find them in Neio's Similarity CI pdf as well): 1.) Trading in the direction of the trend only. You can find similar ideas in the old thread. {image} 2.) I use bollinger bands. 3.) My target is quite...
Apparently if you have entered this bar, you would have netted in ~50 pips by now

Attached Image (click to enlarge)
Click to Enlarge Name: chartexample2.jpg Size: 86 KB
  • #464
  • Jul 31, 2013 2:31pm Jul 31, 2013 2:31pm
Quoting xuenn
{quote} Hi FxGunner, Since I am new to this please take what I say with a huge spoon of salt. I have lots of repainting going on in my CI as well. I do not mind some repaintings actually if you trade in the trend. Most of the time, the price continues off in the main trend. (Check #1) I used some of the methods in the old similarity thread (you can find them in Neio's Similarity CI pdf as well): 1.) Trading in the direction of the trend only. You can find similar ideas in the old thread. {image} 2.) I use bollinger bands. 3.) My target is quite...
Really thank you for sharing your ideas. Wish you all the best pips.

Ok, i will work on that. The cycle identifier is based on two moving averages as i can see in the code. If i have something valuable to share, i will.
Enter Signature
  • #465
  • Jul 31, 2013 3:51pm Jul 31, 2013 3:51pm
Quoting xuenn
{quote} The same parameters on my two different computers produced different charts.
This is really interesting. I wonder if it is a memory thing on your PC's....
  • #466
  • Jul 31, 2013 6:23pm Jul 31, 2013 6:23pm
One of the steps I've completed. The principle of similarity must be used from tf 5 min. 5min tf according to the data used is antipersistent and therefore does not meet the conditions of fractality so the price behaves randomly. Despite having much less data, I think insufficient, the monthly tf is the better show. I will continue my analysis in TF 30, 1H and 4H. This gives us the market testing calculation shows relationships between these tf and the possible application of algorithms to indicate equality between these tf
  • #467
  • Jul 31, 2013 6:33pm Jul 31, 2013 6:33pm
I'm back!

Thank you for those who spotted my earlier post and encouraged me to plow ahead.

I've carried on with Awesome Oscillator, but trying a different approach - constructing a short AweOsc and a longer one on only M1 (previously I compared M5 to M1). The longer inevitably tries to play the role of AweOsc on M5.

So, I compared AweOsc(50,25,Close) with AweOsc(10,5,Close). Results a bit unexpected - only a 50% similarity between signals! For the sake of the similarity principle, I tested different parameter after parameter to get as strong similarity as possible (to ease things I kept the original 2x distance between slow and fast period).

Parameter set of 48,24,Close produced similarity of 90%.

Initial testing on 1 month of data, keeping the same closing point of SL7/TP4, trades opened at bar opening after confirmed dissimilarity:
For 90% similarity: win/loss 0.65, 442 trades.
For 50% similarity: win/loss 0.66, 1184 trades.

I wasn't happy with M5/M1 signals, as they were not even close to relevant turning points, but M1-only-signals are different. For the 90% similarity trades are happening quite close to tops and bottoms even though win/loss ratios are quite sad. It seems that the signals themselves are more nondirectional than directional. It's a fuss to nondirectionalize the signals, but this what I'm gonna do next! I'm yet to test another indicators, but I'm feeling more upbeat about SIM than I was yesterday.
  • #468
  • Jul 31, 2013 6:40pm Jul 31, 2013 6:40pm
Quoting burnssss
One of the steps I've completed. The principle of similarity must be used from tf 5 min. 5min tf according to the data used is antipersistent and therefore does not meet the conditions of fractality so the price behaves randomly. Despite having much less data, I think insufficient, the monthly tf is the better show. I will continue my analysis in TF 30, 1H and 4H. This gives us the market testing calculation shows relationships between these tf and the possible application of algorithms to indicate equality between these tf
Hi,

Are you comparing one tf to another tf for fractality or are you analysing timeframes separately? If it is not a secret, how are you carrying it out?
  • #469
  • Jul 31, 2013 9:23pm Jul 31, 2013 9:23pm
Quoting Me3w
{quote} I thought her CI is used to capture the +/- divergence/reversal where the RSI divergence/reversal did not capture. Is it you called the Dissimilarity in the indicators ? OR your dissimilarity in the indicators is refer to the stochdiff created the zone ? How about her derivative osc, can it use for the same purpose ?
EURUSDD said early in the primary thread that any two indicators can be used to spot dissimilarity as long as those indicators are similar 90 percent of the time or better. The dissimilarity is in the indicators. I have not used her derivative oscillator enough to know.
If you are good at something, never do it for free--Joker
  • #470
  • Jul 31, 2013 11:12pm Jul 31, 2013 11:12pm
Quoting candero
I'm back! Thank you for those who spotted my earlier post and encouraged me to plow ahead
Enter Signature
  • #471
  • Aug 1, 2013 3:23am Aug 1, 2013 3:23am
Quoting candero
{quote} Hi, Are you comparing one tf to another tf for fractality or are you analysing timeframes separately? If it is not a secret, how are you carrying it out?
I am conducting a study of tf separately. Through a series of mathematical expressions to calculate the Hurst exponent, and if this exponent is between 0.5 and 1 indicates that the series has fractal behavior. The calculations are very simple and are as follows:

- First calculate the price returns: return = Ln (Pt/Pt-1)
-After calculating the average of the returns
- Calculate deviations from the mean (x)
-calculate cumulative deviation (y)
- Calculate the range of (y) ---- (R)
- Calculate the standard deviation of returns (S)
- We divide R between S, R and S range the standard deviation
- Calculate the Ln (R / S)
-calculate the ln (n) where n is the number of elements of the series
- H = ln (R / S) / Ln (n)
  • #472
  • Aug 1, 2013 3:26am Aug 1, 2013 3:26am
I believe Hurst was one of the major contributors to this stuff, There is an mt4 indi if I remember but like most indi's wasn't much good.
  • #473
  • Aug 1, 2013 3:35am Aug 1, 2013 3:35am
The indi mt4 not make precise calculations because it is not a mathematical software
  • #474
  • Aug 1, 2013 4:44am Aug 1, 2013 4:44am
Daily Prediction Method

Hi all,
I didn't get time the last 2 days to post daily projections. Yesterday, I eyeballed the charts in the morning GMT and mentioned it could be an up day. It turned out it was. Today, I did the same although I had noted where the price had closed last night. Below is the method.

PREMISE
Eurusdd mentioned in the original thread that we had everything we needed on the chart to predict daily direction. He said he could do it within 1 hour of the new day. This fascinated me.

Looking for the answers, I discovered the following:

  1. ZZ = ZigZag Indicator, but you can also use Semafor
  2. 300EMA = 300 exponential moving average
  3. Price = Price on main chart. You can plot a 1EMA on it if you wish
  4. Stoch = Stochastic set to 75,1,1 for the One Hour and 50 Line dividing the indicator window
  5. PriceBB = 24 period Bollinger Band
  6. StochBB = 24 period Bollinger Band on the Stochastic Indicator window


1H CHART ONLY

  1. If PRICE is above 300EMA, the default daily direction will always be UP. Other things to look for to support an UP projection

    1. PRICE is also traveling the upper half of the bollinger band
    2. PRICE closes the day near a ZZBottom and the next major swing into the new day has to be a TOP.
    3. STOCHBB is above the 50 Line
    4. Stoch is also travelling the upper half of the bollinger band​

​​

  1. If PRICE is below 300EMA, the default daily direction will always be DOWN . Other things to look for to support an DOWN projection

    1. PRICE is also traveling the LOWER half of the bollinger band
    2. PRICE closes the day near a ZZTOP and the next major swing into the new day has to be a BOT.
    3. STOCHBB is below the 50 Line
    4. Stoch is also travelling the LOWERhalf of the bollinger band

​​

  1. Other things to look for

    1. Counter-trends during a main trend occur when PRICE:

      1. Creates a ZZ swing
      2. Travels from one end of the BB to the Other and crosses the MID

    2. If the Counter-trend move is in a Bullish Trend:

      1. When this happens, look for the StochBB to go lower than the 50 Line momentarily to signal a ZZBottom before continuing upwards

    3. If the Counter-trend move is in a Bearish Trend:

      1. When this happens, look for the StochBB to go higher than the 50 Line momentarily to signal a ZZTop before continuing upwards

    4. Changes In Trend

      1. Are Dramatic and tend to be Vertical in change. That means the price travels across Bollinger Bands in vertical fashion or a violent suddent way.
      2. The 300EMA and the 50 Stoch Line and the Bollinger Mid are measurements of the 50% line where it swings from UP/DOWN.

By taking the above into consideration, you are then able to make a projection based on the Hourly Closing price of the day + 1 hour into the new day.
I factor in the movement that was made during New York and London and think about what Asia would want to do.

I don't factor in news or announcements as this dilutes the above "signposts". If you can prove this for yourself - then you can tweak it by adding:

  1. News / Calendar events
  2. Candlestick patterns
  3. MFI volume (thanks Baillie!)
  4. Fibs and Pivots (I used this myself and it worked out well: if projection was up and Open +1 was above Daily Pivot, I was more confident).

Attached is the 1H as of now-ish. See how the StochBB has moved below the 50 Line!.... is that signalling something?

Enjoy!

Attached Image (click to enlarge)
Click to Enlarge Name: eurusdh1.png Size: 72 KB
A pip is only worth it if you know how much you risked to earn it
  • #475
  • Aug 1, 2013 4:54am Aug 1, 2013 4:54am
Quoting vlady1974
Daily Prediction Method Hi all, I didn't get time the last 2 days to post daily projections. Yesterday, I eyeballed the charts in the morning GMT and mentioned it could be an up day. It turned out it was. Today, I did the same although I had noted where the price had closed last night. Below is the method. PREMISE Eurusdd mentioned in the original thread that we had everything we needed on the chart to predict daily direction. He said he could do it within 1 hour of the new day. This fascinated me. Looking for the answers, I discovered the following:...

Thanks Vlady, this is great. Its really good to have clear guidelines. I am a bit of a butterfly when it comes to trading so all the different ideas have me trying out all sorts of things and getting distracted. I have to say that I have been profitable with most dissimmilarity ideas so long as I avoid trading around the news and have just opened a trade explorer in celebration.
  • #476
  • Aug 1, 2013 5:45am Aug 1, 2013 5:45am
Quoting vlady1974
Daily Prediction Method Hi all, I didn't get time the last 2 days to post daily projections. Yesterday, I eyeballed the charts in the morning GMT and mentioned it could be an up day. It turned out it was. Today, I did the same although I had noted where the price had closed last night. Below is the method. PREMISE Eurusdd mentioned in the original thread that we had everything we needed on the chart to predict daily direction. He said he could do it within 1 hour of the new day. This fascinated me. Looking for the answers, I discovered the following:...
This is great stuff!
Thanks a lot for sharing!
  • #477
  • Aug 1, 2013 5:55am Aug 1, 2013 5:55am
Quoting vlady1974
Daily Prediction Method Hi all, {image}
Hi Vlady, i hv similar way to determine general trend, but i use 288EMA (not so different with 300EMA) and BB 72 applied to that 288EMA with dev 1 and level +/- 2 to 10. we can use same setting for H1, M15, or M5.. because the 'time' ratio will more make common sense..
btw sorry if the image too wide..lol

Attached Image (click to enlarge)
Click to Enlarge Name: 20130801-174429 EU H1 288EMA+BB72 on EMA dev 1.jpg Size: 596 KB
  • #478
  • Aug 1, 2013 6:01am Aug 1, 2013 6:01am
Thanks for the feedback guys.

Currently seeing a "black hole" on the 5M chart which is why I've loaded up on longs for the 5M timeframe.... so essentially my time goal is to get a ZZTOP on the 15M. 30M has given a ZZBOT and Dissim ended giving birth to a zone that is higher than price - will it attract and repel price?

The conflicting signals are that 1H has not completed a ZZBOT yet. So I'm on a countertrend trade. But loaded up big style on shorts.
A pip is only worth it if you know how much you risked to earn it
  • #479
  • Aug 1, 2013 7:29am Aug 1, 2013 7:29am
Quoting vlady1974
Daily Prediction Method Hi all, I didn't get time the last 2 days to post daily projections. Yesterday, I eyeballed the charts in the morning GMT and mentioned it could be an up day. It turned out it was. Today, I did the same although I had noted where the price had closed last night. Below is the method. PREMISE Eurusdd mentioned in the original thread that we had everything we needed on the chart to predict daily direction. He said he could do it within 1 hour of the new day. This fascinated me. Looking for the answers, I discovered the following:...
Ha
Thank you very much for your explanetion.
Could you please say us which parameter use ZZ OR SEMAFOR?
Best Regards.
Bülent .
  • #480
  • Edited 7:59am Aug 1, 2013 7:45am | Edited 7:59am
Quoting xuenn
I have also set stop losses after the huge loss on 25th July. Still, I am having problem figuring out the core theory/idea behind the similarity system.
Hi, can you please show how (with charts, 2 chart), where, when you did get that huge loss?
like detailed explanation, thank you

there is a reasons why i want to see this...
first... maybe you was not trading right
or
this could fail sometimes(M30/H1), i think it could fail...when market is veryveryvery slow and get sudden volume coming in... or high impact news when price moves up and down very fast
1 trader viewing now