pedro01
Guest
- Messages
- 1,058
- Likes
- 149
Great - I think you & Naeem should team up as he's been looking too !
Great - I think you & Naeem should team up as he's been looking too !
Will do - I'm going to need some help!! I plan to start at the week-end once I've completed the set-up on my PC
rj,
What are you looking to test? I have a few indicators that I've been watching but I've had data problems so I think I’m going to be dropping IB as a data source and go with IQ feed. If you need some help with the Ehler indicators let me know.
I've got very basic coding experience but I am trying hard to code up the Fisher RVI in a similar way to A_B's brilliant Macci MTF indicator. Most of the Ehler indicators work differently to the Macci so bear that in mind.
If you’re looking to test something may I suggest you take a look at the Inverse Fisher Cyber Cycle. I’ll post the code but I have to tell you I can’t get it to work on my system for some reason. The code has been verified but I keep getting the following message: “ Floating point numerical overload”. This indicator works in a similar way to the Macci so it will be easier to test from the start.
Naeem
Last edited:
I don't know anything about TS, but this is possibly floating point overflow. Maybe dividing by a very small number, or probably less likely multiplying two very large numbers.
At one time I translated some of the Ehlers stuff (from his web site) into Java. The code had some (logical) problems, including divide by zero. Bit sloppy actually.
Hi Naeem
The code you posted works fine for me on all the stock charts I tried but I do get the error you refer to when applied to the INDU chart. Maybe, along the lines of what DC posted, its because the absolute value of INDU is so much higher than a typical stock???
Cheers
Steve
pedro01
Guest
- Messages
- 1,058
- Likes
- 149
Hi guys
I terms of testing - what we need to look for is thie ability of the indicator to call turns on the $INDU when analyzed on multiple time frames.
Note that according to the MESA site, the Inverse Fisher is not adaptive to volatility. That doesn't mean it won't work - just that we need to treat it with care.
https://mesasoftware.com/indicators.htm
I'd be looking at the oscillators & the filters.
Do we really have the code for all of this ? Or just the oscillators ?
Last edited:
Corrected code:
Code:
Inputs: Price((H+L)/2),
alpha(.07);
Vars: Smooth(0),
Cycle(0),
ICycle(0);
Smooth = (Price + 2*Price[1] + 2*Price[2] + Price[3])/6;
Cycle = (1 - .5*alpha)*(1 - .5*alpha)*(Smooth - 2*Smooth[1] + Smooth[2]) + 2*(1 -
alpha)*Cycle[1] - (1 - alpha)*(1 - alpha)*Cycle[2];
If currentbar < 7 then Cycle = (Price - 2*Price[1] + Price[2]) / 4;
If AbsValue(Cycle) > 100 Then ICycle = ICycle[1]
Else ICycle = (ExpValue(2*Cycle) - 1) / (ExpValue(2*Cycle) + 1);
Plot1(ICycle, "Cycle");
Plot2(0.5, "Sell Ref");
Plot3(-0.5, "Buy Ref");A_B,
Cheers for your help there mate. Problem is still there though. I'm still getting the "Floating point numercal" message.
Steve,
I think you're spot on with that. Could you or anyone test out A_B's new code. The problem might not be the code but my set up or something.
Pete,
We only need the oscillators. The oscillators advertised on Mesa have not been treated with the extra filters. The codes me and lwebb have posted are based on his latest work as far as I'm aware.
Cheers
Naeem.
Cheers for your help there mate. Problem is still there though. I'm still getting the "Floating point numercal" message.
Steve,
I think you're spot on with that. Could you or anyone test out A_B's new code. The problem might not be the code but my set up or something.
Pete,
We only need the oscillators. The oscillators advertised on Mesa have not been treated with the extra filters. The codes me and lwebb have posted are based on his latest work as far as I'm aware.
Cheers
Naeem.
It runs in my system (see attached), but is that how the INDU chart should look ?
Charlton
Attachments
Naeem,
The code I posted works on MultiChart.
The problem is that ExpValue(2*Cycle) is too high. I limited AbsValue(Cycle) to 100. Try to limit AbsValue(Cycle) to 10:
Code:
Inputs: Price((H+L)/2),
alpha(.07);
Vars: Smooth(0),
Cycle(0),
ICycle(0);
Smooth = (Price + 2*Price[1] + 2*Price[2] + Price[3])/6;
Cycle = (1 - .5*alpha)*(1 - .5*alpha)*(Smooth - 2*Smooth[1] + Smooth[2]) + 2*(1 -
alpha)*Cycle[1] - (1 - alpha)*(1 - alpha)*Cycle[2];
If currentbar < 7 then Cycle = (Price - 2*Price[1] + Price[2]) / 4;
If AbsValue(Cycle) > 10 Then ICycle = ICycle[1]
Else ICycle = (ExpValue(2*Cycle) - 1) / (ExpValue(2*Cycle) + 1);
Plot1(ICycle, "Cycle");
Plot2(0.5, "Sell Ref");
Plot3(-0.5, "Buy Ref");Or you can divide the price by some value, i.e.
Code:
Inputs: Price((H+L)/20),
alpha(.07);
Last edited:
pedro01
Guest
- Messages
- 1,058
- Likes
- 149
Naeem
In the last seminar, Iraj mentioned one of the problems with the cycles is that they weren't adaptive to volatility - so in different market conditions they don't adjust.
That was why I thought the filters were important...
Any thoughts ?
Pete
This indicator highly depends on price. I think that Inverse Fisher Transform is not suitable in this case.
Hi Pete
I agree and think that 'adapting to volatility' is important here. A lot of potentialy relevant work in this area is published by Tushar Chande. For example, he explains and provides TS code to produce an Adaptive Stochastic Oscillator which 'adapts' based upon volatility specifically.
I can post it here if you want me to?
Cheers
Steve
I’ve enclosed a pdf of the various oscillators that we could test. Most people know about the Inverse Fisher and standard oscillators like the Cyber Cycle, CG and RVI. I’ve enclosed details of how Ehlers’ latest oscillators work. This is a summary and not an exhaustive technical guide. If you would like to delve deeper into how Ehler designes his oscillators then I recommend you read the following John Ehler books:
Rocket Science for Traders and
Cybernetic Analysis for Stocks and Futures
The oscillators in the pdf summary can be found in Cybernetic Analysis for Stocks and Futures.
Pete,
Take a look at the pdf before reading on.
There are three important components we need to measure if we want to extract the Dominant Cycle (DC) from price within a specific TF:
Frequency of various cycles,
Amplitude of various cycles
Phase within various cycles.
From my understanding, the Adaptive Oscillators designed by Ehler factor in potential changes in price volatility because the measurement of the varying amplitudes of market cycles is a core component when measuring the DC. When we test this particular oscillator we will need to bear in mind that there will be no Macci +100 -100 OB/OS thresholds to work from anymore.
The obvious question then becomes; if the Adaptive Oscillator is adapting to the DC within each TF and we don’t have OB/OS conditions as a barometer how will we measure the convergence of all DC’s within each TF equivalent to the Macci’s +100 -100 OB/OS thresholds? In order to stay consist with MLT a possible suggestion could be to measure the time difference between the trigger signals within each TF.
Buy Signal Example (Reverse for Sell Signals):
1min = Trigger @15:00pm GMT
3min = Trigger @15:05pm GMT
5min = Trigger @15:07pm GMT
The smaller the time difference between each Buy trigger signal within each TF the higher the probability that two conditions are true:
A) All lower TF DC's have converged and are in OS conditions
B) All lower TF DC's are turning within the OS conditions
We will obviously need to find the optimal Allowable Time Difference between trigger signals in all TF’s if we are to pursue this method. That doesn’t sound easy.
Charlton & A_B,
An oscillator calculated using the Inverse Fisher Transform (IF) should not look like the jpeg posted. I think A_B could be right about the Cyber Cycle version of the IF applied to INDU. The IF RSI works fine on the INDU charts so that could be a possible indicator to test. I've enclosed the ELA for that below.
Steve that would be great.
Cheers
Naeem
Rocket Science for Traders and
Cybernetic Analysis for Stocks and Futures
The oscillators in the pdf summary can be found in Cybernetic Analysis for Stocks and Futures.
Pete,
Take a look at the pdf before reading on.
There are three important components we need to measure if we want to extract the Dominant Cycle (DC) from price within a specific TF:
Frequency of various cycles,
Amplitude of various cycles
Phase within various cycles.
From my understanding, the Adaptive Oscillators designed by Ehler factor in potential changes in price volatility because the measurement of the varying amplitudes of market cycles is a core component when measuring the DC. When we test this particular oscillator we will need to bear in mind that there will be no Macci +100 -100 OB/OS thresholds to work from anymore.
The obvious question then becomes; if the Adaptive Oscillator is adapting to the DC within each TF and we don’t have OB/OS conditions as a barometer how will we measure the convergence of all DC’s within each TF equivalent to the Macci’s +100 -100 OB/OS thresholds? In order to stay consist with MLT a possible suggestion could be to measure the time difference between the trigger signals within each TF.
Buy Signal Example (Reverse for Sell Signals):
1min = Trigger @15:00pm GMT
3min = Trigger @15:05pm GMT
5min = Trigger @15:07pm GMT
The smaller the time difference between each Buy trigger signal within each TF the higher the probability that two conditions are true:
A) All lower TF DC's have converged and are in OS conditions
B) All lower TF DC's are turning within the OS conditions
We will obviously need to find the optimal Allowable Time Difference between trigger signals in all TF’s if we are to pursue this method. That doesn’t sound easy.
Charlton & A_B,
An oscillator calculated using the Inverse Fisher Transform (IF) should not look like the jpeg posted. I think A_B could be right about the Cyber Cycle version of the IF applied to INDU. The IF RSI works fine on the INDU charts so that could be a possible indicator to test. I've enclosed the ELA for that below.
Steve that would be great.
Cheers
Naeem
Last edited:
leovirgo
Senior member
- Messages
- 3,161
- Likes
- 156
I have been using Inverse Fisher for a few years and I find that Inverse Fisher signals are more reliable than MACCI in terms of getting the direction right. However MACCI OB/OS seems to capture the exact reversal points. So I use both of them in different TF to capture the move. For example, I use 10min IF to determine the direction and then use MACCI at lower TF to determine entry.
Say 10min IF is entering OB zone or at OB level, then I use 1min MACCI OB to enter the reversal short. This method often captures the exact reversal points.
Or 10min IF crosses down Signal line then enter the short trade with 1min MACCI OB or turning down from highzone. This method offers better directional trades but the entry price is often retested before going further in the direcion of the trade.
So to me the question is
1. Are we after the exact reversal points? or
2. Are we after a better directional trade? or
3. we want both.. lol...
Say 10min IF is entering OB zone or at OB level, then I use 1min MACCI OB to enter the reversal short. This method often captures the exact reversal points.
Or 10min IF crosses down Signal line then enter the short trade with 1min MACCI OB or turning down from highzone. This method offers better directional trades but the entry price is often retested before going further in the direcion of the trade.
So to me the question is
1. Are we after the exact reversal points? or
2. Are we after a better directional trade? or
3. we want both.. lol...
Hi
As suggested in an earlier post, the Chande Adaptive Stochastic Oscillator code is as follows. I suppose that the point here is not so much that this particular oscillator is necessarily the one to use but is shows a way that any oscillator can be made adaptive to recent volatility.
{Tushar Chande: Adaptive Stochastic Oscillator}
vars: v1(0), v2(0), v3(0), v4(0);
vars: lenmax(28), lenmin(7), currlen(0);
vars: hh(0), ll(0), stoch(0), stochma(0);
{calculate 20 day std deviation and its 20 day range}
v1 = stddev(c,20);
v2 = highest(v1,20);
v3 = lowest(v1,20);
{create v4: stochastic osc for 20 day std dev}
{if v1=v2 (highest level) => v4 = 1; if v1=v3 (lowest level) => v4=0}
if (v2-v3) > 0 then v4 = ((v1 - v3)/(v2 - v3)) Else v4=0;
{calculate effective current length: if v4=1 then length = minimum}
currlen = Intportion(lenmin + (lenmax-lenmin)*(1-v4));
{calculate stochastic osc and its 3-day exponential average}
hh = highest(h,currlen);
ll = lowest(l,currlen);
if (hh-ll) > 0 then Stoch=((close-ll)/(hh-ll))*100;
if currentbar=1 then Stochma = 0 Else
Stochma = 0.5*stoch + 0.5*stochma[1];
{plot data}
plot1(stoch,"adapt_stoch");
plot2(slowk(18),"stochma");
Plot3(80,"hi_ref");
plot4(20,"low_ref");
{end of code}
Cheers
Steve
As suggested in an earlier post, the Chande Adaptive Stochastic Oscillator code is as follows. I suppose that the point here is not so much that this particular oscillator is necessarily the one to use but is shows a way that any oscillator can be made adaptive to recent volatility.
{Tushar Chande: Adaptive Stochastic Oscillator}
vars: v1(0), v2(0), v3(0), v4(0);
vars: lenmax(28), lenmin(7), currlen(0);
vars: hh(0), ll(0), stoch(0), stochma(0);
{calculate 20 day std deviation and its 20 day range}
v1 = stddev(c,20);
v2 = highest(v1,20);
v3 = lowest(v1,20);
{create v4: stochastic osc for 20 day std dev}
{if v1=v2 (highest level) => v4 = 1; if v1=v3 (lowest level) => v4=0}
if (v2-v3) > 0 then v4 = ((v1 - v3)/(v2 - v3)) Else v4=0;
{calculate effective current length: if v4=1 then length = minimum}
currlen = Intportion(lenmin + (lenmax-lenmin)*(1-v4));
{calculate stochastic osc and its 3-day exponential average}
hh = highest(h,currlen);
ll = lowest(l,currlen);
if (hh-ll) > 0 then Stoch=((close-ll)/(hh-ll))*100;
if currentbar=1 then Stochma = 0 Else
Stochma = 0.5*stoch + 0.5*stochma[1];
{plot data}
plot1(stoch,"adapt_stoch");
plot2(slowk(18),"stochma");
Plot3(80,"hi_ref");
plot4(20,"low_ref");
{end of code}
Cheers
Steve
pedro01
Guest
- Messages
- 1,058
- Likes
- 149
Naeem
I think we are on the right track. Ultimately - we can code anything.
So :
- If the indicator is better
- it doesn't have the traditional OB/OS fixed zones
- it does have a way to give off signals
We should use it & adapt our code to fit the best indicator. We are only using OB/OS zones right now because that is what MACCi gives us.
Cheers
Pete
pedro01
Guest
- Messages
- 1,058
- Likes
- 149
sure - post it up...
pedro01
Guest
- Messages
- 1,058
- Likes
- 149
To start with - we are after something simple that works. Let's just define the right indicator, the best signal, even if that signal only occurs once or twice a week but is very reliable, that's what we need to code first in my opinion. Once that's up & running we can get more sophisticated.
It's a balance between simplicity & reliability we need to aim for at first.
Let's not run before we walk here.
Similar threads
