Muck's Model

[muck]

And as of right now, last weeks call of long is off (by only 1 S&P point). Granted, there is still some time for the market to rally a bit. Not calling you out in any way, just stating what I see. Looking forward to next week though, with elections and Fed meeting, it could get interesting.

 

So, what exactly is it that you see? :confused:

[muck]

Barring an ENORMOUS selloff tomorrow, the model will be long again heading into this weekend.

[muck]

For those who are more technically inclined, here are the dates new signals were generated since March 22, 2002:

 

3/22/2002 Short ... SP500 = 1148.70 -- winner

8/2/2002 Long ... SP500 = 864.24 -- winner

9/6/2002 Short ... SP500 = 893.92 -- winner

9/20/2002 Long ... SP500 = 845.39 -- winner

9/27/2002 Short ... SP500 = 827.37 -- winner

10/4/2002 Long ... SP500 = 800.58 -- winner

11/15/2002 Short ... SP500 = 909.83 -- loser

12/6/2002 Long ... SP500 = 912.23 -- loser

12/20/2002 Short ... SP500 = 895.76 -- winner

12/27/2002 Long ... SP500 = 875.40 -- winner

1/10/2003 Short ... SP500 = 927.57 -- winner

1/17/203 Long ... SP500 = 901.78 -- loser

1/24/2003 Short ... SP500 = 861.40 -- winner

2/14/2003 Long ... SP500 = 834.89 -- winner

2/21/2003 Short ... SP500 = 848.17 -- winner

2/28/2003 Long ... SP500 = 841.15 -- winner

10/1/2004 Short ... SP500 = 1131.50 -- winner

10/15/2004 Long ... SP500 = 1108.20 -- loser

10/22/2004 Short ... SP500 = 1095.74 -- loser

10/29/2004 Long ... SP500 = 1130.20 -- winner

1/11/2008 Short ... SP500 = 1401.02 -- winner

1/25/2008 Long ... SP500 = 1330.61 -- winner

2/1/2008 Short ... SP500 = 1395.42 -- winner

2/8/2008 Long ... SP500 = 1331.29 -- winner

2/15/2008 Short ... SP500 = 1349.99 -- winner

3/7/2008 Long ... SP500 = 1293.37 -- winner

3/14/2008 Short ... SP500 = 1288.14 -- loser

3/28/2008 Long ... SP500 = 1315.22 -- winner

4/4/2008 Short ... SP500 = 1370.40 -- winner

4/11/2008 Long ... SP500 = 1332.83 -- winner

5/16/2008 Short ... SP500 = 1425.35 -- winner

7/11/2008 Long ... SP500 = 1239.49 -- winner

7/18/2008 Short ... SP500 = 1260.68 -- winner

10/24/2008 Long ... SP500 = 876.77 -- winner

11/7/2008 Short ... SP500 = 930.99 -- winner

11/14/2008 Long ... SP500 = 873.29 -- winner

12/19/2008 Short ... SP500 = 887.88 -- winner

1/16/2009 Long ... SP500 = 850.12 -- winner

1/23/2009 Short ... SP500 = 831.95 -- loser

1/30/2009 Long ... SP500 = 825.88 -- winner

2/6/2009 Short ... SP500 = 868.60 -- winner

2/13/2009 Long ... SP500 = 826.84 -- winner

2/20/2009 Short ... SP500 = 770.05 -- loser

3/13/2009 Long ... SP500 = 756.55 -- winner

4/3/2009 Short ... SP500 = 842.50 -- loser

4/10/2009 Long ... SP500 = 856.56 -- winner

5/8/2009 Short ... SP500 = 929.23 -- winner

5/15/2009 Long ... SP500 = 882.88 -- winner

5/22/2009 Short ... SP500 = 887.00 -- loser

6/5/2009 Long ... SP500 = 940.09 -- winner

8/13/2010 Short ... SP500 = 1079.25 -- winner

8/27/2010 Long ... SP500 = 1064.59 -- winner

9/3/2010 Short ... SP500 = 1090.10 -- loser

9/10/2010 Long ... SP500 = 1109.55 -- ...winning...

 

Including the current trade:

11 losers

43 winners

 

 

...pretty interesting...I'd never looked at this signal like that before...hmmm...

Edited November 5, 2010 by muck

[The Irish Doggy]

...pretty interesting...I'd never looked at this signal like that before...hmmm...

 

 

Indeed. What's the annual % yield over that time?

[muck]

A little over 30% / yr (before accounting for commissions, slippage, interest, dividends, fees, etc.).

[ljbrun]

I'm curious. Why are you always 100% in or out? I would think that the values that trigger your long or short signals would not always give you the same confidence level. If so, wouldn't there be value in risking more or less dollars, or using funds that are more or less leveraged, when your signal comes with a higher or lower level of confidence?

[muck]

I'm curious. Why are you always 100% in or out? I would think that the values that trigger your long or short signals would not always give you the same confidence level. If so, wouldn't there be value in risking more or less dollars, or using funds that are more or less leveraged, when your signal comes with a higher or lower level of confidence?

 

To be clear, it's not 100% in or out ... it's either 100% long or 100% short (not 100% long or 100% cash). FWIW.

 

Regarding "confidence levels" ... you can run the risk of over optimization, in which case you really have nothing. So, I'm trying to keep this robust ... do you really want to see all 50-some-odd different levels of signal strength and signal trend? I think that that is just noise. By being either "long" or "short", it allows for it to be a less emotional process, too (imo).

 

 

Good luck to you.

[ljbrun]

To be clear, it's not 100% in or out ... it's either 100% long or 100% short (not 100% long or 100% cash). FWIW.

Yeah, your OP was perfectly clear. Not sure why I typed in/out instead of short/long. Probably just lazy.

 

Regarding "confidence levels" ... you can run the risk of over optimization, in which case you really have nothing. So, I'm trying to keep this robust ... do you really want to see all 50-some-odd different levels of signal strength and signal trend? I think that that is just noise. By being either "long" or "short", it allows for it to be a less emotional process, too (imo).

 

 

Good luck to you.

Not trying to complicate things, understood. I have a Profunds account. So I was just thinking that if there were occasions where "the force is strong" I might be more inclined to choose an ultra fund.

 

Thank you for sharing this. I look forward to your weekly updates.

[muck]

I understand your point ... I just don't want to get too cute.

 

Said another way, I'd rather be generally right than precicely wrong.

 

And, said yet another way, just because you flip heads three times in a row doesn't mean you're going to get heads the fourth time. Nothing works every time it's tried.

[Brentastic]

What is the model projecting for next week?

 

Thanks for doing this btw

[muck]

Still long.

[Brentastic]

Still long.

Just so I have this right, your model is setup for a week long swing trade (entry on Friday and an exit the following week on Friday)? OR is it an entry (in this case long) on a Friday that you stay in until the signal reverses, which could be a minimum of 1 week or longer?

[muck]

New signal each week. One weeks' signal may be in the same direction of the previous weeks (or previous several weeks') signal.

 

The entry and exit is always at Friday close.

 

If you're trading futures, if the signal is the same from week to week, you may want to adjust the number of contracts to account for changes in account value (i.e., if you made money in the most recent week and the signal is the same, you may need to add some contracts ... and vice versa).

[muck]

Not trying to complicate things, understood. I have a Profunds account. So I was just thinking that if there were occasions where "the force is strong" I might be more inclined to choose an ultra fund.

 

Thank you for sharing this. I look forward to your weekly updates.

 

You're welcome.

 

For giggles, I went back to take a look at adding a "the force appears to be strong" aspect of this and here's what I found (in VERY simple logic):

 

1372 weeks (including the week to come)

 

Original Model (as shown above):

1208 weeks long (average market performance during those weeks of +0.32%)

163 weeks short (average market performance during those weeks of -0.90%)

 

"Tweaked" Model (per your request):

634 weeks long on leverage (average market performance during those weeks of +0.46%) ... I am calling this a "2" ...

575 weeks long w/o leverage (average market perforamnce during those weeks of +0.18%) ... I am calling this a "1" ...

163 weeks short (average market performance during those weeks of -0.90%) ... I am calling this a "-1" ...

 

********************************

 

This Friday's signal is a "long on leverage" type of signal (aka, a "2").

 

This "2" signal has been unchanged since the week ending September 10, 2010. The market is up about 8% since then.

 

********************************

 

Good luck to you.

Edited November 12, 2010 by muck

[Brentastic]

You're welcome.

 

For giggles, I went back to take a look at adding a "you REALLY need to be long" aspect of this and here's what I found (in VERY simple logic):

 

1372 weeks (including the week to come)

 

Original Model (as shown above):

1208 weeks long (average market performance during those weeks of +0.32%)

163 weeks short (average market performance during those weeks of -0.90%)

 

"Tweaked" Model (per your request):

634 weeks long on leverage (average market performance during those weeks of +0.46%) ... I am calling this a "2" ...

575 weeks long w/o leverage (average market perforamnce during those weeks of +0.18%) ... I am calling this a "1" ...

163 weeks short (average market performance during those weeks of -0.90%) ... I am calling this a "-1" ...

 

********************************

 

This Friday's signal is a "long on leverage" type of signal (aka, a "2").

 

This "2" signal has been unchanged since the week ending September 10, 2010. The market is up about 8% since then.

 

********************************

 

Good luck to you.

So to dummify - the signal for next week is essentially a 'strong buy'?

 

It's interesting because I am trying to time the 'biggie' (the major decline I keep forecasting that WILL happen) and there are essentially 3 scenarios that could play out for the coming weeks and more.

 

Scenario 1 is that the market topped last Friday and that the next major phase of the bear market, Primary wave 3 (circle) down is underway, a decline that should draw the Dow thousands of points lower over the coming year or more. This scenario can also be defined as the 'biggie'.

 

The 2nd scenario is that we have begun a several week decline that should reach the July lows (1011 S&P and 9614 DOW) or lower. Once these lows are reached we will then rebound to a new recovery high (that tops last Friday's highs) which will complete this bear market rally that started in Mar 09. After this new high, we will decline thousands of points in the dow.

 

The 3rd and final scenario is a multi-week net-sideways move that swings back and forth well into December. Once the sideways move is complete, prices will rebound to a new recovery high which will end this bear market rally. Like the other 2 scenarios above, the next phase will see a few thousand points shed from the dow.

 

All of these scenarios have the same final outcome in the coming years - a major stock market decline that will see the Mar 09 lows surpassed and then some. It's interesting combining your signals to my bigger picture view. IF your signal is right for next week, then it looks like the 3rd scenario could be in play. Although your signal has been off the last 2 weeks, it seems to have a pretty high success rate so next week could be an increase in price. It will be very interesting to see how the next month or so pans out. And I'm very interested in your signals so I hope you keep posting them. Thanks again, much appreciated

[muck]

A point of order ... net, over the past two weeks the market is up and this model has been long.

[Brentastic]

A point of order ... net, over the past two weeks the market is up and this model has been long.

You're right, my mistake. Last 3 weeks, model has been long, long, long - resulting in scratch, profit, loss, BUT overall net profit. I looked at a small sample from the historical data above (new signal) and it appears to need a trend confirmed before it gives a new signal. In other words, it doesn't seem to have a forecasting ability on trend reversal, rather confirming a larger trend and being one with the trend. That is actually a good thing for investing, IMO and my observation/presumption is not a knock on your model.

 

The 'new' signal success rate is pretty impresive.

 

E2A: So, is a 2 rating essentially a strong(er) buy (than normal)?

 

E2A2: Also curious about the results of this model from late April through early August this year. Being long into the flash crash would devastate a portfolio.

Edited November 13, 2010 by Brentastic

[muck]

You're right, my mistake. Last 3 weeks, model has been long, long, long - resulting in scratch, profit, loss, BUT overall net profit.

 

Glad we got that sorted out.

 

I looked at a small sample from the historical data above (new signal) and it appears to need a trend confirmed before it gives a new signal. In other words, it doesn't seem to have a forecasting ability on trend reversal, rather confirming a larger trend and being one with the trend. That is actually a good thing for investing, IMO and my observation/presumption is not a knock on your model.

 

I am not sure I understand your point.

 

If you are saying that it doesn't appear that my model would ever generate a signal today that says that you should go short (or long) ______ weeks into the future, that is correct. It is not a cycle-based model.

 

However, if you're saying that a trend needs to be established before giving a new signal, I would disagree. Crtain aspects of the technical component of the model are decidedly counter-trend in nature. As "Exhibit A" I would provide the frequent trading during the down markets of 2002 and 2008/2009.

 

The 'new' signal success rate is pretty impresive.

 

Backtests tend to look wonderful. While I did the out of sample and multi-periodicity tests one would expect, nothing is perfect and nothing works forever. Caveat emptor.

 

E2A: So, is a 2 rating essentially a strong(er) buy (than normal)?

 

Yeah, I guess so.

 

The way I have it set up in my spreadsheets would allow someone to pick a targeted max notional exposure and to size positions around (i) the max notional exposure and (ii) the account size. A "2" would result in having 2x the targeted notional exposure.

 

E2A2: Also curious about the results of this model from late April through early August this year. Being long into the flash crash would devastate a portfolio.

 

It would only devestate a portfolio if the leverage was extraordinary and/or if the diversification was minimal.

 

As posted earlier in this thread, the "basic" model (i.e., the one originally posted) was long the entire period from the Friday close on January 1, 2010 through the Friday close on August 20, 2010.

 

In looking at when it would have scored a "1" or a "2" ... for 2010 ...

 

Trade Date:

1/8/2010 "1" ... which stays a "1" until 2/26/2010 ... during this period, the S&P500 went from 1144.98 to 1104.49

2/26/2010 "2" ... which stays a "2" until 3/12/2010 ... during this period, the S&P500 went from 1104.49 to 1149.99

3/12/2010 "1" ... which stays a "1" until 3/26/2010 ... during this period, the S&P500 went from 1149.99 to 1166.59

3/26/2010 "2" ... which stays a "2" until 5/14/2010 ... during this period, the S&P500 went from 1166.59 to 1135.68

5/14/2010 "1" ... which stays a "1" until 6/18/2010 ... during this period, the S&P500 went from 1135.68 to 1117.51

6/18/2010 "2" ... which stays a "2" until 7/2/2010 ... during this period, the S&P500 went from 1117.51 to 1022.58

7/2/2010 "1" ... which stays a "1" until 8/20/2010 ... during this period, the S&P500 went from 1022.58 to 1071.69

8/20/2010 "-1" ... which stays a "-1" until 8/27/2010 ... during this period, the S&P500 went from 1071.69 to 1064.59

8/272010 "1" ... which stays a "1" until 9/3/2010 ... during this period, the S&P500 went from 1064.59 to 1090.10

9/3/2010 "-1" ... which stays a "-1" until 9/10/2010 ... during this period, the S&P500 went from 1090.10 to 1109.55

9/10/2010 "2" ... which stays a "2" through today... during this period, the S&P500 has gone from 1109.55 to 1199.21

 

Excluding commissions, etc.:

Long Only from 1/8/2010 to 11/12/2010 = +54.23 S&P500 points

Trading Original Signals from 1/8/2010 to 11/12/2010 = +29.53 S&P500 points

Trading "Weighted" Signals (i.e., you're 2x as long as normal when the signal is "2") from 1/8/2010 to 11/12/2010 = -10.26 S&P500 points

 

***************************************

 

ETA: The "real value" of this sort of thing (vs. the long-only portfolio) really comes in down markets (and being long on leverage in strongly rising markets). To wit:

 

From Friday 12/28/2007 through Friday 1/2/2009, the S&P500 was -37%. During the same period, the "original" signal was +143% and the "occasional strong buy" signal was +186%.

 

From Friday 1/2/2009 through Friday 1/1/2010, the S&P500 was +20%. During the same period, the "original" model was +52% and the "occasional strong buy" signal +89%.

Edited November 13, 2010 by muck

[muck]

For grins, I ran some more statistics on the "the force appears to be strong" model covering the period from March 22, 2002 through November 5 ,2010 in eleven equal 41 week periods. Here's what I found:

 

Period 1 -- March 22, 2002 through December 27, 2002

2 Score = 15 weeks @ +0.55% / week average

1 Score = 0 weeks

-1 Score = 26 weeks @ -1.47% / weeks average

 

Period 2 -- January 3, 2003 through October 10, 2003

2 Score = 23 weeks @ +0.33% / week average

1 Score = 12 weeks @ +0.47% / week average

-1 Score = 6 weeks @ -1.02% / week average

 

Period 3 -- October 13, 2003 through July 23, 2004

2 Score = 16 weeks @ +0.26% / week average

1 Score = 25 weeks @ -0.09% / week average

-1 Score = 0 weeks

 

Period 4 -- July 26, 2004 through May 6 ,2005

2 Score = 18 weeks @ -0.00% / week average

1 Score = 20 weeks @ +0.21% / week average

-1 Score = 3 weeks @ +0.36% / week average

 

Period 5 -- May 9, 2005 through February 17, 2006

2 Score = 20 weeks @ +0.21% / week average

1 Score = 21 weeks @ +0.15% / week average

-1 Score = 0 weeks

 

Period 6 -- February 21, 2006 through December 1, 2006

2 Score = 22 weeks @ +0.13% / week average

1 Score = 19 weeks @ +0.19% / week average

-1 Score = 0 weeks

 

Period 7 -- December 4, 2006 through September 14, 2007

2 Score = 18 weeks @ +0.27% / week average

1 Score = 23 weeks @ -0.05% / week average

-1 Score = 0 weeks

 

Period 8 -- September 17, 2007 through June 27, 2008

2 Score = 17 weeks @ +0.28% / week average

1 Score = 8 weeks @ -0.09% / week average

-1 Score = 16 weeks @ -1.64% / week average

 

Period 9 -- June 30, 2008 through April 10, 2009

2 Score = 15 weeks @ +0.66% / week average

1 Score = 0 weeks

-1 Score = 26 weeks @ -1.93% / week average

 

Period 10 -- April 13, 2009 through January 22, 2010

2 Score = 27 weeks @ +0.33% / week average

1 Score = 11 weeks @ +0.33% / week average

-1 Score = 3 weeks @ +0.30% / week average

 

Period 11 -- January 25, 2010 through November 5, 2010

2 Score = 20 weeks @ +0.02% / week average

1 Score = 19 weeks @ +0.54% / week average

-1 Score = 2 weeks @ +0.56% / week average

 

***************************

 

Entire 451 Week Period from March 22, 2002 through November 5, 2010

2 Score = 212 weeks @ +0.26% / week average

1 Score = 158 weeks @ +0.17% / week average

-1 Score = 82 Weeks @ -1.43% / week average

 

***************************

 

I continue to find this little signal to be pretty interesting...

 

Good luck to you.

[Azazello1313]

 

[muck]

Long ... score is "2" ...

 

IOW, unchanged since Friday, September 10, 2010.

 

Caveat emptor.

[muck]

Market closed last Friday at 1199.21 and closed today at 1199.73 ... a whopping gain of +0.04%.

 

However ... by closing above last Friday's close, the model actually clocked in at score of "1" (not a "2"). Had the market had a down week, we'd still be at a "2".

 

ETA -- For Brentastic's info ... the model is a long way from triggering a short sell.

ETA 2 -- This is a bit of an unusual combination of signals I'm looking at here ... Specifically, since July 27, 1984, the combination of signals I'm looking at has only occured 12 times; the average market return in those weeks was -0.44%. Since March 22, 2002, this combination has occured three times; the average market reutrn in those weeks was +1.09%. The last signal like this was all the way back at the beginning of 2006 (and that week, the market was up +1.76%). ...we'll see...

Edited November 20, 2010 by muck

[muck]

The score for this coming week is a "1" (iow, "buy"). I do not expect this to change at the close like it did last week.

[muck]

The score for this coming week is a "1" (iow, "buy"). I do not expect this to change at the close like it did last week.

 

...same as last week... it looks like we'll have a "+1" score again

[Perchoutofwater]

Asked in another thread that you may not have looked at, but if the capital gains taxes revert to the pre-shrub rate, do you see a significant sell off in the market the latter part of this month? If so, any guess on how low it would go and when it would be back to where it is now?