Attachments: Mean Reversion TMA

Mean Reversion TMA

#1
First Post: Edited 5:22am Jul 22, 2019 5:07am | Edited 5:22am

SwingMan
Joined Aug 2007 | Status: Trader | 2,725 Posts

Many thanks for posting ideas and programs for a nice mean reversion system to two members:
@robots4me and @bishopdotun!

MM (Money Maker) Detective Indicator: https://www.forexfactory.com/showthread.php?t=911239
and
Custom Indicators for TMA-ADR Reversal Trading and BBMA: https://www.forexfactory.com/showthread.php?t=920925

RBA system: https://www.forexfactory.com/showthread.php?t=930790

----------------

I do not want to burden their threads with my comments, so I opened a new thread, and who has time to read and apply some things is OK.
(I will occasionally post a few more analyzes if I have programmed additional indicators).

#2
Edited 6:22am Jul 22, 2019 5:16am | Edited 6:22am

SwingMan
Joined Aug 2007 | Status: Trader | 2,725 Posts

The main thing in the R4M-BD system is the mean reversion to the nice moving average - the TMA.

For lazy traders like me, it's unclear how I should take a proper value for ADR percent to calculate the bands.
1. - For this reason I have calculate the standard deviations of the High/Lows to the TMA and compared with the ATR, period=20 bars.
(It is possible that my StdDev calculation is not correct or the best, so I still have some tests to do, but the idea remains valid.)

Attached Image

2. - One problem mentioned was the repainting of the bands for the last bars, so I also have a comparison of the bands for TMA on median and on open prices. The differences would not be too big ... (Dotted lines are the ATR bands)

Attached Image (click to enlarge)

Click to Enlarge

Name: MeanReversion-2.PNG
Size: 82 KB

3. - It was noticed that sometimes there are too many signals. I think that the system is stable enough not to trade W1/M15.
That's why I would take for example:
MN1/D1, W1/H4, H4/M15, H1/M5, M15/M1

#3
Jul 22, 2019 7:01am Jul 22, 2019 7:01am

bishopdotun
Joined Jun 2011 | Status: Trader | 840 Posts

An interesting approach here @SM. So the % of ADR could be estimated dynamically and applied based on pair STD.

In this case, can you make it fixed and what impact will it have? This opens up an entirely new path and I will follow closely on how it impacts trading outcomes.

Referring to Std from the TMA line, I assume this is default 1 STD.

Was the TMA centre line drawn from W1 or D1?

#4
Jul 22, 2019 7:14am Jul 22, 2019 7:14am

bishopdotun
Joined Jun 2011 | Status: Trader | 840 Posts

Quoting SwingMan

Disliked

The main thing in the R4M-BD system is the mean reversion to the nice moving average - the TMA. For lazy traders like me, it's unclear how I should take a proper value for ADR percent to calculate the bands. 1. - For this reason I have calculate the standard deviations of the High/Lows to the TMA and compared with the ATR, period=20 bars. (It is possible that my StdDev calculation is not correct or the best, so I still have some tests to do, but the idea remains valid.) {image} 2. - One problem mentioned was the repainting of the bands for the last...

Ignored

Can you upload STD for W1 and H4. I would love to compare their ratios to ADR.

#5
Edited 8:33am Jul 22, 2019 7:58am | Edited 8:33am

SwingMan
Joined Aug 2007 | Status: Trader | 2,725 Posts

Quoting bishopdotun

Disliked

An interesting approach here @SM. So the % of ADR could be estimated dynamically and applied based on pair STD. In this case, can you make it fixed and what impact will it have? This opens up an entirely new path and I will follow closely on how it impacts trading outcomes. Referring to Std from the TMA line, I assume this is default 1 STD. Was the TMA centre line drawn from W1 or D1?

Ignored

- In the title one can see the price "Median" and the Deviation factor = "1.0"
I suppose that values of 0.5 - 0.8 would be ok.

It will not have a major impact on trading, but you do not have to calculate or optimize yourself the ADR fractions.

- In the charts, the TMA and bands are drawn for the corresponding time frame (in the charts is D1). Later in the trading TFs they can be taken as MTF lines.

As I wrote, I still have to check my StdDev calculation and then I post the indicators.

#6
Jul 22, 2019 10:29am Jul 22, 2019 10:29am

robots4me
Joined Dec 2017 | Status: Trader | 4,378 Posts

Quoting SwingMan

Disliked

The main thing in the R4M-BD system is the mean reversion to the nice moving average - the TMA. For lazy traders like me, it's unclear how I should take a proper value for ADR percent to calculate the bands. 1. - For this reason I have calculate the standard deviations of the High/Lows to the TMA and compared with the ATR, period=20 bars. (It is possible that my StdDev calculation is not correct or the best, so I still have some tests to do, but the idea remains valid.) {image} 2. - One problem mentioned was the repainting of the bands for the last...

Ignored

@SwingMan -- interesting idea.

In looking at the Standard Deviation data the ATR column looks to me like the ADR -- or very similar, right?

And then looking at EURUSD the ADR is 458 points -- which makes sense. But the Std Deviation is 358 points? Isn't that a huge variation?

#7
Jul 22, 2019 10:54am Jul 22, 2019 10:54am

SwingMan
Joined Aug 2007 | Status: Trader | 2,725 Posts

Quoting robots4me

Disliked

{quote} @SwingMan -- interesting idea. In looking at the Standard Deviation data the ATR column looks to me like the ADR -- or very similar, right? And then looking at EURUSD the ADR is 458 points -- which makes sense. But the Std Deviation is 358 points? Isn't that a huge variation?

Ignored

- ADR is calculated as the average of the differences High/Low

- ATR is calculated as the average of the differences High/Close etc. (see the definition ...), and I used "iATR()".
A fraction of this value is added or subtracted from the TMA (dotted lines).

- For me it seems logical to calculate directly the average of the distances between TMA and High or Low, which is greater.
Adding or subtracting this value from the TMA, gives me a closer look at the volatility.

But because the distances are positive or negative, one calculates the Mean Deviation with squares of the values (then the average).
I did that in the existing indicators.

There is even one more step to calculate the standard deviation, but the StdDev values seemed a bit strange to me, and I still have to think about whether they might be right. That will take some time (to think about, and that hurts sometimes), or a mathematician can help us.

#8
Last Post: Jul 25, 2019 4:43am Jul 25, 2019 4:43am

SwingMan
Joined Aug 2007 | Status: Trader | 2,725 Posts

For analysis purposes I programmed a small help indicator.

Attached File(s)

smMeanReversion TMAcentered_v3.0.ex4 24 KB | 518 downloads

How to use it:
- One can calculate TMA for the Median or Open price
- For Shift=1 the TMA is shifted 1 bar to the right, and the calculated values at the last bar remain stable.

Attached Image (click to enlarge)

Click to Enlarge

Name: MeanReversion-2charts.PNG
Size: 96 KB

- One can get a similar line if one calculates the TMA with the Open price and Shift=0.

- For the bands one can use the ATR, and as an alternative the difference between TMA and Highs/Lows.

Inserted Code

         ib=iBar+i;
         if(Deviation_Price==HighLow)
           {
            bufDeviation[i]=MathMax(MathAbs(High[ib]-bufAverage[ib]),MathAbs(Low[ib]-bufAverage[ib]));
           }

- A variant is to take the distance between TMA and Open/Close. For me, this variant would be preferable, because long wicks have no influence.

Inserted Code

         if(Deviation_Price==OpenClose)
           {
            bufDeviation[i]=MathMax(MathAbs(Open[ib]-bufAverage[ib]),MathAbs(Close[ib]-bufAverage[ib]));
           }

and

Inserted Code

      for(int i=0;i<numberBars;i++)
        {
         sum+=MathAbs(bufDeviation[i]);
        }
      medSum=sum/numberBars;

If you take half this distance,

Inserted Code

      if(Draw_TMA_Bands==true)
        {
         bandUP[iBar]=bufAverage[iBar]+medSum*TMA_Bands_Deviation;
         bandDN[iBar]=bufAverage[iBar]-medSum*TMA_Bands_Deviation;
        }

you get approximately the 20% value of ATR which one finds in the charts of @bishopdotun and @robots4me.

With this value, I dont see in the visual analysis a great influence whether one calculates the TMA with Median and Shift=1, or with Opens.

- It would be important for the trading algorithm to consider the slope of the TMA!

Attached Image (click to enlarge)

Click to Enlarge

Name: MeanReversion-4charts.PNG
Size: 78 KB

Trading Discussion
/
Mean Reversion TMA
Reply to Thread

0 traders viewing now

Options

Similar Threads

Mean Reversion TMA