GDX Next Buy signal

[Pascal]

As can be seen below, as soon as the MF crosses over the Buy Limit level, a non-confirmed buy signal is produced, together with a confidence level.

As a reminder, this confidence level is a statistical measure of the probability that the intraday signal will be confirmed by the close. This probability is calculated using the distribution of MF variations every day in the past 1260 trading days at different period of a trading day. A confidence level higher than 90% indicates that "in the past 1260 days, 9 times out of 10, the MF variation until the end of the day was not strong enough for the signal to reverse back below the buy limit." A reversal back below the buy limit would cancel the buy signal.


Pascal

16486

[Pascal]

As we can see below, the confidence level is still less than 2%. This is due to the fact that there is still a long time before the market closes. However, if 10 minutes before the close, the MF was still at about the same level (0.1%), then the confidence level would rise to 65% and then to 90.5% two minutes before the close.


Pascal

16487

[Pascal]

With the confidence level higher than 75% 20 minutes before the close, I believe that we can close the short position and turn long.


Pascal

16488

[pdp-brugge]

Pascal,

Could you publish an updated table with the GDX RT Strong Signals ?
Like table 3 on page 6 of your document published October 22 but updated with all the recent trades.
Is it possible to mark each trade in this table with the signal type: Bought, Bought Oversold, Shorted, Shorted Overbought ?

Thanks

PdP

[lisa]

pascal,

i love the new gdx rt! it seems very straightforward and easy to follow without needing a lot of background rules to remember. just wondering, is there supposed to be information in the "intraday signal" box at all times? there was information in it at one point but not presently.

thanks,
lisa

[Pascal]

pascal,

i love the new gdx rt! it seems very straightforward and easy to follow without needing a lot of background rules to remember. just wondering, is there supposed to be information in the "intraday signal" box at all times? there was information in it at one point but not presently.

thanks,
lisa

Thank you Lisa. I am still collecting data and writing a document to post in the next days.
The Intraday Signal is used only when there is a signal change during the day.
For example, we are now in a "Bought Protected" stage. if we cross below the short level during the day, then a non-confirmed short signal will be issued, and from that point on, as far as we stay below that short level, you will see a "confidence level" appear in that box.

When that confidence level is higher than a certain value (originally it was 90%, but we lowered it to 80%) then a "confirmed short" signal is issued and the Model status changes to "Short"

Also, you do not need to follow the model during all day. Just watching for the last 30 minutes will allow you to get the most important changes on which to act.


Pascal

[lisa]

great, thanks, pascal. i'll look forward to reading the document.

lisa

[davidallison@gmail.com]

Also, you do not need to follow the model during all day. Just watching for the last 30 minutes will allow you to get the most important changes on which to act.Pascal

Hi Pascal, With the RT now shorting and if one was in a long position, then checking in 30 min before the close, what action would you suggest if we appear to be closing in an unconfirmed RT short? Close the long position and go to cash?

Thanks, Dave

[Pascal]

Hi Pascal, With the RT now shorting and if one was in a long position, then checking in 30 min before the close, what action would you suggest if we appear to be closing in an unconfirmed RT short? Close the long position and go to cash?

Thanks, Dave

If we are in non-confirmed short 30 minutes before the close, I would wait and see if the signal gets a confirmation.
If I see a few minutes before the close that we will have a confirmation then I will place a short.

I believe that the confidence level will be clear.



Pascal

[davidallison@gmail.com]

If we are in non-confirmed short 30 minutes before the close, I would wait and see if the signal gets a confirmation.
If I see a few minutes before the close that we will have a confirmation then I will place a short.

I believe that the confidence level will be clear.

Pascal

It's interesting to watch it unfold. For daytraders, it seems if you take the RT signal, the question of closing or maintaning the position is answered in the confidence level. Thanks Pascal, Dave

[Pascal]

It's interesting to watch it unfold. For daytraders, it seems if you take the RT signal, the question of closing or maintaning the position is answered in the confidence level. Thanks Pascal, Dave

The confidence look at the time left until the end of day and the distance to the signal.
This means that if the distance to the signal does not change, as we come closer to the EOD, then the confidence level will rise to 80% and then the short signal will be issued.

For a short signal not to be issued, then we need to move above the dotted short level.


Pascal

[DJones]

If we are in non-confirmed short 30 minutes before the close, I would wait and see if the signal gets a confirmation.
If I see a few minutes before the close that we will have a confirmation then I will place a short.

I believe that the confidence level will be clear.
Pascal

So...just to be clear...if there is no confirmation before the close, you would stay long even though the bot has issued a (unconfirmed) short signal.

Personally, I'm inclined to stay long here, even into the weekend though I recognize the risk with that. Why? In the past, looking at a longer term (5, 10, even 15 day) comparison of MF and price has given a good sense of where the price is relative to where we would expect it to be given the MF over that time period. In some cases, and this is shaping up to be one, the drawdown gets a little frisky, but eventually that longer time frame has correctly predicted a winning trade. Looking at the longer term (5/10/15) now sure makes one think the next move is to the upside. I'll admit, though, looking at a 20 day time period that answer is a little less clear.

[Pascal]

So...just to be clear...if there is no confirmation before the close, you would stay long even though the bot has issued a (unconfirmed) short signal.

Personally, I'm inclined to stay long here, even into the weekend though I recognize the risk with that. Why? In the past, looking at a longer term (5, 10, even 15 day) comparison of MF and price has given a good sense of where the price is relative to where we would expect it to be given the MF over that time period. In some cases, and this is shaping up to be one, the drawdown gets a little frisky, but eventually that longer time frame has correctly predicted a winning trade. Looking at the longer term (5/10/15) now sure makes one think the next move is to the upside. I'll admit, though, looking at a 20 day time period that answer is a little less clear.

You are indeed right: the short signal was canceled and I really do not expect to be forced out now.
In fact, the confidence level works like the porosity level, except that it is "flexible" in terms of time and distance to the signal... and it goes back to see what the GDX MF did in the past 1158 days.


Pascal


Pascal

[davidallison@gmail.com]

So...just to be clear...if there is no confirmation before the close, you would stay long even though the bot has issued a (unconfirmed) short signal.

Personally, I'm inclined to stay long here, even into the weekend though I recognize the risk with that. Why? In the past, looking at a longer term (5, 10, even 15 day) comparison of MF and price has given a good sense of where the price is relative to where we would expect it to be given the MF over that time period. In some cases, and this is shaping up to be one, the drawdown gets a little frisky, but eventually that longer time frame has correctly predicted a winning trade. Looking at the longer term (5/10/15) now sure makes one think the next move is to the upside. I'll admit, though, looking at a 20 day time period that answer is a little less clear.

Good to talk about this. I am long as well, but was perhaps considering what to do if things had deteriorated. The 'Next signal strenght' would indicate that shorting, or maybe position sizing if one wanted to short, is answered there. In other words, shorting is ill advised. Looks like the dip was a great buying opportunity. Hey Djone, Dave

[Pascal]

By the way, I noticed that the reversal signal came quicker than the chart update, because a chart is heavier to send and slower to be processed by the browser.


Pascal

[pdp-brugge]

Pascal,

According to what I read in the new Trading Models section, a confidence level of 80% is needed to trade a signal intraday.
Yesterday, the intraday Bought Protected signal had a confidence of 78.0%.
What do we do now:
- go long at the open today ?
- stay short and wait until the bought signal gets a confidence level above 80% ?

Please advice

[Pascal]

Pascal,

According to what I read in the new Trading Models section, a confidence level of 80% is needed to trade a signal intraday.
Yesterday, the intraday Bought Protected signal had a confidence of 78.0%.
What do we do now:
- go long at the open today ?
- stay short and wait until the bought signal gets a confidence level above 80% ?

Please advice

Good question! I was going to post a message regarding yesterday's Confidence level evolution.

We are issuing reversed signals simply because the GDX Model is whipsawing around the 0 level.

I attach below two screen shots taken at different time, but very close to the EOD.
You can see that the confidence level varied much.
The reason is simple: All instruments experience and end of day spike, which correspond to "at the close" orders.

For GDX, the average spike is 0.0582%. However, 20% of the end of day spikes are above 0.085%
Since the distance between the signal and the Buy level was always lower than 0.085% in the last minute of trading, then an intraday confirmed buy signal could NOT be issued.

I myself however closed the short when the confidence level was around 55% and took a long position at a confidence level of 70%, but the system is programmed for 80% (I "gambled" that the spike at the end of the day would not be strongly negative.)

Since the signal is issued at the close, for our track record, we will note a change to a buy signal at the price of the open of the day.

Please understand that the confidence level is only a guide that uses past trade statistics. I advise everyone to take trading decisions as they see fit. When the confidence level is 70%, it means that in 7 cases out of 10, the signal did not reverse by the close.

What you might do is close a previous position when a new signal reaches 50% and then open a new position afterward when you see it fit, depending on you judgement of the probability.

The GDX Model is now in a BUY PROTECTED mode.

We will most probably also add in a new information box the probability that a new "NON-CONFIRMED" signal is issued during the day. Once such a signal is issued, then the system will turn on the confidence level.


Pascal

16548


16550

[pdp-brugge]

Thanks for the reply!
I know now what to do.

[DJones]

[QUOTE=Pascal;24304]Good question! I was going to post a message regarding yesterday's Confidence level evolution.
Since the signal is issued at the close, for our track record, we will note a change to a buy signal at the price of the open of the day.
Pascal

1) The track record will show a buy at what price? The next day's open, or that day's open?

2) Could you post the current GDX RT results spreadsheet?

Thanks

[Pascal]

[QUOTE=Pascal;24304]Good question! I was going to post a message regarding yesterday's Confidence level evolution.
Since the signal is issued at the close, for our track record, we will note a change to a buy signal at the price of the open of the day.
Pascal

1) The track record will show a buy at what price? The next day's open, or that day's open?

2) Could you post the current GDX RT results spreadsheet?

Thanks

The track record will show a buy at today's open, whihc I believe was around 46.2

The GDX RT spreadsheet can be seen in the GDX Model return section. See below.
I'll update that link whenever I'll remember to do so or when someone will nicely remind me.


Pascal

16557

[pdp-brugge]

gdx is whipsawing :(
I have opened a long trade at the open that I have closed once the short signal had a confidence of 50%

Th confidence is now (11:37 New-York Time) above 70%.
Maybe a stupid question: what happens if intraday the confidence goes above 80% - is there a possibility that it then again drops below?
If so, should we wait until 30 minutes before the close to act

[Pascal]

gdx is whipsawing :(
I have opened a long trade at the open that I have closed once the short signal had a confidence of 50%

Th confidence is now (11:37 New-York Time) above 70%.
Maybe a stupid question: what happens if intraday the confidence goes above 80% - is there a possibility that it then again drops below?
If so, should we wait until 30 minutes before the close to act

At 80%, the short trade will be opened and the confidence level will stop being shown.
At 80%, it means that "that far in the trading day, only 20% of the time did the MF move back up to the short limit"



Pascal

[pdp-brugge]

Ok. Clear. Thanks

[pdp-brugge]

Pascal,

I find the current information on the "GDX Real-Time View" somewhat confusing.
It states that the last close was "Bought Protected" although the current state is "Short Protected".
In the next signal message there are two lines that say the same:
- We will buy when the MF crosses above 0.014%
- We will buy when the MF crosses above Average + 0.014%

I noticed that a few minutes before the close yesterday there was a buy protected target at +/- -0.46%.
This level was reached whitin seconds of the close but the confidence level was very low.
Is the false "last close" message coming from this target that was just hit within seconds of the close?
Was that buy level (- 0.46%) a false one?
Is the GDX RT now still short?

Please advice

[Pascal]

There is indeed a software error in the programming of the logic.
The RT system had one status change yesterday and tried a second status change, which should have been disabled, because the EOD system itself can only have one status change a day.

The GDX model is still short and will turn long if we cross over the 0 level.

If we had closed below the average (and we closed just above it), then the model would have turned long on a move above that average; this is what the RT system tried to do at the EOD, but such a decision should not have happened as we already had one signal change during the day.

We are correcting this so that we start the day in a short mode as should be the case.


Pascal




Pascal,

I find the current information on the "GDX Real-Time View" somewhat confusing.
It states that the last close was "Bought Protected" although the current state is "Short Protected".
In the next signal message there are two lines that say the same:
- We will buy when the MF crosses above 0.014%
- We will buy when the MF crosses above Average + 0.014%

I noticed that a few minutes before the close yesterday there was a buy protected target at +/- -0.46%.
This level was reached whitin seconds of the close but the confidence level was very low.
Is the false "last close" message coming from this target that was just hit within seconds of the close?
Was that buy level (- 0.46%) a false one?
Is the GDX RT now still short?

Please advice