GDX Models Under Pressure - April 23, 2012

[Billy]

The long GDX robot position has me worried a lot. First, last Friday, physical gold didn’t react at all to the USD weakness and is getting very weak in the futures market as I am writing this commentary. Second, GDX is very close to the lowest support line (45.98) of the ongoing consolidation that began on April 4th. Third, the Precious Metals MF closed in negative territory and below the MF average. Large buyers were totally absent in the final minutes of trading, contrary to their recent automatisms. The MF is just 0.05% shy from issuing a short signal today at the porosity limit of -0.150%. The RT model is most likely to trigger a short signal early today and the EOD model will have to wait for a confirmation by the close to sell its position at the open tomorrow.
To avoid a possible ultimate shakeout followed by an intraday price and MF reversal to the upside, I plan to set my discretionary hard stop just below Quarterly S1 (45.75), the strongest and most significant nearby floor support. If the stop is not hit and the EOD model closes below -0.15% with a sell signal, I will close my long position EOD.
The robot’s secondary limit entry at 45.15 should not be taken into consideration if the RT MF is trading below -0.15%.
13912

The IWM robot remains short and still advises a secondary entry at the same limit price as the initial entry of 80.03.

IWM will start the week with the confluent resistance of Quarterly pivot (80.28) and Weekly pivot (80.21). Now that the opex bullish bias is behind us, It will likely cautiously trade below these resistances and in the lowest area of the consolidation until Wednesday’s morning reaction to AAPL’s earnings or Wednesday’s afternoon FOMC statement. The whole world and media networks will be focused on these two events this week and we just need to wait for the market’s reaction and 20 DMF response to the news.
Billy
13913

[Billy]

The GDX RT model has turned short at 9:57 am when the MF hit -0.15% at a price of 44.81.
The EOD robot was stopped out previously at 45.06.
Since then, the MF has already traded below average – porosity, so the new protection level for the RT model where it would turn long again is now at average + porosity (0.15%).
Otherwise, hitting the oversold level of -1.45% is needed to reverse the RT position from short to long.
Billy

[TraderD]

The GDX RT model has turned short at 9:57 am when the MF hit -0.15% at a price of 44.81.
The EOD robot was stopped out previously at 45.06.
Since then, the MF has already traded below average – porosity, so the new protection level for the RT model where it would turn long again is now at average + porosity (0.15%).
Otherwise, hitting the oversold level of -1.45% is needed to reverse the RT position from short to long.
Billy

Billy,

What I find interesting is that the peaks (or if you like, long position bailout points) in GDX MF the last 3-4 weeks occurred around +0.60% (twice) which is quite off an overbought level. This signifies to me a longer timeframe downtrend in GDX which isn't captured by a 20-day summation window for MF. It may be useful to generate a set of longer-period MF curves alongside the standard 20-day window IMHO.

Trader D

[Pascal]

We tested longer time frames, but they are slower to react and in general offer poor returns.
We will however retest shorter/longer time frames when the database will be ready to back-test RT signals.

Pascal



Billy,

What I find interesting is that the peaks (or if you like, long position bailout points) in GDX MF the last 3-4 weeks occurred around +0.60% (twice) which is quite off an overbought level. This signifies to me a longer timeframe downtrend in GDX which isn't captured by a 20-day summation window for MF. It may be useful to generate a set of longer-period MF curves alongside the standard 20-day window IMHO.

Trader D

[Billy]

Billy,

What I find interesting is that the peaks (or if you like, long position bailout points) in GDX MF the last 3-4 weeks occurred around +0.60% (twice) which is quite off an overbought level. This signifies to me a longer timeframe downtrend in GDX which isn't captured by a 20-day summation window for MF. It may be useful to generate a set of longer-period MF curves alongside the standard 20-day window IMHO.

Trader D

This is of course tempting to do and it is beyond my personal competence.
However, it seems to me that the longer the MF summation timeframes you will use, the biggest drawdowns you will get because the longer you’ll need to wait for a fresh signal. Or you’ll need to find new rules with more protection levels and more whipsaws. I know, nothing can ever be perfect in this world…
Billy

[TraderD]

This is of course tempting to do and it is beyond my personal competence.
However, it seems to me that the longer the MF summation timeframes you will use, the biggest drawdowns you will get because the longer you’ll need to wait for a fresh signal. Or you’ll need to find new rules with more protection levels and more whipsaws. I know, nothing can ever be perfect in this world…
Billy

I agree with your observation and actually not suggesting to generate signals off of the longer timeframe MF but rather use it to help put the 20-day MF in context, as typically done in traditional TA.

A second observation is the bounce off of the -0.60% MF point today. Is it possible that the market is "learning" MF gyrations and/or reduced volatility results in range compression of the OB/OS MF oscillator?

Trader D

[Pascal]

Billy,

What I find interesting is that the peaks (or if you like, long position bailout points) in GDX MF the last 3-4 weeks occurred around +0.60% (twice) which is quite off an overbought level. This signifies to me a longer timeframe downtrend in GDX which isn't captured by a 20-day summation window for MF. It may be useful to generate a set of longer-period MF curves alongside the standard 20-day window IMHO.

Trader D

Below are a few different time frames with some green-red arrows (I did not use the OB/OS levels but the simpler earlier model). We can see that the longer the time frame the mode difficult it is to make money, except if you accept larger drawdowns.

The shorter 15D time frame produces many signals. Maybe too many for position traidng, but these could be suitable in a RT system. At least, we should back-test this idea.

What would be your idea to use long-term time frames to modeulate the shorter time frame signals?


Pascal

13930
13926
13929
13928

[TraderD]

Below are a few different time frames with some green-red arrows (I did not use the OB/OS levels but the simpler earlier model). We can see that the longer the time frame the mode difficult it is to make money, except if you accept larger drawdowns.

The shorter 15D time frame produces many signals. Maybe too many for position traidng, but these could be suitable in a RT system. At least, we should back-test this idea.

What would be your idea to use long-term time frames to modeulate the shorter time frame signals?

Pascal


Hi Pascal,

As mentioned in my earlier reply, I'm not suggesting to generate signals off of the longer timeframes' MF curves. To answer your question, if we suspect that the market is "learning" the MF gyrations, we may be able to use the dynamic range of a long timeframe (e.g. 200-day) MF to continually adapt the OB/OS thresholds of the 20-day MF. The purpose of that would be to help make the 20-day MF based signals be more "adaptive". Maybe even dealing with non-stationarity of the zero line if that effect manifests itself. In short, I see it as means to improve the 20-day MF for signal generation. To a secondary degree, the longer term MF may be used to create a trading bias (e.g. take only long or short trades during certain times, etc.)

Trader D