MurreyMath

Quote from wd78:



Exactly. If that were not the case then one could easily demonstrate that MM was THE best trading system, which has not been done.
More...

Yeah. The 8 ball gives it a good run for ya money.

Hey, Oddi! I thought you had gone,man! Or that they had caught ya!!! :-)

I'm getting a 'sell the s+P below 814.5', 'buy it over 834' feeling.

What does the great Murrey Line Method predict?
 
Quote from pretzel:

wd78,

This script produces the same lines as in your chart but dont know if it will still do for the the next frame.

pretzel
More...

Looks OK. Suggest, however, checking it out with other time frames and a variety of stocks with different volatilities. That maze should find any problems, if there are any. I have not looked at your code, but if you are concerned about stability between frames then just watch it for a while. However, my experience is that changing volatility and time frame should be enough, absent anything unusual in the code.
 
Quote from wd78:
Have you actually seen a program that is coded for MM produce correct MMLs one day and not the next? I routinely see MMLs on two programs (Tradestation and AmiBroker) and have never seen a variation from MM's software. Others have made your claim, but so far nobody has demonstrated what their claim is valid. Your 25% claim, I am sure, is incorrect for a correctly coded program since that would easily show up. Bottom line is that day after day, these program duplicate the MMLs in MM's software. Of course, whether MM is the best way to go with or without other supportive analysis, etc. is another matter. Until hard evidence is presented, claims that MMLs cannot be duplicated in other software are not meaningful. All it takes is a couple of gifs to prove that the coding is not equivalent. Enuf ced.
More...

If you haven't seen discrepencies, then you're not looking at enough different price ranges. As I said, the Kruzel algorithm WILL produce the correct #s part of the time, but it will always produce the wrong #s for a bunch of different price ranges. If you're looking at mostly daily charts for a handful of symbols you might not notice it because the price ranges don't change enough or don't represent a large sample of diverse price ranges. It's more obvious when comparing dozens of symbols using several different frame sizes and different intraday timeframes throughout the day for a few days - then you end up seeing thousands of different price ranges to compare the results over.

If you don't believe the error rate I noted, that's up to you, believe what you want - but you missed the actual message. The Kruzel algorithm is wrong 100% of the time when presented certain price ranges - the 25%+ error rate came from hitting those price range conditions (in which the algorithm fails) in about 25% of the cases. It's possible that if it was done again over a different set of days (and thus a different sample of price ranges) it might be 10% or it might be 50%. Frankly after seeing that it wasn't reliable, I didn't bother trying it out over a whole year's worth of intraday data to see what would happen.

If you're looking for a example of an error - compare the WL and TS charts that someone posted earlier at http://www.elitetrader.com/vb/attachment.php?s=&postid=214104 and http://www.elitetrader.com/vb/attachment.php?s=&postid=214109 to the one attached to this message.

Same day (3/5), same 64 bar frame setting - but the WL and TS #s are wrong. Someone else previously posted a chart using eSignal that was noted at the time as having the wrong #s (had nothing to do with colors as the person later suggested - the #s were just wrong).
 

Attachments

Quote from ArchAngel:



If you haven't seen discrepencies, then you're not looking at enough different price ranges. As I said, the Kruzel algorithm WILL produce the correct #s part of the time, but it will always produce the wrong #s for a bunch of different price ranges. If you're looking at mostly daily charts for a handful of symbols you might not notice it because the price ranges don't change enough or don't represent a large sample of diverse price ranges. It's more obvious when comparing dozens of symbols using several different frame sizes and different intraday timeframes throughout the day for a few days - then you end up seeing thousands of different price ranges to compare the results over.

If you don't believe the error rate I noted, that's up to you, believe what you want - but you missed the actual message. The Kruzel algorithm is wrong 100% of the time when presented certain price ranges - the 25%+ error rate came from hitting those price range conditions (in which the algorithm fails) in about 25% of the cases. It's possible that if it was done again over a different set of days (and thus a different sample of price ranges) it might be 10% or it might be 50%. Frankly after seeing that it wasn't reliable, I didn't bother trying it out over a whole year's worth of intraday data to see what would happen.

If you're looking for a example of an error - compare the WL and TS charts that someone posted earlier at http://www.elitetrader.com/vb/attachment.php?s=&postid=214104 and http://www.elitetrader.com/vb/attachment.php?s=&postid=214109 to the one attached to this message.

Same day (3/5), same 64 bar frame setting - but the WL and TS #s are wrong. Someone else previously posted a chart using eSignal that was noted at the time as having the wrong #s (had nothing to do with colors as the person later suggested - the #s were just wrong).
More...

You are behind the times. I pointed out those errors yesterday, and couple of messages ago the prezel noted that the code was corrected. As I noted just above your message, the MMLs are now correct.
 
Quote from ArchAngel:

Here's what it should look like (uses 3/5 as the last bar like WL did).
More...

Yes, I agree that Pretzel has a coding error, which is why I suggested he check other stocks and volatilities before assuming everything was OK. However, that does not mean that a correctly coded program will not produce the same result. For example, the attached shows the correct MMLs in AmiBroker. And if it can be coded correctly in one program it can be coded correctly in another.
 

Attachments

Quote from Wappers2:



Yeah. The 8 ball gives it a good run for ya money.

Hey, Oddi! I thought you had gone,man! Or that they had caught ya!!! :-)

I'm getting a 'sell the s+P below 814.5', 'buy it over 834' feeling.

What does the great Murrey Line Method predict?
More...

Hi Wappers, have you made it to Stockholm yet???

Long yesterday at 827, is what the system says.
 
You must log in or register to reply here.
Back
Top