Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
Register for free to join our community of investors and share your ideas. You will also get access to streaming quotes, interactive charts, trades, portfolio, live options flow and more tools.
WLD, I'm also looking forward to Chris' study.
I assume the study was done around the same time Fosback was revising the system (2000-2001), it could just be a coincidence that the study was done in 2000. My problem with the in-sample period is 1987.
Given the historic run during the 90s, I would assume every day (M-F) would be a net gainer. That's why the results for Mondays are so shocking.
WLD, I think you misunderstood Fosback's rule concerning Mondays (avoiding the first trading day of the week during the end-of-month period). Mondays are bearish (see table below). My original point is this: How much abberant effect did the '87 crash have on these results? In other words, are Mondays really as bad as they look on this table?
TOTAL GAIN 1/31/85 - 12/20/00____________
NASDAQ NASDAQ NASDAQ
100 Composite Industrials
Monday (39.9)% (51.0)% (62.4)%
Tuesday +56.5 +4.9 (24.4)
Wednesday +399.4 +190.7 +187.0
Thursday +85.3 +102.7 +108.1
Friday +90.3 +176.6 +165.6
Total +591.6% +423.9% +373.9%
Ned Davis to be guest on Lou Rukeyser's Wall Street:
CNBC on July 4 at 8:30 pm ET
does that make sense?
Sort of. I understand and agree with your first point entirely. I don't know if I completely understand or agree with your second statement. I see very few "outliers" on the scale of the '87 crash. Yes, "sh-t happens", but the market is a lot less random, less volative than it seems to those of us who are too close to it (the ol' can't see the forest through the trees adage applies here).
Also, I agree that it makes sense to exclude both positive and negative days from the test. Interested to hear how you would test if Mondays are seasonally significant - what data would you test and would you exclude outliers?
WLD, I understand your point, but why would anyone want to use a system where the historical make-or-break performance is totally dependant on a single day? If I want to determine if Mondays are seasonally significant, the only way to know for sure is to exclude the most extreme days from the test. I never took a statistics course but this seems to be a logical approach to me.
Still working on my Sector Selector Spreadsheet:
I am currently using 100 days of data for each industry group in my Relative Strength Composite (RSC) indicator. After observing the results for a couple days, I have concluded that I'll need to increase the period substantially in order to use the rankings for intermediate term trading. I'll probably go with 200 days.
I am also including 3 other indicators based on suggestions and input from others. First, I want track the 10 day ROC of each sector's RSC so I can gauge which sectors are gaining or losing relative strength the fastest. Second, I'll track the 1-40-1 PPO of each sector index and use it as a deviation from trend indicator by generating OB or OS alerts. Lastly, I'll track sector/SPX ratios and their 1-40-1 PPO. This will allow me gauge the relative performance of each sector against the S&P 500.
I'll post another update next week.
Chris, I think the Keystone system is based on the changes Fosback made around 2000 or 2001. I presume Hulbert is continuing to track the older system from 20 years ago. Actually, I think the system described in Market Logic was also slightly different but Fosback updated it in the late '70s - early '80s.
The latest version accounts for the propensity for Mondays to underperform. This smacks a bit of curve-fitting if you include the 1987 period - If I were backtesting this system, I would intentionally exclude '87 from my sample. The latest system also attempts to reduce the number of round trip trades by lumping the end-of-year holiday periods together.
You should read Tim Hayes' book The Research Driven Investor. He outlines a system testing methodology that will help to minimize over-optimization and curve fitting. I continue to notice that most people who back test systems tend to use all the historical data they have and then optimize from there.
Hayes' suggests testing a system against a sample period, optimizing the system, then testing it again against an out-of-sample period. If the system holds up in out-of-sample testing, the next step is to test in in real time (NDR tests indicators in real time, sometimes for years before they deploy an indicator).
Good luck with your test - hope you'll post the results.
WLD, don't worry, I'm not a one indicator type of guy. On the other hand, Hulbert is pretty rigorous in tracking methods. Note I say "tracking", not "testing". What's interesting is these results aren't the derived from curve-fitting, data-torturing, or the over-optimization you frequently see when somebody is touting a trading system. These are the real-time performance results during the tenure of HFD.
What's also interesting is that end of month seasonality has a logical explaination - end of month cash flows to institutions and fund managers.
Chris, found this posted on Suite101:
By Mark Hulbert, CBS.MarketWatch.com
Last Update: 12:02 AM ET Jan. 8, 2003
ANNANDALE, Va. (CBS.MW) -- First the bad news. The stock market timing system followed by the Hulbert Financial Digest with the best 20-year record is no longer published.
Now the good news: It doesn't matter -- you can profit from it anyway. It is an entirely mechanical timing model, based on three simple rules. With knowledge of those three rules, it is possible to specify in advance when buy and sell signals will occur -- months, or even years, before hand.
I am referring to the so-called Seasonality Timing System. And I'm going to tell you what these three rules are in a moment.
But first I want to review the history of this remarkable timing system.
Its origin in the investment newsletter arena can be traced back to Market Logic in the mid-1970s. Norman Fosback, one of that newsletter's editors, had come up with this Seasonality Timing System while conducting research for his book "Stock Market Logic." Fosback devoted a chapter of that book to this timing model, and he regularly reported on it in the newsletter.
The model proceeded to have the best long-term risk-adjusted return of any the Hulbert Financial Digest has followed.
Consider the performance of a hypothetical portfolio that switched between the Wilshire 5000 index and 90-day T-Bills on this timing system's signals. Over the 20 years ending this past Dec. 31, this portfolio gained 13.5 percent annualized, in contrast to 12.3 percent for buying and holding.
But what is really impressive is that this hypothetical portfolio was 45 percent less volatile, or risky, than the overall market. In other words, this timing model outperformed the market by more than a percentage point per year with little more than half the risk.
To be sure, this timing system is not for everyone. It trades frequently, for example -- about 17 round trips per year, in fact. So it should be employed in a tax-free or tax-deferred portfolio, and only with investment vehicles for which there are no transaction costs (such as no-load funds that levy no penalties for short holding periods).
In the late 1990s, Time Warner (AOL: news, chart, profile) (now AOL Time Warner) bought Market Logic and folded into Mutual Funds Magazine. While this monthly magazine continued - episodically -- to report on the Seasonality Timing System, AOL discontinued the magazine altogether this past Fall.
As a result, as far as I can tell, the Seasonality Timing System no longer is published. To be sure, Fosback started another newsletter of his own in March 2002, Fosback's Fund Forecaster, and in it Fosback does recommend several different "Seasonality Model Portfolios." However, the seasonality model that Fosback employs appears to be significantly different from the one he originally recommended in Market Logic.
If you can find a newsletter that publishes Fosback's original seasonality timing system, I recommend that you subscribe to it if you want to follow that timing system. No doubt its editor can do a much better job of explaining that system and its rationale than can I.
However, since AOL Time Warner seems to have abandoned the publication of this information, what follows is my effort to make the information available until such time as some publication gets back into the business of providing this information to investors. I'm doing this primarily at the request of investors who are interested in the system -- and in the hope that its regular publication will be resumed and make my efforts unnecessary.
What follows are the three rules that make up the Seasonality Timing System. I retrieved them from the May 2001 issue of Mutual Funds Magazine, the last time I saw the magazine provide a full-scale review of this timing model.
The first two rules outline the two types of seasonality that the system hopes to exploit, while the third deals with exceptions:
To exploit positive seasonality around the turns of each month: Buy at the close of the third-to-last trading of each month, and sell at the close of the fifth trading session of the following month.
To exploit positive pre-holiday seasonality: Buy at the close of the third-to-last trading day prior to exchange holidays, and sell at the close of the last trading day before a holiday.
Exceptions: If the holiday falls on a Thursday, sell at Friday's close rather than Wednesday's. Also, if the last day before a holiday is the first trading day of the week, don't sell until the day after the holiday. Finally, never sell on the first trading day after options expire; instead wait an extra day.
Following these rules, the Seasonality Timing System became invested in stocks as of the close of the market this past Dec. 27 (the third to last trading day of December) and will get out at the close of the market on Wednesday, Jan. 8 (the fifth trading session of January). Through the close of Jan. 7, the Dow Jones Industrial Average ($INDU: news, chart, profile) has risen by more than 5 percent over this period.
Those following this timing model are now in cash. They will next be in the market between the close of Wednesday, Jan. 15, through the close of Friday, Jan. 17, immediately prior to the exchange holiday for Martin Luther King day on Jan. 20.
Stay tuned. I will continue to monitor this model and report on its performance.
Nope. I would see if you can get a recent copy of Hulbert Financial Digest at the Library or contact Fosback's newsletter. Found this on Fosback's site: http://www.fosback.com/
"In March 2001, The New York Times reported that a timing methodology with precisely those characteristics, known as the Seasonality System, "outperformed every other timing model . . . monitored over two decades." Times columnist Mark Hulbert particularly lauded the System for its "conservatism"; in other words, its ability to earn returns with significantly below-average risk."
Nasdaq Intermediate Index: Now at "0" (Dead Neutral)
Nyse Intermediate Index: -.53 (Moderately Bearish)
Culmus, thanks for the link. A great toolbox here.
Reminds me of a joke:
A man inquired about an ad from a surgeon who was promoting brain transplants. The surgeon had a variety of brains, all priced by the ounce. The doctor told the man the brains of a laywer sell for $1000 an ounce, the brains of a nuclear physicist go for $2,000 an ounce, and the brains of a stock market guru go for $20,000 and ounce.
The man exclaimed, "why are the brains of a stock market guru so expensive?" The surgeon replied, "do you have any idea how many stock market gurus it takes to get an ounce of brains?"
If you can turn $100K into $50K, we can easily get you a job as a fund manager.
Steve, Wealth-Lab should work. It's easy to register and free. You start with $100K. I'll post all the details next week.
Cool. I'll post my email when it's ready - drop me a note and I'll reply with an attachment. I'd like Augieboo to do the same. Basically, I just want your feedback.
I don't have the means to backtest this ranking system, but I thought it would be fun to test it in real time through paper trades. Each of us could use our own timing systems, but use the Selector Selector to pick our funds. We'll need to find an online portfolio simulator to log our trades: http://www.wealth-lab.com/ OR http://moneycentral.msn.com/investor/home.asp (Portfolio section) are two suggestions.
Steve, was wondering if you'd like to be a beta tester of my Sector Selector spreadsheet.
It uses a methodology similar to NDR's techno ranks - my spreadsheet calculates a relative strength composite for 21 sector indices (based on Profunds underlying indices) and ranks them accordingly. The best long candidates are at the top of the list, the best short candidates (via ETFs) are at the bottom. I should be finished with this project by the weekend. Interested?
Can you run a scan on ratios???
For example, I have a Favorites List composed of Dow Jones Sector Indices as a ratio to the S&P 500 (ie. $DJUSAR:$SPX). I want to run a very simple scan on this list but it always returns "0" results. Is it because the list isn't composed of pure stocks, mutual funds, or indices? If the answer is yes, why should it matter?
Augieboo - Ignore my previous Excel question. I solved the problem by creating a macro. Hopefully, by next week, my "sector selector" spreadsheet will be complete. You'll be the first to look at it, then I'll send copies to anyone who wants it.
My Intermediate Models continue to slide:
Nasdaq Model: -.27 (Mildly Bearish)
Nyse Model: -.33 (Moderately Bearish)
Current Equity Allocation: 30%L 70%C 0%S
Trend indicators continue to weaken, breadth is contracting from ultra-high overbought territory. On the positive side, T-Bond Momentum is still very favorable for stocks, the downward MO of the US dollar is begining to come back, AND the volatility indicators are settling down - if they continue to rise this week I'll score them as neutral.
My guess is the market is just catching it's breath before it sets sail to new highs, HOWEVER - I never, never (did I say "never"?) second-guess my models.
Augieboo, got an Excel question for you.
I'm working on the sector momentum project this week. I solved the first part pretty easily - using a web query to pull the index data into my workbook. So far so good...
I have 21 columns set up to keep the list of prices for each DJindex. Question: Is there an easy way to automatically add a new entry to the list each day without having to manually insert a new cell (to all 21 columns)?
For example, suppose the data for Friday's close is in cell A1. At the end of the day today, I want to update A1 to reflect today's price and push Friday's close to A2, the data in A2 to A3, etc. You would think this would be an easy thing to do dynamically in Excel, but after a good deal of searching, I haven't found the answer.
Note, I am referencing cells on sheet1 from the sheet my data is being displayed through the web query (sheet3). So, A1 is actually referencing the contents of a cell on sheet 3. It would be great if I could run the query and it would dynamically update all my lists on sheet1.
Thanks in advance for any tips!
My Nasdaq Intermediate Index turned bullish on 3/13.
My NYSE Intermediate Index turned bullish on 3/18, turned slightly bearish on 5/19, and was back to bullish by 5/28.
My Nasdaq Intermiediate Index just turned negative: Mildly Bearish at -.07
http://stockcharts.com/def/servlet/Favorites.CServlet?obj=ID368318
Steve, both my NYSE and NASDAQ Intermediate models have been moving downward lately. The NYSE model just hit zero, a dead neutral interpretation. Anything below zero is bearish. My Nasdaq model isn't much better: it registered a +.13 today - the lowest it's been in quite a while.
Why? Breadth has stopped expanding and is starting to contract. My Micro-trend indictors now hold the balance here. Another down day may be enough to put both models in negative territory.
Augie, sure thing. I probably won't have time to work on it until sometime next week.
BTW, I have another question for you. You've done a ton of great work in regards to market breadth indicators. Was wondering if I could have your take on how to best use breadth indicators (see point #2 below). I'm currently writing a stock market manifesto of sorts with a detailed description of how my intermediate models work.
Conceptually my models are based on six behaviors (tendencies, anomallies, principles, etc.) the stock market consistantly exhibits. These behaviors can be quantified as indicators which can be used to maximize returns by minimizing exposure to equities. The behaviors are:
1. Price Trends: Prices tend to trend. The best bet is to stay on the right side of the trend.
2. Breadth Trends: Trends in breadth tend to mirror trends in price. Beware of reversals when breadth trends reach extremes, HOWEVER, because you can never know how high is too high or how low is too low, don't buck the current trend. Also - When the breadth trend diverges from the price trend, there is a strong tendency for a reversal.
3. Price Momentum: Price momentum tends to lead market prices. The longer the divergence between price and momentum, the more likely a reversal will occur.
4. Breadth Momentun: Breadth thrusts (short-term periods of overwhelmingly positive breadth) tend to have bullish implications for intermediate/long term prices. Selling climaxes(short-term periods of overwhelmingly negative breadth) also tend to be bullish EXCEPT the market may experience more down side action over the short term.
5. Sentiment: Extremes in sentiment tend to signal risk of a reversal is growing. The longer the extreme exists, the more likely a reversal will occur. Note: The best bet is to wait until sentiment peaks or troughs. Anticipating sentiment extremes is risky.
6. Monetary Conditions: Falling interest rates and an accommodative monetary policy tends to be bullish for stocks. Don't fight the Fed.
Thoughts?
Augieboo, it's very similar except Pring uses KST primarily as a timing indicator for a single index or security. NDR uses this method as a ranking system - similar to relative strength except using multiple time periods.
More info on NDR momentum ranking systems:
Lance Stonecypher's chaper from "Being Right or Making Money":
"We call this ranking system are Techno Ranks. It's really a very simple formula that applies rates of change over a 26 to 52 week time frame to each stock in the S&P 500 and then ranks those stocks in descending order based on the results of the formula. The component rates of change used in this formula have been well documented by independent academic research.
[A table illustrates the performance of this system over a 27 year period. In practice, NDR combines sector momentum with the relative momentum of stocks. They have a Traders Buy List, basically a ranking system that selects the top 15 industry groups (out of 100)and the top 15% of stocks out of those groups (out of a universe of 2000 stocks)]
Believe me, I ain't bucking the trend...but the breadth trend indicators are begining to roll over. That isn't to say this market can't go higher, BUT - when the breadth trend begins to diverge from the price trend, the outcome is usually pretty ugly.
Both my Nasdaq and NYSE Composite models have been bullish for months now, but they're starting to slip.
Augieboo, I'm piecing together the NDR strategy from a variety of sources. In the book Research Driven Investor, Tim Hayes states:
"A single rate of change can be subject to misleading messages since a spike or temporary low in the data at the start of the period can produce a temporary high or low momentum reading. The sum of three rates of change will reduce the impact of an aberrant change."
[Note, Mr. Hayes is somewhat secretive/misleading in the statement above - he never mentions NDR's actual method, number of periods sampled, or the lengths of those periods. In the paragraph above, he mentions using 3 rates of change, however, I've read elsewhere that NDR actually samples ROCs in one day increments.]
"You should recognize that if your combination includes a relatively long rate of change, the momentum message will be dominated by that long rate of change since percent changes over long periods tend to be greater than percent changes over short periods. One way to de-emphasize the percent change of the longest momentum is to equal-weight your momentum sum by dividing each rate of change by its momentum period. If you were using rates of change of 10, 20, and 40 weeks, for instance, you would divide the 10-week change by 10, the 20-week change by 20, and the 40-week change by 40 before adding them together."
NYSE Composite Summation Index just turned bearish:
Don, my guess is that it wouldn't matter much whether you're using rydex or profunds. Rydex has 17, Profunds 22 - five more. I would assume sector funds from both families would behave similar to the DJ industry group indexes - except with leverage.
Thanks Augieboo! Thanks Hiker! You've given me some great leads. I'll let you know when I get this figured out.
Sector selection system - need help!
I posted this for Augieboo on the MDA board. Thought someone on this board could help:
I want to set up sector momentum system similar to what Ned Davis Research uses. Here's what I have in mind:
I want to gauge the relative momentum of each DJ sector index used as benchmarks for Profunds sector funds. I think there are 22 Profunds sector funds. First, I would need to calculate the ROC of each indice for several periods; 5 day ROC, 6 day ROC, 7day ROC….all the way to 60 days (I don't know the actual time periods NDR uses, but from what I gather, this would be a pretty good data sample for intermediate term cycles). The ROC for each period would need to be divided by the number of days in the period (this is so longer periods don't dominate the composite). Then, every period would be added together to form a momentum composite for each index.
Next, I want to rank the sectors by their momentum composite, from strongest to weakest every day. If my intermediate timing model is on a buy signal, I want to own the 4 sector Profunds that mirror the top ranked indices. These funds would be sold when they drop below the top 6 ranking OR if my timing model gives a sell signal. New funds would be bought as they enter the top 4. This ranking system could also be used to short ETFs by shorting sectors in the bottom rankings when my model is bearish.
Based on NDR's research and other academic studies I've read, I'm pretty confident this system is a winner. My challenge is in implementing it.
First, the only data I need is closing prices of each DJ sector index. On the surface, it sounds easy. But I don't want to be burdened with entering these manually every day OR paying a wad for data feeds. Do you know of an inexpensive but easy solution?
Second problem, I'm not a math wizard or software programmer and I'm a novice in Excel. I assume my system could be developed just using Excel (?), but I wouldn't have the faintest idea of how to do it. Any suggestions or tips would be appreciated. Do I need to take a class in Excel or is this so simple even an idiot like me could learn just by tinkering?
Augieboo, was wondering if I could pick your brain?
I want to set up sector momentum system similar to what Ned Davis Research uses. Here's what I have in mind:
I want to gauge the relative momentum of each DJ sector index used as benchmarks for Profunds sector funds. I think there are 22 Profunds sector funds. First, I would need to calculate the ROC of each indice for several periods; 5 day ROC, 6 day ROC, 7 day ROC...all the way to 60 days (I don't know the actual time periods NDR uses, but from what I gather, this would be a pretty good data sample for intermediate term cycles). The ROC for each period would need to be divided by the number of days in the period (this is so longer periods don't dominate the composite). Then, every period would be added together to form a momentum composite for each index.
Next, I want to rank the sectors by their momentum composite, from strongest to weakest every day. If my intermediate timing model is on a buy signal, I want to own the 4 sector Profunds that mirror the top ranked indices. These funds would be sold when they drop below the top 6 ranking OR if my timing model gives a sell signal. New funds would be bought as they enter the top 4. This ranking system could also be used to short ETFs by shorting sectors in the bottom ranking when my model is bearish.
Based on NDR's research and other academic studies I've read, I'm pretty confident this system is a winner. My challenge is in implementing it.
First, the only data I need is closing prices of each DJ sector index. On the surface, it sounds easy. But I don't want to be burdened with entering these manually every day OR paying a wad for data feeds. Do you know of an inexpensive but easy solution?
Second problem, I'm not a math wizard or software programmer and I'm a novice in Excel. I assume my system could be developed just using Excel (?), but I wouldn't have the faintest idea of how to do it. Any suggestions or tips would be appreciated. Do I need to take a class in Excel or is this so simple even an idiot like me could learn just by tinkering?
augieboo, thanks for the insight. This is the kind of market where the trend is truly your friend. Don't fight the system, just profit from it.
What's your take on total volume? SPX volume has been below it's smoothing 5 out of 6 days although the market's going higher. Any ideas?
Augie, thanks for the breadth update. I don't see breadth divergences developing like we've seen in past rallies ie. summer of '98. I wouldn't bet heavily on shorts during the next leg down unless I see the internals wither.
Hiker, I think it really depends on how bonds do during subsequent bears. The allocation considerations are primarily between stocks and bonds with cash as a safe haven.
I really don't mind a fund holding lots of cash during a bear market - capital preservation provides enormous compounding benefits over the long haul. Also note that the NDR tape model would probably give some exposure to stocks during bear market rallies. For example, they're currently overweight in stocks, although Ned thinks we're in a secular bear. In other words, I wouldn't have expected them to be 100% in cash from 3/00 to 10/9.
I wish NDR/Gabelli would provide some examples on how their model performed during out-of-sample periods and real time. Because their models are always being updated, don't know if that is even possible.
Thanks hiker!
Thanks for posting hightecheast.
Mr. Davis has an excellent track record of forecasting the markets and the economy. Personally, I hope his economic forecast is wrong this time.