DMI/ADX consistently works

Hi NYC-Hotshot,

I used the "DX", as opposed to the "ADX", as your later posts indicated.

What I didn't do however, was look for the "explosive" move as you indicate. Any time DX crossed and was above PDI/MDI, I entered the requisite Long or Short trade. When DX crossed again, I exited accordingly.

What are your subjective measurements in terms of "explosive" moves ? In other words, how much above PDI and MDI lines (in absolute or % terms) does DX have to be, before you enter a position ?

I understand this is a look and feel strategy, but I'm curious to see if I can get better results using the answer you provide.

Thanks in advance,

Doug
 
That's the problem there is no finite % or amount that I use. I just want to see the white (DX) line ascending rapidly. Don't get me wrong, I do enter on some less "explosive" moves, but that's when I'll be happy with .02 or .03.

Another thing that is faulty in that back test is I do not always exit when the white crosses back below the others. Often just a descent in DX will cause me to bail out even if it hasn't crossed.

Sorry, I know that probably doesn't help you to program parameters, but thats what I was saying about this type of trading being somewhat subjective. I do not have a program that rings "enter now" or "exit now" alerts. I have been doing this for years now entry and exits are just something you come about through practice and experience.
 
As a result of this useful thread I have been experimenting using Pmi and Dmi scans in the evening, which has produced what appears to be good swing trade prospects.

1. Scan for Adx>20 and where the difference between pmi and dmi is less than 17, but have not yet recently crossed, but look like they may cross soon. With pmi green and dmi red, a buy candidate would be one where the pmi is lower than the dmi and is sloping up heading towards the dmi line and about ready to cross the dmi. A sell candidate would be where the pmi is higher than the dmi and the dmi is sloping up about ready to cross the pmi.

2. Visually scan this list for candidates where pmi and dmi are closing together, preferably converging or closing together at a steep angle, rather than meandering sideways.

3. Further refine this list to those stocks that may have other confirming indicators. Like those that appear to have been consolidating and ready to breakout or breakdown by looking at the chart patterns for volume or any signs that would indicate distribution or accumulation, or for any useful candlestick pattern. The general idea is to look for pmi/dmi candidates where the upward momentum is slowing down, reversing and and about ready to tip over for shorts. Or for longs, where the downward momentum has slowed and the buying is starting to pick up about ready to breakout.

4. I'm experimenting with an intraday trade setup that would trade these candidates for breakouts above yesterday's high if the market is trending up. And entering shorts for breakdowns only if the market is trending down.

If you look at enough charts, it looks like the most profitable entries are ones that anticipate and enter before the pmi/dmi crossovers.

At this point, I'm not sure if positive results would be any better just using 1 and 2 versus just 3 and 4, versus 1,2,3, and 4. But it has produced some interesting results.


Thanks nyc-hotshot and everyone for contributing.
 
Quote from kowboy:

If you look at enough charts, it looks like the most profitable entries are ones that anticipate and enter before the pmi/dmi crossovers.
More...

A word of caution here. One thing to realize is that virtually every single trade that is profitable by taking a crossover would have been even more profitable by jumping in early on anticipation of a cross. What that doesn't tell you however is how many times there was no eventual crossover trade, yet the anticipation entry would have occured (since both crossovers and fakeouts will look the same before the outcome is known). That is a dangerous problem with visual backtesting. Its easy to see only what you are looking for during the test, but when you go live, you get them all.
 
Quote from WarEagle:

A word of caution here. One thing to realize is that virtually every single trade that is profitable by taking a crossover would have been even more profitable by jumping in early on anticipation of a cross. What that doesn't tell you however is how many times there was no eventual crossover trade, yet the anticipation entry would have occurred. That is a dangerous problem with visual backtesting. Its easy to see only what you are looking for during the test, but when you go live, you get them all.
More...

I agree wholeheartedly War Eagle. There can be many false entries with anticipation, just like there also appear to be many false cross overs. The thing I was suggesting, is to use dmi/pmi to obtain a list of candidates. Then use other criteria to confirm a potential candidate and then to use the market trend for that day to help the trade in your favor.

This has been a really interesting thread. Thanks
 
I wouldn't make "anticipation" trades. There are plenty of opportunities daily just for explosive DX crosses over MDI/PDI. Worst case (if you take enough shares) you can always jump out at .02 (or even .01). Those are filler trades and add up to decent profits over time (even if you only net around $10). There will be a few times each day when you can take the .05-.10+ ride that makes your day. However, there is no such thing as a bad positive trade.
 
nyc,

How long have you been trading this method?

just wondering since your normal trade is from 0.02-0.04. Did you do this scalping when stocks were trading in fractions? If so how did the transition to decimals affect your trading?

Thanks
Eric
 
NYC

I opened an account with speedtrader and used the same DX study as you.

I papertraded for about 2 weeks and did ok with your method.

Thanx.

I then funded the account and traded for ONE day net -$200 loss BUT
I had $500 in commission including all the exchange fee's.

This was at $11 per trade.

I only did 15 trades during the day but this was insane.

Thanks for the help but I will stick to futures $4.20 r/t.

Good luck and thanks for sharing. :)
 
Very interesting thread, lot of good ideas.

My post will try to look at DMI/ADX from different point of view.
When I look as the indicators are calculated, there is nothing about crossing DMI-,DMI+,ADX or about which line should be at the top. All is about change or direction+rate of change.

Purely from math long trend is starting when: (first derivative of DMI+ > 0) AND (first derivative of DMI- < 0) AND (first derivative of DMI+ > abs(first derivative of DMI > 0) AND ). Maybe it sound ugly, for chartist: DMI+ goes up, DMI- goes down and DMI+ goes up steeper than DMI- declines. In other words, if DMI+ is below DMI- you CAN anticipate crossing.
The last condition is the same as second derivative of ADX > 0. Again for chartist: when ADX goes down, declining rate slows, line starts go flat and then up. When ADX goes up, the angle will be steeper. For example parabola has second derivative > 0.

End of the long trend is simpler to explain. Math: local maximum of ADX ... first derivative of ADX = 0. Chart : Local peak in ADX chart .. ADX line starts go down.

I'm not sure about thresholds and strength of trend (20,30,40,...), I hope better is look at first derivative of DMI+ for up trend, this tells how strong trend is and how quick the price grows per bar. For down trend strength you can use first derivative of DMI-. In chart you can see how is the DMI+,DMI- line steep or flat, because first derivative shows the angle of the line.

In praxis should be DMI/ADX line smooth, (Jurik JMA, spline functions, EMA for 3-5 bars) and then derivative should be approximated by differences. When I'll have spare time I try some simulation of this ideas.

For short trend only change DMI+ and DMI-.

Please tolerate my English.
 
50 pages of posts- Phew! Thanks nyc-hotshot for starting this thread and offering up your technique. Thanks to everyone else for your comments and analysis.

-peerless
 
You must log in or register to reply here.
Back
Top