Combo-MF, a simple and robust trading strategy ?!?

[pdp-brugge]

No changes

[pdp-brugge]

I have updated the theoretical equity graph of my strategy. September was a small but positive month. The streak of losing months is broken, I think. My NoFear has protected me against larger draw downs.

[manucastle]

I have updated the theoretical equity graph of my strategy. September was a small but positive month. The streak of losing months is broken, I think. My NoFear has protected me against larger draw downs.

Thanks very much Pascal. Very informative.

Trev

[pdp-brugge]

No changes in signals or trades.
The slope of the 14SMA of the S&P500 has started to go down yesterday. This enhances the odds for a short signal in the coming days.

[pdp-brugge]

No changes

[pdp-brugge]

The VSTpro has signaled a SHORT signal at the close of Friday. When considering the 20DMF is still LONG, then my Combo-MF model would go to neutral (CASH) because of the conflict between the two models. I am still considering the 20DMF (based on my personal conviction) to be LONG since the launch of QE3.

The models XLB and XLF have both a SHORTED OVERBOUGHT signal. The basic rules of my strategy would suggest going short for these models. I will not follow this because, in my humble opinion, both these trades have had a length when the "juice is out of the bottle".

The models XLI and XLK are in CASH so no need to do anything.

The XLY model has a SHORTED OVERBOUGHT signal. The statistics of my back tests have learned me that following a SHORTED OVERBOUGHT signal for the XLY model is not beneficial if the 20DMF is not SHORT.

Conclusion: I will close the open long trade in TQQQ and do nothing else. The odds for the VSTpro to signal a new LONG signal are strong. Let's wait and see

[pdp-brugge]

No changes

[manucastle]

The VSTpro has signaled a SHORT signal at the close of Friday. When considering the 20DMF is still LONG, then my Combo-MF model would go to neutral (CASH) because of the conflict between the two models. I am still considering the 20DMF (based on my personal conviction) to be LONG since the launch of QE3.

The models XLB and XLF have both a SHORTED OVERBOUGHT signal. The basic rules of my strategy would suggest going short for these models. I will not follow this because, in my humble opinion, both these trades have had a length when the "juice is out of the bottle".

The models XLI and XLK are in CASH so no need to do anything.

The XLY model has a SHORTED OVERBOUGHT signal. The statistics of my back tests have learned me that following a SHORTED OVERBOUGHT signal for the XLY model is not beneficial if the 20DMF is not SHORT.

Conclusion: I will close the open long trade in TQQQ and do nothing else. The odds for the VSTpro to signal a new LONG signal are strong. Let's wait and see

PdP,

The 20DMF appears to be more neutral than long. Do you agree, and how would that affect your model.

Thanks in advance.

Trev

[pdp-brugge]

Trev,

It is my personal judgement that I consider the 20DMF now in a LONG state (since the announcement of QE3).

In the weekend I have done the simulation to see what would change if the 20DMF was indeed still considered to be in CASH. Over the last 17 trading days, this change would be an increase of the performance of my strategy of 0.0029% over that period (17 trading days).
The Combo-MF would now be in SHORT mode because a SHORT for the VSTpro and a CASH for the 20DMF results in a SHORT for the Combo-MF. I would have skipped this trade because my back tests statistics have learned me that following a SHORT trade for the Combo-MF model when the 20DMF is in CASH and the 50MA is above the 200MA is not beneficial.
The SHORTED OVERBOUGHT signal for the XLB model could have resulted in a SHORT trade but I still would hesitate because that SHORTED OVERBOUGHT signal was last Friday 11 days old ("juice out of the bottle" syndrome).
The same for the SHORTED OVERBOUGHT signal for the XLF model.

As a reference I attach the Strategy Status in the consideration of a 20DMF in CASH mode.

I know my judgement is now somewhat emotional but my back tests statistics have also shown that periods where the 20DMF is CASH and the VSTpro is SHORT are not the greatest in performance. They also do not last very long.

If and when the 20DMF also would go SHORT, then we would have an other type of ball game. Then things would really be different and I would follow my theoretical strategy again.

The odds for a change of the VSTpro to signal LONG again are high. We need a high higher then yesterday, a low below the 14SMA and a close above the 14SMA. With the increase volume compared to yesterday I suspect the volatility will also be higher today.

[manucastle]

Trev,

It is my personal judgement that I consider the 20DMF now in a LONG state (since the announcement of QE3).

In the weekend I have done the simulation to see what would change if the 20DMF was indeed still considered to be in CASH. Over the last 17 trading days, this change would be an increase of the performance of my strategy of 0.0029% over that period (17 trading days).
The Combo-MF would now be in SHORT mode because a SHORT for the VSTpro and a CASH for the 20DMF results in a SHORT for the Combo-MF. I would have skipped this trade because my back tests statistics have learned me that following a SHORT trade for the Combo-MF model when the 20DMF is in CASH and the 50MA is above the 200MA is not beneficial.
The SHORTED OVERBOUGHT signal for the XLB model could have resulted in a SHORT trade but I still would hesitate because that SHORTED OVERBOUGHT signal was last Friday 11 days old ("juice out of the bottle" syndrome).
The same for the SHORTED OVERBOUGHT signal for the XLF model.

As a reference I attach the Strategy Status in the consideration of a 20DMF in CASH mode.

I know my judgement is now somewhat emotional but my back tests statistics have also shown that periods where the 20DMF is CASH and the VSTpro is SHORT are not the greatest in performance. They also do not last very long.

If and when the 20DMF also would go SHORT, then we would have an other type of ball game. Then things would really be different and I would follow my theoretical strategy again.

The odds for a change of the VSTpro to signal LONG again are high. We need a high higher then yesterday, a low below the 14SMA and a close above the 14SMA. With the increase volume compared to yesterday I suspect the volatility will also be higher today.

Thanks for your comprehensive and valuable reply Pascal :O)

Trev

[manucastle]

Thanks for your comprehensive and valuable reply Pascal :O)

Trev

Pascal, you use TQQQ for the market. Have you tested if the model would improve or be more stable if you used XLK as a further confirmation of trend for risk control ?

Trev

[pdp-brugge]

Trev,

A while back, I did a very comprehensive research to find the "best" instrument for trading the Combo-MF signals. Out of that work, I deducted the use of TQQQ.
Because I always want to keep my eyes open for a changing environment, I have just made a small comparison between TQQQ and TECL (the 3x ETF for XLK) with the Combo-MF signals between January 2010 and now. This back test confirms that TQQQ is still the best vehicle.

[manucastle]

Trev,

A while back, I did a very comprehensive research to find the "best" instrument for trading the Combo-MF signals. Out of that work, I deducted the use of TQQQ.
Because I always want to keep my eyes open for a changing environment, I have just made a small comparison between TQQQ and TECL (the 3x ETF for XLK) with the Combo-MF signals between January 2010 and now. This back test confirms that TQQQ is still the best vehicle.

Thanks Pascal, good to see you are constanly adapting to current conditions. Very neccassary in todays markets.

Trev

[Harry]

Pascal,

I recently questioned SPX "Open" values and confirmed (via this forum) that they are not really indicative of the market open since not all securities not starting trading at exactly 9:30. I know VST uses Open values in it's calculations and wonder if you looked at the difference in signals when comparing to say SPY? I did a rough check and found quite a few signal differences.

I question aloud how much better VST could be if it's signals were not generate off of fictitious data? Could a MA be optimized for SPY for even better returns? Maybe not? Have you considered this?

Harry

[pdp-brugge]

Harry,

The last days I have been busy with optimizing exactly that: finding the most optimal MA parameters.
It takes a lot of testing but my goal is to combine the different EV signals (20DMF, the 9 S&P sub-components and PM) with their appropriate VSTpro counterpart signals.
For each tradable EV signal could/should be a comparable VSTpro signal.
Out of this ongoing work I hope I will be able to digest more efficient combined signals.
I will present my results once they are done.

PdP

[pdp-brugge]

The XLK model has now a SHORT signal. I will not follow this signal because my back test statistics have learned me that following a SHORT trade when the 20DMF is not SHORT is not beneficial.

[pdp-brugge]

No changes in signals nor trades.

[pdp-brugge]

The 20DMF has signaled a SHORT signal yesterday. All the models that I follow have now a SHORT signal. I am going to open the corresponding new trades at the open today. Let's hope I won't be trying to catch a falling knife...

[pdp-brugge]

No changes

[pdp-brugge]

No changes

[pdp-brugge]

The VSTpro has a LONG signal since yesterday. This has turned the Combo-MF to CASH because of the conflict between the VSTpro and the 20DMF signals. Since the 14SMA is rising, a new SHORT signal for the VSTpro is not possible today.

[pdp-brugge]

The 20DMF is in CASH since yesterday. The Combo-MF will now follow the VSTpro LONG signal.
The XLF model goes gone to CASH. The XLI model has a BOUGHT signal.
I will close the SHORT trades for the XLF, XLI, XLK and XLY models.
I will open a new LONG trade in TQQQ.
As clearly indicated with the low Kelly and the high NoFear, my strategy is performing very poor (for now).

[pdp-brugge]

The XLB model has signaled a BOUGHT signal yesterday.

[pdp-brugge]

The 20DMF has a Short signal since the close of last Friday. This means that the Combo-MF will go to CASH.
The XLF model has a Shorted Overbought signal.
The trade signals from my strategy are mixed: XLB & XLI both LONG while XLF, XLK and XLY are SHORT.
I think this gives me a hedge situation to keep me floating through these rough waters.

[pdp-brugge]

No changes

[Harry]

pdp-brugge,

20DMF moved back to cash today, not short as you have listed? Are you making a subjective call?

Harry

[pdp-brugge]

Harry,

Thanks for noticing this!
I suppose I was not already awake this morning when I looked at the 20DMF.
You are right: the 20DMF is neutral again since the close of yesterday.
This changes a lot!
The Combo-MF goes back to LONG.
I will not follow the signals from the XLF, XLI, XLK and XLY models because of their poor statistical performances learned from my back-tests.

[pdp-brugge]

No changes

[manucastle]

No changes


Hi PdP,

Seems strange the VSTPro signal is long in this current market ?

Trev

[pdp-brugge]

Trev,

I know and it is somewhat frustrating.
Last Friday, the S&P500 turned down very fast while the 14SMA was not going downwards for at least two days. There have been other occasions in the past where the same thing happened. May 4, 2012 serves as an example. At that time, the 20DMF caught the down-move and went short. Last Friday, the 20DMF also signaled a short signal. This short signal was however cancelled this Monday.
There are several possibilities for adapting the VSTpro formula to try catching these fast moves down. I have tried to adapt the VSTpro formula to obtain better results, all with poorer results when I perform my back-tests.
In the broader picture, I have come to the conclusion that the combining of the VSTpro signals with the 20DMF EOD signals is still the best model.
There have been, according to my back-tests, similar periods in time where the performance of the Combo-MF model was not so good. None however so long as the one we are in now.
In the mean time, I am waiting for times to change.
I am waiting very eager on Pascal's detailed back-tests results for the new 20DMF RT model.

PdP

[Pascal]

The new 20DMF's back-test results are here:

16240


Pascal

[pdp-brugge]

Pascal,

I admire the work you did regarding the RT system.
After reading your latest document several times, I can not construct a clear table with the new RT 20DMF signals.
Is it possible for you to supply a simple table with the dates and the appropriated signals going back as far as possible?
Thanks in advance.

PdP

[pdp-brugge]

The XLK model has a Bought Oversold signal and I will follow this signal.
For the XLB model, the ticker that I use to trade this model (UYM) has had a bullish signal as of yesterday. The 50MA is now above the 200MA. My back-tested rule for this model is to ignore Bought signals if the 50MA is above the 200MA. I will close the current long trade in UYM at the open today.

[pdp-brugge]

The 20DMF has signaled a SHORT signal yesterday. This means that my Combo-MF is in CASH again.
The XLB model has also signaled a SHORT signal. The XLK model has gone to CASH.
The only model still in buy position is XLI. Following my rules, I should open a new long position for this model. I will not follow this mainly because RSI, MacD and Williams'%R are all bearish. I know, this is very subjective. For the sake of the statistics, this new long trade in XLI will only be opened in my paper-trading account.
I real life, I am now a bear.

[pdp-brugge]

The XLK model is Bought Oversold since the close last Friday.
I am short for the models XLB, XLF and XLY in my real-life account
I am long for the model XLI in my paper-account and will open a new long trade for the XLK model in that paper-account.
Let's wait and see what the post Sandy trading days deliver.

[pdp-brugge]

No changes

[pdp-brugge]

The XLB model has a Bought signal since yesterday. My back test statistics tell me that this is not a good trade when the 20DMF is Short.
Meanwhile the FAS trade for the XLF model has hit his stop-loss yesterday.

It is painful to see that the real trades I have made following my strategy did not work out and those in my paper account did well. I will see how the futures react to the jobs report at 8:30am. If the reaction is positive I will open the long trades for the XLI and XLK models in my real account. This is a very subjective and therefore personal decision.

[pdp-brugge]

It looks like we are going to have a short signal for the VSTpro today.
The 14SMA has been going down for at least two days. The high was above that 14SMA and the close should be below the open. Unless the S&P500 goes up 17 points between now and the close, we will have a SHORT signal for the VSTpro. Together with the current, still valid, SHORT signal for the 20DMF, this will move the Combo-MF in SHORT state.
I think a powerful signal !

[pdp-brugge]

The VSTpro issued a SHORT signal last Friday. This changes to Combo-MF also to SHORT.
The long signals for the XLB and XLK models are ignored due to their poor statistical performance from my back tests under the current circumstances.

[pdp-brugge]

No changes

[pdp-brugge]

The VSTpro has gone long again yesterday. Therefore the Combo-MF will be neutral due to the conflict between the 20DMF and the VSTpro.
The XLY model has signaled a Bought signal yesterday.
The current short trades for the Combo-MF and XLF models had a stop-loss yesterday.
I will close the current short trade for the XLY model at the open today
I will open a new short trade for the XLF model and new long trades for the XLK and XLY models at the open today.

[pdp-brugge]

The market is producing whipsaws for my strategy :(

There are three rules for the VSTpro to signal a short signal:
- slope of 14SMA going down for minimum the last two days
- high above the 14SMA
- close below the open

Because the slope of the 14SMA of the S&P500 is still going down and the high of the day of an index is the close of the day before and this close was yesterday above the 14SMA, all we need for (again) a new short signal for the VSTpro model is a close below the open today.

[Harry]

... high of the day of an index is the close of the day before ...

Why is this?

[pdp-brugge]

Harry,

I do not know. Sorry.

Most of the time, when there is a very serious down gap, the open of the day is not the real "open" but more or less equal to the close of the previous day. If the day remains negative this means that the virtual open is also the high of the day (and more or less the close of the day before). The open (when a gap down or up) is a fictitious number. My experience is limited but that is how I have seen it.

Maybe someone with more experience can correct me...

[Harry]

Pascal,

My point of asking is that whenever we have gaps, I have seen the index post open numbers different than the appropriate high/low of the previous day, so that's why I asked why?

I have learned (via this site) that the opening value of an index is usually off when we have large gaps either way due to not all the securities start trading at exactly 9:30. I personally have switched to the ETF proxy SPY that trades on the open.

As I'm sure you already know, the signals are different for VSTPro using ^GSPC vs. SPY due to the opening values discussed above. I do not know if this hurts or helps VSTPro performance?

Harry

[pdp-brugge]

Harry,

In my back tests, I have compared the signals generated from .SPX (the MetaStock ticker for the S&P500) and those generated by the ticker SPY
The performance of my Combo-MF model with signals from SPY where significant lower than those from .SPX
I have done no research for the how's/why's of these differences
Maybe they are partially caused by the behaviour that we have experienced today
Jose Silva, the creator of the VSTpro add-on for Metastock, also did his development and testing based on .SPX
Jose has warned his users to use the VSTpro formula only with indexes
For common stocks and ETFs he has a (completely different) trading strategy based on band crossing
Therefor, I will stick with observing the behavior of the VSTpro formula applied to .SPX for determinating the state of the VSPro model
To my experience, fooling around with such essential determinant parameter has not brought any improvement

Pascal

[Pascal]

When you compare, you need to adjust SPY for dividends.
The difference is coming from there.


Pascal

[Harry]

When you compare, you need to adjust SPY for dividends.
The difference is coming from there.Pascal

I don't believe so. The difference I was referring to was in regards to when signals were issued, not returns. Due to the different opening prices between .SPX/SPY and the VSTpro rule of when to go short regarding the open price, there can be a matter of days difference between signals being issued (if they even are sometimes). I discovered this through my backtesting and recall the .SPX gave better returns concurring with pdp and Silva observations. At this point, I stopped my backtesting.

[pdp-brugge]

As expected, the VSTpro model has (again) a short signal.
This turns the Combo-MF model also to short.
The VSTpro model needs a close above 1420 today for signaling a new long signal.
This is about 1.8% gain which is very unlikely under the current market situation.

The XLB and XLI models have a short signal since yesterday and the XLK model has turned into Cash Oversold state.
The current long signal for the XLY model will be ignored due to poor statistical back tested performance when the Combo-MF is short.

I had a stop-loss for my long trade for the XLK model.
I will close the existing long trades for the XLI and XLY models at the open today.
I will open new short trades for the Combo-MF, XLB and XLI models.

I hope the market now shows a clear sign.
I do not care up or down.
Changing direction every few days only brings losses and frustration.
Meanwhile, the whipsaw song keeps on playing...
http://www.youtube.com/watch?v=LiE1VgWdcQM

[pdp-brugge]

The XLY model is now in cash.
The XLF model has a Bought Oversold signal.
I will close the short trade for the XLF model and open a new long trade for this model at the open today.

[pdp-brugge]

The XLF model returned to cash last trading day.
In my strategy, only Combo-MF (QQQ), XLB and XLI remain short.

[pdp-brugge]

No changes

[pdp-brugge]

The 20DMF issued intraday a buy signal that was cancelled at the close.
There is now confusion between the EOD 20DMF and the RT20DMF.
This is frustrating for me. For now, I will follow the EOF 20DMF for my strategy. This is a personal choice. I hope not a bad one...

[pdp-brugge]

The XLB and XLI models have a Bought Oversold signal since yesterday.
I will close the short trades and open new long trades at the open today for these models.

The distance between the close of the S&P500 and his 14SMA is very large. If there is a bounce today, the odds for a long signal for the VSTpro model are very small. Unless the 20DMF signals a Bought Oversold signal today, nothing will change for tomorrow.

[pdp-brugge]

The XLB and XLI models have gone to cash yesterday.
I will close the two long trades for these models at the open today.
The only trade left is the short one for the Combo-MF model.
I will hold this trade until a 20DMF buy or a VSTpro buy is triggered.

[pdp-brugge]

The 20DMF has a long signal since the close last Friday. This puts the Combo-MF into cash.
The distance between the close of the S&P500 and his 14SMA is big so a VSTpro buy signal is not likely. The S&P500 needs to jump more than 2.2% for this to happen.
The XLB model has now a Bought Oversold signal.
I will close my short trade for the Combo-MF model and open a new long trade for the XLB model at the open today.

[manucastle]

The 20DMF has a long signal since the close last Friday. This puts the Combo-MF into cash.
The distance between the close of the S&P500 and his 14SMA is big so a VSTpro buy signal is not likely. The S&P500 needs to jump more than 2.2% for this to happen.
The XLB model has now a Bought Oversold signal.
I will close my short trade for the Combo-MF model and open a new long trade for the XLB model at the open today.

PdP,

Has the recent past made you consider changing anything in your trading strategy ?

Trev

[pdp-brugge]

All the XLX moels that I use in my strategy have now a Bought Oversold signal.
It will take a close of the S&P500 above 1388.42 to generate a long signal for the VSTpro.
If that happens, all my models will be long.

To answer the question from Trev: I have done a lot of research for improving my strategy but have not found anything that fundamentally improves the results based on the same VSTpro, 20DMF and XLX signals.

[pdp-brugge]

No changes

[pdp-brugge]

The VSTpro model has signaled a buy signal at the close Wednesday. This means that the Combo-MF model is now also long.
All my models are now signaling long trades.

[manucastle]

The VSTpro model has signaled a buy signal at the close Wednesday. This means that the Combo-MF model is now also long.
All my models are now signaling long trades.

Missing your model today PdP. Is there a problem ?

Trev

[pdp-brugge]

Trev,

There is no problem; just nothing has changed.
I will come back to the forum when there is a signal change in one of the models.

[manucastle]

Trev,

There is no problem; just nothing has changed.
I will come back to the forum when there is a signal change in one of the models.

I take a keen interest in your Risk Factors (Kelly, No Fear and Risk%) so would appreciate the model if you can find the time :O)

Thanks in advance.

Trev

[pdp-brugge]

here they are

[pdp-brugge]

The VSTpro has signaled a short signal yesterday. This puts my Combo-MF into cash because of the conflict between VSTpro and 20DMF.
The XLF and XLK model have still a Bought Oversold signal but I will close these trades because my back tested statistics learned me that following those Bought Oversold signals when Combo-MF is in cash is not profitable.
If the 20DMF signals a short today, then Combo-MF will also go short.
In that case then the long trades for the XLY model will also need to be closed.

[Pascal]

Thank you for your daily analysis.
Regarding the possibility of the 20DMF issuing a short signal today, it is rather remote, I would say.
Indeed, the 20DMF can issue a short signal either

- if the OB/OS indicator falls back below -30, which is rather far. If this occurs, then we would probably be below the 200MA.
- or if the 20DMF first crosses above the Buy level (+0.11% on an EOD base or +0.176% on an intraday base) and then falls back below -0.11% on an EOD base. We are in Bought Protected 30 mode and need first to close above the normal "Buy" level before registering a down move to a new short level. On an EOD base, you obviously would need two days to complete these two steps.



Pascal

16485

[pdp-brugge]

No changes

[pdp-brugge]

No changes

[pdp-brugge]

No changes

[pdp-brugge]

The XLK model has now a short signal. Following my strategy I did not have a long trade for this model.
If the 20DMF issues a short signal (MF cross below -0.176%) then a lot will change.
For now, I only have (small) long trades for the XLB, XLI and XLY models.

[pdp-brugge]

The XLB model is now short and the XLK model has now a buy signal.
I will ignore the XLB short signal because of poor back tested performance under these circumstances.
I will open a new long trade for the XLK model at the open.

[pdp-brugge]

The XLB model turned long again yesterday. I will open a new long trade for this model at the open.

The VSTpro just missed a new long signal yesterday. If the conditions are right then a buy signal today is within the odds.
The two most important conditions are:
- a low below the 14SMA which will be about 1400 today
- a high higher then the high yesterday 1415.56
This means a intraday volatility of about 1% which is quite possible in the current market situation.

An attentive reader of my posts will notice that my Kelly factor has gone negative today. This means that based on all the trades made in the last 250 trading session, the Kelly formula thinks that the risk is to high. This is caused by the large number of trades with a negative result. The end result is still positive but there where (in my strategy) 65 positive trades and 102 negative ones.
A negative Kelly formula results in a negative risk which means theoretical no trade at all.
I think this is just a momentary situation.

Meanwhile I am working on a review of the filter conditions. Based on the back tested performances of each model, I selected certain trades to ignore based on their poor results. This study was first done in August. I am reviewing these conditions to see if a adjustment should/could be made. Also I am looking at the XLX models that I have ignored so far (like XLE). There is a possibility that their trades over the last 4 months have improved their performance.

[manucastle]

The XLB model turned long again yesterday. I will open a new long trade for this model at the open.

The VSTpro just missed a new long signal yesterday. If the conditions are right then a buy signal today is within the odds.
The two most important conditions are:
- a low below the 14SMA which will be about 1400 today
- a high higher then the high yesterday 1415.56
This means a intraday volatility of about 1% which is quite possible in the current market situation.

An attentive reader of my posts will notice that my Kelly factor has gone negative today. This means that based on all the trades made in the last 250 trading session, the Kelly formula thinks that the risk is to high. This is caused by the large number of trades with a negative result. The end result is still positive but there where (in my strategy) 65 positive trades and 102 negative ones.
A negative Kelly formula results in a negative risk which means theoretical no trade at all.
I think this is just a momentary situation.

Meanwhile I am working on a review of the filter conditions. Based on the back tested performances of each model, I selected certain trades to ignore based on their poor results. This study was first done in August. I am reviewing these conditions to see if a adjustment should/could be made. Also I am looking at the XLX models that I have ignored so far (like XLE). There is a possibility that their trades over the last 4 months have improved their performance.

Thanks for your feedback PdP.

The current risk metrics for our type of investing are very valuable.

Trev

[Harry]

"... Also I am looking at the XLX models that I have ignored so far (like XLE). There is a possibility that their trades over the last 4 months have improved their performance."

PdP, I know you are aware of this already but just wanted to put this out there for others that consider building financial models. You should not backfit your model. True backtesting involves two stages: calibration and testing. One should always break their data into a minimum of two sections. Calibration determines the parameters for your model and the second is where you test the performance of those parameters. Consider 1/2 & 1/2, 1/3 & 2/3, or even 1/4 & 3/4 partitioning. If the XLX models did not far well during the calibration testing, using the past 4 months performance to calibrate will probably lead to poor future performance.

[pdp-brugge]

No changes

[manucastle]

The XLB model turned long again yesterday. I will open a new long trade for this model at the open.

The VSTpro just missed a new long signal yesterday. If the conditions are right then a buy signal today is within the odds.
The two most important conditions are:
- a low below the 14SMA which will be about 1400 today
- a high higher then the high yesterday 1415.56
This means a intraday volatility of about 1% which is quite possible in the current market situation.

An attentive reader of my posts will notice that my Kelly factor has gone negative today. This means that based on all the trades made in the last 250 trading session, the Kelly formula thinks that the risk is to high. This is caused by the large number of trades with a negative result. The end result is still positive but there where (in my strategy) 65 positive trades and 102 negative ones.
A negative Kelly formula results in a negative risk which means theoretical no trade at all.
I think this is just a momentary situation.

Meanwhile I am working on a review of the filter conditions. Based on the back tested performances of each model, I selected certain trades to ignore based on their poor results. This study was first done in August. I am reviewing these conditions to see if a adjustment should/could be made. Also I am looking at the XLX models that I have ignored so far (like XLE). There is a possibility that their trades over the last 4 months have improved their performance.

PdP.

We all know this is an extremely difficult market to trade and I think your risk metrics are doing a very good job of highlighting this fact !

This market is not for the feint hearted !

Trev:o)

[pdp-brugge]

Thanks Trev.

The odds for a buy signal for the VSTpro are growing. The high today is already higher then yesterday. If we get a low below 14SMA (+/- 1404.75) then the second condition is set. After that only a close above the 14SMA is sufficient to trigger the buy trigger. Let's wait and see.

[manucastle]

No changes

Hi PdP,

Would it be possible to get the update for today please.

Trev

[pdp-brugge]

sorry for this late update

[manucastle]

sorry for this late update

Thanks Pascal.

Trev

[pdp-brugge]

No changes

[pdp-brugge]

No changes

[pdp-brugge]

The XLB and XLI models have a Shorted Overbought signal as of yesterday.

The Kelly factor stays negative for the moment. My personal feeling about this situation is that the market is very difficult to trade under these circumstances. A lot of the recent trades (October and November) are a string of negative ones. This is not good for the way the Kelly Formula calculates the odds. Recently there have been a couple of wining trades but overall the result of my strategy is not prosperous.
The main indexes are cought between a decending 50MA and a rising 200MA. Until we get a decisive break out of that triangle we are currently in, I do not see large changes in the very near future.
The Nasdag is going to signal a death cross today, I presume.
I expect a decisive break out of the triangle very soon. This can change the uncertain situation and possibly offer some rewarding short trades.

[pdp-brugge]

No changes

[pdp-brugge]

No changes

[pdp-brugge]

Maybe premature but a buy signal from VSTpro is rather possible today: low (open) was below the 14SMA, the high of the the day is already higher then the previous high. All we need is a close above the 14SMA, which is about 1416.
Let's wait and see.

[pdp-brugge]

As expected, the VSTpro signaled a buy signal yesterday. This turns my Combo-MF model to LONG.
The Shorted Overbought signals for the XLB and XLI models where cancelled yesterday.
I will close the short trades for the XLB and XLI models and open a new long trade for the Combo-MF using TQQQ as vehicle.

Tomorrow, due to an external appointment, I will not be able to post an update until just before the open.

[pdp-brugge]

No changes

[pdp-brugge]

No changes

[pdp-brugge]

No changes

[pdp-brugge]

No changes

[pdp-brugge]

No changes

[pdp-brugge]

The XLI model signaled a Shorted Overbought yesterday. I will open a new short trade for this model at the open today.

[pdp-brugge]

The XLF model has a Shorted Overbought signal since yesterday. I will not follow this signal due to the poor back-tested performance with the combination of a XLF Shorted Overbought and a long 20DMF.

[pdp-brugge]

No changes

[pdp-brugge]

Happy New Year to everyone!

The XLI model turned to cash last Monday. I will close the current short trade for this model at the open today.
In my strategy, only the Combo-MF, XLK and XLY models are active as of the first trading day of the New Year and all in long mode.

[pdp-brugge]

No changes

[pdp-brugge]

No changes

[pdp-brugge]

The XLK model is now also in Cash. The only models that are active in my strategy are now my Combo-MF and XLY.

[pdp-brugge]

No changes

[pdp-brugge]

No changes

[manucastle]

No changes

pdp,

Thanks for all your work.

Your Kelly and NoFear metrics are indicating to stay out of the market but the market has had a decent rise since the 20DMF buy signal of 11/15/2012.

Does this change your view of these metrics ?

Thanks in advance.

Trev

[pdp-brugge]

Trev,

I agree with your remark. The market has been rather well performing since the long signal of the 20DMF on November 16.

My calculation for the Kelly Criterion is based on a moving window of 250 days. In the last 250 days, my strategy had a return of 39,39% non-cumulative. There where 59 positive trades and 99 negative ones. This large number of negative trades causes the current negative Kelly criterion. This has me puzzled.
Personally, I am still trading with a very small position size.

I must admit that I fear that there is something fundamental wrong with my position size calculations. Since November 16 there have been 16 trades in my strategy with a non-cumulative result of 8,81% based on the raw trades. With my position size technique, this positive raw result has turned into a negative growth of my equity of 0.51%. Frustrating!
I am working on a review of my position size technique and will report once I have found a better one with consistent equity growth over the now 3 years of trading signals that I have.

PdP

[pdp-brugge]

No changes in signals

[pdp-brugge]

No changes in signals

[pdp-brugge]

I am reviewing my strategy from the bottom up.

Examining the trades my strategy generated in 2012 delivered not a satisfactory result. Trading the XLX signals nor the Combo-MF signals gave me a decent impression. If I compare the results from the naked signals with the results of my strategy after implementing my position management rules, than it is even poorer. Although my strategy is still performing better than the S&P500, I do not have a good feeling about it.

I do not think there is something wrong with the signals. In the last weeks I have had a thoroughly review of my position sizing rules. Here the same: I have not found a serious error in my logic nor my programming.

I honestly think that the solution lies in expanding the number of models/instruments that I use. I am examining the use of currencies and commodities.

Until I have found a broader spectrum, I am postponing my daily posts.

PdP

[brrim]

PdP; I really appreciate all the work that you have shared and your daily posts. I look forward to seeing
where your work will lead you.
Best regards,
Robert