DMI/ADX consistently works

Quote from WarEagle:


I am working on the TS code for it (the built in version of DMI does not plot the actual DMI (DX) (only the DM+/- and ADX). I am running into a few problems in the calculation results. The plot looks the same as nyc's, but the actual values for DX are different, causing them to have different cross points with the DM lines. I'll post what I have soon and maybe someone can tell me where the differences are.
More...

That's commendable, but I suggest that those who are interested play with whatever ADX/DMIs come with their software to determine what has potential and what doesn't. Whatever is of greatest benefit to the greatest number of people will generate the most discussion.

One area to investigate might be divergences between the ADX and price and/or divergences between DM+ and DM- and price. Crosses above the DMs may not be the best trades unless one also knows how to recognize trend exhaustion.
 
If anyone wants to discuss other ways to use ADX/DMI then that is fine. For the sake of the current discussion I wanted to see if I could duplicate nyc's indicator so that we could be comparing apples to apples. It has always interested me how much reliance many traders put on an indicator (especially one that is working for them) but another trader can use it with a different platform and/or data feed and get completely different results. Its actually a little scary but I have seen it happen over and over.

Ok, so here is where I am at so far...

I plotted the DX along with the DM lines to get it as close to nyc's 9/17 CSCO chart as possible (see attached charts). I tried to alter the indicator to start from the beginning of the day (ignoring data from 9/16), but couldn't get the code to work, so for the sake of comparison, we have to ignore the first 15 minutes after the open when the plots would be different. But after that, with the same lookback period of 15 the values should be the same.

My DM- and DM+ lines are pretty much the same as his. Its the DX line that is different. The general shape of the DX line is the same on both charts, but the actual values that are plotted are different. To double check this I looked up the formula for DX, which is:

DX = |DM+ - DM-| / |DM+ + DM-|

where || is the absolute value.

So, to double check this for the plot I plugged in the numbers from our charts. On my chart at the 10:48ET bar (the white line) the DM+ is .1655 and the DM- is .2291. This also lines up with the DM numbers on nyc's chart (#4 circle). So...

DX = |.1655 - .2291| / |.1655 + .2291| = .1612

That is what is plotted as DX on my chart (see data window on the chart). nyc's DX is approximately .3 at that time.

So, its close at times, but not the same formula for DX. Initially I thought it might be a scaling issue on the chart (which would really make it amazing that it worked since the crosses would be random based on the vertical scale), but if you look at the peaks in DX that occur at 10:58ET (9:58 on my chart) and 11:32ET, on nyc's chart they are approximately the same value whereas on my chart the second peak is higher than the first one. So it must be in the formula.

Any math whizzes out there that might be able to figure out the difference? Maybe none of this matters (certainly not to nyc since it is working for him), but its fun to try and figure out.

And in reply to the guy that thinks nyc is spamming, all I can say is that I have spoken with him over pm a few times to try and figure this out and he answered my questions without once pushing his broker or their software. I hope he will hang around and not let the negativity get to him.
 

Attachments

Thanks for the work WarEagle. I do want to figure out how speedtrader is calculating that ADX...which is some sort of an DX obviously, because I did download the trial and was very impressed with nyc's methods. Hope he does keep posting here.
 
some things i noticed warlord...you are using 2-min data off of even numbered minutes: 8:30, 8:32, 8:34....nyc's chart uses odd number minutes for the 2 minute data, notice how the tick marks on the bottom of his chart are in 2 min segments, but the peaks and troughs on his DX line land in between the tick marks...telling me that it's using those odd-numbered minutes.


also, bar data needs to be exactly the same for dm+ dm- to be the same, since the formula takes the avg of all positive/negative changes in the last X bars. i noticed that on some of his bars, he's got crosses (no change), where your data has a slight change. this is a minor difference though, and im not sure if it can explain the large discrepancy in some of the dx values.

good work though! hope it can be figured out.
 
simstimlord, (just kidding :) )

I also noticed the difference in bar label, but I think that is just a difference in how the software displays the time lables and not in the data itself. If you look closely, the price bars are identical except for a few that differ on the close by 1 cent (tick). That can occur from one data feed to another just because one of their clocks changes a split second before the other. If it were actually bars built with even versus odd minutes, then the bars would look quite a bit different from each other, and what you said would certainly be true.

There are also slight differences in the DM lines, but as you said, I don't think its enough to explain the differences in the DX. Even if there were bigger differences, the places where they equal each other, like in my previous example, should lead to an equal DX value at that point in time since DX is calculated on a single data point (the current DM+/- values). There must be something else going on.
 
Here is another screen shot that will hopefully help you guys figure it out. Check out the 2 nice short calls today. Both at least .10+ moves. Good day today...so far.:D

The screen shot makes the DMI lines look broken up, in the real software it doesn't look like that (it's much cleaner).

Happy Trading
 

Attachments

nyc,

Have you ever thought about just trading the emini's instead of stocks? Commish lower and shorting on downticks as well. Just wondering. Thanks for hanging around and posting more charts.
 
Quote from Stok:

nyc,

Have you ever thought about just trading the emini's instead of stocks? Commish lower and shorting on downticks as well. Just wondering. Thanks for hanging around and posting more charts.
More...

I just got a little annoyed the other day by the negativity and stupid accusations. I won't let a few jerks drive me away. I came here to offer up my strategy to those that would like to see something that works.

I trade stock and I own some investment real estate properties and I do ok with both those ventures. The stock trading affords me the flexibility I need to run my properties. I have considered trading the emini's among other things, but my stock strategy is working so well for me that I see no reason to change anything at this point.
 
You must log in or register to reply here.
Back
Top