- This topic has 169 replies, 16 voices, and was last updated 9 years, 9 months ago by
.
- Posts
Can you please add alert on the first time that a currency becomes the strongest (CIX_v_2_2_post.mq4, or later version)?
Thanks.
G.
We can currently agree on few key elements of the EA’s Logic: Trend: ClearMethod: simplex latest version with current and 1 or 2 higher TF’s, Trend direction according to HTF, entries on areas of current TF in opposite to Trend direction. TVI (or TickVolume): (Optional) Trend direction defined by above/below zero, entries areas are in opposite direction. Exact bar for entry: VZO (or VolumeOsc): when above/below range, in opposite direction of Trend. Currency-Strength: Here we have a problem… This one should indicate strong/weak currencies (Sentiment) by their location above/below the “0” line. Comparing my ForexGT_Spaghetti with simplex CIX, I found that currencies location in relation to the “0” line are different. We need conclusion on this one. Attached is comparison (CIX is default params, USD majors even, Timeframe normalization) USD both in Green. G.
This is good post G , thanks a lot
I own a car with a six-step gearbox. I’ve never had the idea to use only one of them, let’s say third or fifth gear. When I drive my car I take the freedom to choose between all six gears with regard to the current situation, reflecting my experience as a seasoned driver. It’s the same way I’m using the input parameters of my indicators. It just does not make any sense to use default inputs for each and every comparison with another indicator. If comparison with ForexGT_Spaghetti is that important, I’m ‘again’ calling everybody who has access to that one to make use of the CIX gearbox (input parameters) and try to adjust it in a way that provides the best fit. Starting from that point we might turn this part of our discussion and development in a new direction. I do not have access to ForexGT_Spaghetti, so I can’t provide this specific detail. Apart from that, any idea from any thread contributor leading to an improvement of CIX is most welcome. Best, s. Edit: since I’m not Phil Connors, I’m out of this specific CS part of our discussion and development, until new information is on the table. Following scientific principles, ‘new information’ could be a so far unknown theory, preferrably based on replicable math, or new experimental evidence including detailed setup for everybody competent enough to reproduce the proposed results. This
while (true)loop requires abreak, re-entry not excluded. Please consider this a call for action targeted at everybody interested in our community project.Good discussion Simplex … and thanks for the zip file … will try that new toys
Can you please add alert on the first time that a currency becomes the strongest (CIX_v_2_2_post.mq4, or later version)?
Sorry G., did not have time for that today. I was fully occupied working on the basics of CIX 4. It’s not yet mature enough for a public test, just want to post a pic for review to ensure my direction is ok from your point of view. Local testing and debugging will certainly take one more day. It’s not yet completely stable.
You wrote: Here we have a problem… This one should indicate strong/weak currencies (Sentiment) by their location above/below the “0” line.
Current status:
- All basic algorithms are identical to the most recent public release CIX 2.2, only BIS 2013-based index added.
- Default settings changed towards more reactive nature of the CS lines.
- ZigZag algorithm applied to CS lines – see dots in the pic.
- ZigZag end points are basis for trendlines drawn towards a chosen bar offset (orange vertical line in the pic, bar 1 by default, trendlines see left side)
- Currency table added, sorted by trendline slope
- Interface to the EA has to be discussed – this will probably be too CPU intensive for iCustom use (@VlanFx: what’s your opinion about global variables to transfer CS and slope to the EA?)
Discussion:
I reduced the CIX chart to 4 lines for easier viewing on that pic. Please note the time interval between the orange and the green vertical line. EURJPY has a weak downtrend here. Looking at CS lines, JPY is trending slightly up, EUR a little more. So slope alone would be very slightly bullish, but certainly no valid signal, since both lines are travelling north.
The actual trend of the pair is better reflected by the location of EUR and JPY lines below and above zero.
Leaving aside our fundamental discussion about correct application of the index: would you say this is similar to the behaviour you’re suggesting for CS?
s.
A good trader is a realist who wants to grab a chunk from the body of a trend, leaving top- and bottom-fishing to people on an ego trip. (Dr. Alexander Elder)
Call For Coders
It’s still too early to publish my latest CIX full version – there’s some code cleanup to do.
But if anybody would like to backtest our EA prototype based on CIX data, I’ve got something to offer: the Offline Version of CIX 4.3.
It works like this: I equipped the full version with a csv export interface, that dumps timestamps as well as all CIX readings in a comma delimited file. The offline version imports these files (several timeframes and indicator settings to choose from) and displays them on screen – see large pic.
The output buffers: 8 currency specific strength values, and 8 slope values, see small pic.
CS calculation is basically the same as my latest public CIX version, just tuned for higher reactivity. Slope calculation is done by a linear regression over the individual strength lines.
So if anybody is interested to start some backtests based on currency strength and post information about the strategy used and the results, just reply. I can also assist you in implementing the offline CIX in your EA.
I think it would be good to collect some more ideas about how to integrate CS in our strategies on a detailed level, so I’m hoping for some prompts and some interesting results.
CIX full version will be published later, may still take some days from now on.
s.
A good trader is a realist who wants to grab a chunk from the body of a trend, leaving top- and bottom-fishing to people on an ego trip. (Dr. Alexander Elder)
Call For Coders It’s still too early to publish my latest CIX full version – there’s some code cleanup to do. But if anybody would like to backtest our EA prototype based on CIX data, I’ve got something to offer: the Offline Version of CIX 4.3. It works like this: I equipped the full version with a csv export interface, that dumps timestamps as well as all CIX readings in a comma delimited file. The offline version imports these files (several timeframes and indicator settings to choose from) and displays them on screen – see large pic. The output buffers: 8 currency specific strength values, and 8 slope values, see small pic. CS calculation is basically the same as my latest public CIX version, just tuned for higher reactivity. Slope calculation is done by a linear regression over the individual strength lines. So if anybody is interested to start some backtests based on currency strength and post information about the strategy used and the results, just reply. I can also assist you in implementing the offline CIX in your EA. I think it would be good to collect some more ideas about how to integrate CS in our strategies on a detailed level, so I’m hoping for some prompts and some interesting results. CIX full version will be published later, may still take some days from now on. s.
CS should be implemented as already described: by calling a function that returns 1, -1, or 0 (Long, Short, None).
Long signal for currency pair AB is when currency A is above 0 and pointing up, currency B is below 0 and pointing down.
The opposite is for Short.
Other variations returns 0 (None).
Optional addition to the above is a confirmation from LOWER TF CS.
The above is just ONE part of the complete logic. The other two parts are the Volume (VZO or similiar) & ClearMethod (Min. 2 TF’s histogram).
ClearMethod can be enhanced with some ADX method (such as ADXcandle variation).
Implementation of the EA should be by calling the relevant functions:
if (CS()==1 && VZO()==1 && CM()==1)
… go LONG …
G.
Thank you for clarifying:
CS should be implemented as already described: by calling a function that returns 1, -1, or 0 (Long, Short, None).
In my Penguin_CIX.mqh, it looks like this, because I prefer to work with enumerations in such cases:
enum CIX_TRADE { CIX_TRADE_NONE, CIX_TRADE_BUY, CIX_TRADE_SELL };Called like this:
CIX_TRADE cixEntry = getCixEntry(symbolIndex, tf, CIX_INDI_NAME, cixSmoothing, cixTrendDepth, Prefix, Suffix);The rest of my indicator calls is about to be modified in a similar way now, resulting in one-line function calls.
My current logic depends on one instance of CIX and one instance of CLEAR, both in H4. VZO now is about to be implemented in the trading decisions.
When this works successfully to a certain degree in backtest, I would consider to add higher timeframes and switch to forward testing. When forward testing starts, it might be the right time to switch from single pair to multi pair also.
When using VZO, do you just wait until VZO has achieved a certain counter trend value, anticipating a reversal, or do you wait for some VZO reversal to establish?
I’m further considering to use CIX slopes to decide upon dynamic TP, because ADX Main tends to reverse too late for this purpose. The idea is to stop progressing TP when the slope of one of the CIX lines involved turns against us, so the trade can rather hit TP instead of a trailed SL. What’s your opinion about that?
s.
A good trader is a realist who wants to grab a chunk from the body of a trend, leaving top- and bottom-fishing to people on an ego trip. (Dr. Alexander Elder)
A little motivational post regarding backtests with Offline CIX.
This is the result of a 4 months backtest run on EURUSD H4. Simple entry combination of just CIX and CLEAR, no VZO yet.
Purposefully hardly any trade management yet, because I wanted to get an idea about the entry quality. It’s 1 % risk per trade, 1.0 ATR SL and TP, with moving SL to BE when 0.5 ATR profit are achieved. Moving to BE so early causes the high rate of about 50 % of the trades to be closed just above BE.
Partial close disabled for this run, also dynamic TP and SL.
s.
A good trader is a realist who wants to grab a chunk from the body of a trend, leaving top- and bottom-fishing to people on an ego trip. (Dr. Alexander Elder)
Thank you for clarifying:
CS should be implemented as already described: by calling a function that returns 1, -1, or 0 (Long, Short, None).
In my Penguin_CIX.mqh, it looks like this, because I prefer to work with enumerations in such cases:
enum CIX_TRADE { CIX_TRADE_NONE, CIX_TRADE_BUY, CIX_TRADE_SELL };Called like this:
CIX_TRADE cixEntry = getCixEntry(symbolIndex, tf, CIX_INDI_NAME, cixSmoothing, cixTrendDepth, Prefix, Suffix);The rest of my indicator calls is about to be modified in a similar way now, resulting in one-line function calls. My current logic depends on one instance of CIX and one instance of CLEAR, both in H4. VZO now is about to be implemented in the trading decisions. When this works successfully to a certain degree in backtest, I would consider to add higher timeframes and switch to forward testing. When forward testing starts, it might be the right time to switch from single pair to multi pair also. When using VZO, do you just wait until VZO has achieved a certain counter trend value, anticipating a reversal, or do you wait for some VZO reversal to establish? I’m further considering to use CIX slopes to decide upon dynamic TP, because ADX Main tends to reverse too late for this purpose. The idea is to stop progressing TP when the slope of one of the CIX lines involved turns against us, so the trade can rather hit TP instead of a trailed SL. What’s your opinion about that? s.ClearMethod Histogram:
This requires at least 2 TF’s, current and Higher. Current defines Peak or Trough area, Higher TF’s define the longer trend.
Current ClearMethod can be enhanced by combining it with ADX (ADXcandle or SuperADX variations).
VZO:
Defines entry bar, either on Peak or Trough, in the direction of the Higher TF of the ClearMethod. No need for Higher/Lower TF’s for VZO.
The Entry signal is when VZO breaks the range limit (on bar close).
No need to further confirmation, since we have 2 other logics from Price-Action (CM) and Currency-Strength (CM) in order for the EA to enter a trade, with reasonable SL.
Currency-Strength:
Logic already explained and it’s quite simple. The use of 1, -1, 0 or WORDS isn’t critical, but it would be nice if ALL functions return values in the same manner.
From experience with CS, I found it beneficial to consult LOWER TF, to confirm that the direction didn’t move against you. This requires that LOWER TF of CS currency A is pointing UP or flat, and currency B is pointing down or flat. Above/Below 0 doesn’t matter for LOWER TF CS.
Consulting lower TF’s IS NOT a must, just beneficial.
G.
Thank you for yout hints!
Re. enumerations: yes, all trading decisions can be represented by one common enumeration instead of indicator-specific enums. That would be even easier to read and maintain. I will modify my functions in that way.
Re. higher and lower TFs: I fully support your view. Just at the moment I would like to stay with one TF to further enable for backtesting as long as possible, mostly to experiment with modified Dynamic TP and SL until a satisfying performance can be achieved when using those functions. This is not yet the case.
CS @ LOWER TFs: yes, I also have this in mind for the future. It would certainly lead to even more precise entries.
Re. VZO, let me describe my current view on it: Let’s say we define ranging limits at +40, -40 (to be discussed and tested). So if VZO breaks -40 upwards, this is a buy signal, and when it breaks +40 downwards, we consider it a sell signal.
s.
A good trader is a realist who wants to grab a chunk from the body of a trend, leaving top- and bottom-fishing to people on an ego trip. (Dr. Alexander Elder)
Thank you for yout hints! Re. enumerations: yes, all trading decisions can be represented by one common enumeration instead of indicator-specific enums. That would be even easier to read and maintain. I will modify my functions in that way. Re. higher and lower TFs: I fully support your view. Just at the moment I would like to stay with one TF to further enable for backtesting as long as possible, mostly to experiment with modified Dynamic TP and SL until a satisfying performance can be achieved when using those functions. This is not yet the case. CS @ LOWER TFs: yes, I also have this in mind for the future. It would certainly lead to even more precise entries. Re. VZO, let me describe my current view on it: Let’s say we define ranging limits at +40, -40 (to be discussed and tested). So if VZO breaks -40 upwards, this is a buy signal, and when it breaks +40 downwards, we consider it a sell signal. s.
Regarding VZO:
I think you get better entries playing with VZO parameters and range (try VZO 6, range 55 & -55), thus trading just the breakout of the range (i.e. on lower range breakout – signal Long, Upper range breakout – signal short.)
Your method will also work Ok (a bit lagging), without fine-tuning VZO.
G.
Thanks for this:
I think you get better entries playing with VZO parameters and range (try VZO 6, range 55 & -55), thus trading just the breakout of the range (i.e. on lower range breakout – signal Long, Upper range breakout – signal short.)
I also experimented with VZO settings different than the defaults suggested in Khalil’s article. I found that VZO 4 … 8, limits around 60 looked interesting, so very much similar to your suggestion.
I will certainly try your breakout, anticipating the reversal! I’m curious how the exact timing will meet CIX and CLEAR.
s.
A good trader is a realist who wants to grab a chunk from the body of a trend, leaving top- and bottom-fishing to people on an ego trip. (Dr. Alexander Elder)
plugged the str weights into another one of my indices code. it should show the same direction as gtarrow aside from being squished from normalization.
if anyone wants to see the code give a shout.
“I believe the very best money is made at the market turns. Everyone says you get killed trying to pick tops and bottoms and you make all
your money by playing the trend in the middle. Well for twelve years I have been missing the meat in the middle but I have made a lot of
money at tops and bottoms.”
– Paul Tudor Jonesplugged the str weights into another one of my indices code. it should show the same direction as gtarrow aside from being squished from normalization. if anyone wants to see the code give a shout.
Try 3 & 5 sma’s on the USDx. More responsive.
I’m interested in your code for comparison. SHOUT!!
G.
its a W.I.P. so the code is not finished. i have 3 additional indices added to see what it looks like. being normalized means everything is relative to normalization length so i was thinking of adding a rank based on how much it has moved and the direction it has moved from the last bar.
“I believe the very best money is made at the market turns. Everyone says you get killed trying to pick tops and bottoms and you make all
your money by playing the trend in the middle. Well for twelve years I have been missing the meat in the middle but I have made a lot of
money at tops and bottoms.”
– Paul Tudor Jonesif anyone wants to see the code give a shout.
I’m interested in your code for comparison. SHOUT!!
You beat me: thanks for shouting, G. and thanks for sharing, L. !
s.
A good trader is a realist who wants to grab a chunk from the body of a trend, leaving top- and bottom-fishing to people on an ego trip. (Dr. Alexander Elder)
Call For Coders It’s still too early to publish my latest CIX full version – there’s some code cleanup to do. But if anybody would like to backtest our EA prototype based on CIX data, I’ve got something to offer: the Offline Version of CIX 4.3. It works like this: I equipped the full version with a csv export interface, that dumps timestamps as well as all CIX readings in a comma delimited file. The offline version imports these files (several timeframes and indicator settings to choose from) and displays them on screen – see large pic. The output buffers: 8 currency specific strength values, and 8 slope values, see small pic. CS calculation is basically the same as my latest public CIX version, just tuned for higher reactivity. Slope calculation is done by a linear regression over the individual strength lines. So if anybody is interested to start some backtests based on currency strength and post information about the strategy used and the results, just reply. I can also assist you in implementing the offline CIX in your EA. I think it would be good to collect some more ideas about how to integrate CS in our strategies on a detailed level, so I’m hoping for some prompts and some interesting results. CIX full version will be published later, may still take some days from now on. s.
I can do a backtest I got a lot of free time.TMНичто не ново под луной:
Что было, то и будет.
Я завершаю круг и - вновь
Готов бежать по кругу...I’d like to come back to Ron Black’s CLEAR Method indicator much discussed here as well as in the main thread. Recently I stumbled upon a video showing an interesting approach to utilize CLEAR in a more sophisticated way than proposed here before.
Note: that YouTube video is posted by a company, linking to their commercial website. Please don’t take my posting here as a recommendation to buy any of their products! It’s just a proposal to think about CLEAR indicators in a different way than discussed here at Penguin’s before.
A good trader is a realist who wants to grab a chunk from the body of a trend, leaving top- and bottom-fishing to people on an ego trip. (Dr. Alexander Elder)
- You must be logged in to reply to this topic.