No. of Recommendations: 26
Jim- Thanks for the methodology-
here's another one, that fairly simple.

1) Take the 40 week moving average (200 day) of the S&P 500
2) If this week's figure is better than that of 2 weeks ago,
then buy (or hold, if in already).
3) If less, sell or wait it out.

Back to 1997, you would have done about 50.6% better than
buy and hold. About 7 switches over that time period.

Now, if you are into the ETFs that do the double &/or
inverse, here is a table of the theoretical gains.

For example, if rule 2 applies and you buy the S&P ETF,
and when rule 3 applies, you buy the double inverse; you'll
end up with 133.5% better than S&P buy and hold.

 % Better (worse) than B&H

SPY (1X) SSO (2X)
50.6% 181.1%
SH (-1X) -10.1% 99.7% 272.7%
SDS (-2X) 5.2% 133.5% 335.7%



Note: none of the figures incluce any interest obtained if
out of the market nor transaction costs.

See the following for doubles and inverses of the S&P500 ETF-
SPY,SSO,SH,SDS
http://seekingalpha.com/article/36908-leverage-and-inversion...
Print the post Back To Top
No. of Recommendations: 0
Quite an interesting study. Because you have often examined SP500 data by comparing equal-weighted versus cap-weighted performance, I wonder: have you tried that with respect to this particular signal?

hirundo
Print the post Back To Top
No. of Recommendations: 26
Jim- Thanks for the methodology-
here's another one, that fairly simple.

1) Take the 40 week moving average (200 day) of the S&P 500
2) If this week's figure is better than that of 2 weeks ago,
then buy (or hold, if in already).
3) If less, sell or wait it out.

Back to 1997, you would have done about 50.6% better than
buy and hold. About 7 switches over that time period.

Now, if you are into the ETFs that do the double &/or
inverse, here is a table of the theoretical gains.

For example, if rule 2 applies and you buy the S&P ETF,
and when rule 3 applies, you buy the double inverse; you'll
end up with 133.5% better than S&P buy and hold.

 % Better (worse) than B&H

SPY (1X) SSO (2X)
50.6% 181.1%
SH (-1X) -10.1% 99.7% 272.7%
SDS (-2X) 5.2% 133.5% 335.7%



Note: none of the figures incluce any interest obtained if
out of the market nor transaction costs.

See the following for doubles and inverses of the S&P500 ETF-
SPY,SSO,SH,SDS
http://seekingalpha.com/article/36908-leverage-and-inversion...
Print the post Back To Top
No. of Recommendations: 2
have you tried that with respect to this particular signal?

It doesn't work on the S&P equal weight total return index since 1970.
Returns are high in both bear and bull states, and a bit higher in bear times.
One other contributing factor may be that it's bullish too much of the time---84.5% of the time.


But if my theory is right, it has to be the index that people are watching.
For example, in the UK you'd want the FTSE 100, and in Japan you'd want the Nikkei 225.
On the Nikkei, bullish periods have been 4.5% higher CAGR than bearish periods since 1989 (though both have been negative!).
For the FTSE, bullish periods have been 16.7% higher CAGR than bearish periods since 1990.
For the CAC40 in France, bullish periods have been 18.9% higher than bearish periods since 1993.
All of these are with the exact same 99-day rule, though using closes rather than intraday highs.

The rule has not provided much value in the last 10-20 years for the
Dow, which I speculate is because the Dow has fallen from prominence
among the bulk of investors. New recent highs are no longer noticed as much.

On the other hand, it could all be a fluke. But, it does pretty well in
quite a large number of separate bear markets, and it makes some sense.
Incidentally, overall CAGR during bullish phases since 1930 (index only) is
+10.2% versus -7.9% for bearish periods. The distinguishing ability has been
pretty good through the years, too. It "worked" (bullish return higher
than bearish return) in most of the 16 separate 5-year periods, with the
exceptions of 1975-1980, 1985-1990, and 1990-1995. The latter two were
probably because it was only bearish a very small portion of each
of the two periods (8-9% of the time), and the overall returns were
just fine as a result, at around 16.8% for that decade. In the
1975-1980 it was a more pure failure: the signal was bullish a more
typical 61% of the time, bullish returns CAGR +6.5% and bearish +11.7%.
Still, the bullish periods were positive, so the five year period was
again positive, with 12% CAGR overall. Interest rates were high then.

Jim
Print the post Back To Top
No. of Recommendations: 11
Jim- Thanks for the methodology-
here's another one, that fairly simple.


Yes, I've used a lot like that. The difference is subtle, in that it's more
of a trend follower rather than a bear market detector: with 87% more
signals per year, it lets you out of the market a lot more easily during a
bull market, which is rarely worthwhile. This shows up as a less capable
distinction between bull and bear periods. Compared to the "new recent high",
the bullish periods on your signal are 1.7% lower CAGR, and the bearish
priods 6.6% higher CAGR, plus the nuisance of more signals.

Mine isn't the greatest signal in the world, but it's the best I've
found with so few signals---it's very hesitant to make you jump in and out.
There was only one short "erroneous" cash period in the 2002-2007 bull, for example.
By construction, you can't have a short term bull whipsaw, as all
bull signals have to be at least 100 days long. You can have
a short term bearish whipsaw, but you can't lose money per se during it
if you're in cash, you can only forego profit.

Jim
Print the post Back To Top
No. of Recommendations: 1
For whatever it's worth, I've often mentioned that I consider Berkshire
Hathaway more like an asset class than a stock. Does it trend like the market?

Using this signal on Berkshire's share price since 1990, bullish periods
have been 9.6% higher CAGR than bearish periods. But, most of the
advantage was in the early years. It hasn't added value in the last decade.

Jim
Print the post Back To Top
No. of Recommendations: 0
tpoto,

A couple of followup questions:

1. I am assuming that you check once every 2 weeks and disregard what happens in-between; correct?

2. How did you arrive at the 2 week number? Is there an automated way to check for different values of N?

3. Ditto for 200-day MA. I wonder if there is a mound of toast...

Thanks!
-dr.nonlinear-
Print the post Back To Top
No. of Recommendations: 2
dr. nonlinear-
The excel program looks every week at the value
from two weeks earlier (and compares is with the current week).

I did not do any maximizing for both of the look back periods,
just a few trials to get something simple. I suppose I
could write a macro to do it (when golf season is over...)
Print the post Back To Top
No. of Recommendations: 17
For those of you who don't feel like doing a spreadsheet to calculate this,
I think this is right---
http://stockcharts.com/h-sc/ui?s=$SPX&p=D&yr=1&m...

If the last movement in the upper line was upwards, stay in the market.
If the last movement was downwards (as is the case now), go to cash.
Conveniently, you can also see how much the market would have to rise
for it to become a bull market according to this signal.

Jim
Print the post Back To Top
No. of Recommendations: 0
Jim,

Nice system.

In fact this cash period is now guaranteed to be profitable, since the "99 day high" line is below the point that we left the market---we will get back in again lower than where we got out in May, though we
don't yet know when that will be, so we will definitely be ahead
of the people who stayed in the market the whole time.


I'm struggling to understand this bit. Also your forward looking line appears dodgy.
For example, if the S&P jumps to 1460 tomorrow, it seems that your statement above would not be true.

Or am I in error?

Steve
Print the post Back To Top
No. of Recommendations: 0

But if my theory is right, it has to be the index that people are watching.
For example, in the UK you'd want the FTSE 100, and in Japan you'd want the Nikkei 225.
On the Nikkei, bullish periods have been 4.5% higher CAGR than bearish periods since 1989 (though both have been negative!).
For the FTSE, bullish periods have been 16.7% higher CAGR than bearish periods since 1990.
For the CAC40 in France, bullish periods have been 18.9% higher than bearish periods since 1993.
All of these are with the exact same 99-day rule, though using closes rather than intraday highs.


Jim,
Thank you for the grate post and well detailed description.
It looks like you have considered many variations on this timing system.
Have you looked at using this method in a port with all (S&P500, FTSE100, Nikkei225, CAC40, BRK, Bearx, etc.) equally weighted (20% each etc) holding the index or cash whenever the time system indicates?
Best Regards

Syvash
Print the post Back To Top
No. of Recommendations: 0
I'm struggling to understand this bit. Also your forward looking line appears dodgy.
For example, if the S&P jumps to 1460 tomorrow, it seems that your statement above would not be true.

Or am I in error?



You're right, I goofed, it would not be true. I implicitly made the
assumption that the market would not soar at an breathtaking pace.
So, this means I ignored the one day lag, and made the assumption that
the market isn't going to be able to catch up to the 100 day line before
it's below the 1326 level. The 1-day lag assumption is merely assuming
the market won't rise 20% overnight, which is reasonable.
The other assumption is an error, though I think probably still true.
The market would have to rise to 1326 (the exit level) before the 99 day
line gets down below that on Nov 13th. That would be a 14.2%
rise in 31 trading days, which is a bit of a reach but not impossible at all.
It would be the 98.5th percentile of performance for a period that long.
The odds are probably better than 1.5% since it's starting from
a sharply oversold level.

Jim
Print the post Back To Top
No. of Recommendations: 0
Jim. I'm a newbie and am very interested in your system but am struggling with a few points. I'm understanding what triggers converting to cash but I'm having trouble understanding what triggers a buy in. I'm looking at your chart:

http://stockcharts.com/h-sc/ui?s=$SPX&p=D&yr=1&m......

If I am understanding this, if the S&P had jumped to 1450 around June 1st, that would have triggered you to go back into the market, right?

This may be an oversimplification but it appears that whenever the S&P line intersects your "high" line pushed out 100 days, you buy back in. Is that correct?

I really appreciate your input. I'm not experienced in the stock market but I pulled half of my 401K into a stable fund on Sept 29th and am trying to find some methodology on how to get back in!
Carl
Print the post Back To Top
No. of Recommendations: 0
Oops. Just realized the link didn't work. The chart I'm looking at is

http://stockcharts.com/h-sc/ui?s=$SPX&p=D&yr=1&m...

Carl
Print the post Back To Top
No. of Recommendations: 6
tpoto writes:
1) Take the 40 week moving average (200 day) of the S&P 500
2) If this week's figure is better than that of 2 weeks ago,
then buy (or hold, if in already).
3) If less, sell or wait it out.

Because SMA = (sum of n values)/n, and n is invariant, the rule boils down to:
Buy/Hold iff (c0 + c1)>(c41 + c42)
where cn is the close n weeks ago.
Pretty simple, with no need to do any averaging. You can probably imlement it by eyeball.

Just an observation.
Print the post Back To Top
No. of Recommendations: 0
Jim - just trying to catch-up and so a reply to an old post. The update for the chart referred to in your post has the upper line (in fact, all three lines) currently flat. I assume that means you would stay whatever course you were currently on, correct? And since the course was bullish, the last change in the upper line being upward,you would be in the market. Do I understand correctly?

Is there importance to the other two lines that I need to understand.

Appreciate your patience as I try to catch up.

zaksdad

P.S. Do you mid that I email things like this to you since it is a replay to a post that is 16 months old? If so, let me know and I will stop.
Print the post Back To Top
No. of Recommendations: 0
Mungofitch,
How has this strategy performed since you reported it in 2008?
adiabatic
Print the post Back To Top
No. of Recommendations: 5
Mungofitch,
How has this strategy performed since you reported it in 2008?
adiabatic


A little trip down memory lane. The 2008 OP (original post) is Jim's first proposal (at least on this board) of the 99 day rule. You can see all his posts using the DataHelper (http://www.datahelper.com/mi/search.phtml) and doing an "Exact Match" on "99 day."

Zeelotes renamed it "Dying Bullish Euphoria (DBE)" (http://boards.fool.com/mr-macaroni-asked-its-very-interestin...), which was BCIII in the 3 bear catchers he had implemented.

Robbie then implemented the BCC timing signal in gtr1 (http://boards.fool.com/gtr1-and-bear-catchers-30893319.aspx).

So how has it worked? You can test it yourself using gtr1 with just BCIII enabled. I think this is the URL for the S&P 500 with BCIII.

http://gtr1.backtest.org/2013/?gprc%281%29gt0:cashif:BCC:et0...

Charlie
Print the post Back To Top
No. of Recommendations: 9
Mungofitch,
How has this strategy performed since you reported it in 2008?
adiabatic


It has been bullish 84.4% of the time in the 7.11 years since the first post.

The CAGR during the bullish moments has been +13.2%/year, total return 110.9%.
The CAGR during the bearish moments has been -4.2%/year, total return -4.6%.
(as you can see from those figures, it has been bearish a total of about one year)

A long/cash portfolio has had an ulcer index of 5.40% versus 10.56% for buy and hold (assuming no interest earned).
A long/cash portfolio has had a worst drawdown of -17.3% versus -43.2% for buy and hold (assuming no interest earned).

$1 turned into $2.11 for long/cash switch, or $2.01 for buy and hold.
Total of 6 signals = 3 round trips.

So, all in all, I'd say a high level of value relative to the conveniently low number of signals.
More or less everything is about in line with what the backtest suggested.
Very modest performance increase, substantially better risk metrics, about a signal a year on average.

Jim
Print the post Back To Top