Hi,One of the better performing newsletters in the Hulbert Financial Digest is the Contrarian's View. It's "NYSE Timer's Trend portfolio" is highly rated for the past 5 and 10 years considering overall performance as well as for risk adjusted returns. This is strictly a timing signal and has no recommended stocks. One metric that stood out is the 10 year annual performance of 12%, since that period includes much of the "lost decade". Last 6 months 11%, last 5 years 12%, last 10 years 12%, and overall annual return since 12/31/01 is 10%Its author, Nick Chase, states " 'Timer's Trend' is based on 4% and 10% exponential moving averages of the New York Stock Exchange advance/decline lines (that is, the ratio of advancing to declining stocks)." Further he comments about this signal "We're on a BUY signal as of June 15, 2012."Because of this success, I tried to backwards engineer his signal and evaluate it. Although I can hypothesize, I am not familiar with advance decline lines being defined in "percents" and cannot figure out how to construct the 4% and 10% scores which produced a buy on June 15th. I have tried out some exponential moving averages at Stockcharts.com using the ticker, $NYAD but have had no luck. Can anyone help?Newsletter: http://contrariansview.org/Site/About_Me.htmlResults from 1982 to present, including an explanatory footnote at the end of the page: http://contrariansview.org/TT1982onward.htmlNow, at almost 1:00 a.m. it is sleepville for me. Thanks,Elliot
He posts the formula in BASIC right on his website.
I am not familiar with advance decline lines being defined in "percents" and cannot figure out how to construct the 4% and 10% scores ...The 4% and 10% probably do not refer to the AD lines, but to theconstruction of the EMA.If you look at the formula for any EMA, there is a key percentage used, call it P. EMAday2 = (P*(inputvalueday1-EMAday1))+EMAday1That percentage is generally calculated from a notional number of days N as P=2/(1+N), but if you use the percentage directly that works.Thus,His 10% is presumably the same as a "19 day" EMA.His 4% is presumably the same as a "49 day" EMA.As you can see from the formula, there is no specific lookback windowand referring to an EMA by its number of days N of smoothing is a bit misleading.Using the daily percentage weight factor P directly actually makes much more sense.It looks like the larger difference in his construction is that heis applying the EMAs to the A/D line, not the daily difference of A-Dwhich would be a close analogy of the NH-NL signals.An A/D line is normally constructed as "yesterday's A/D + today's advances - today's declines", so it's an integral of the daily difference.This might (?) be like what he's using.http://stockcharts.com/h-sc/ui?s=$NYAD&p=D&yr=0&...Note the very important "cumulative" selection in the chart type pulldown.I expect he's interpreting signals as crossover dates of the EMAs, buthis use of the word "confirmation" leaves a tiny little doubt in my mind.Note, I'm very leery of using breadth based indicators based on theNYSE because the majority of things listed there in recent yearsare not simple US companies. Note how different the Nasdaq equivalent looks because there aren't any bond funds giving strength to the upswing:http://stockcharts.com/h-sc/ui?s=$NAAD&p=D&yr=0&...Jim
EMAday2 = (P*(inputvalueday1-EMAday1))+EMAday1
He posts the formula in BASIC right on his website. Saw this comment in the code: 20 REM recoded from VMS BASIC That's impressive...I haven't seen any real references to VMS in nearly 20 years...
Thanks, I never would had thought that 4% & 10% would be equal to 19 & 49 days. Yes, the NASDAQ and NYSE provide signals at different times and the length of time of the signals differs also. Also, for the time period of the charts, the NASDAQ provides more signals than the NYSE. Would I be correct in saying that for a signal to be true, the daily must past by both the 19 and 49 exponential moving averages?Thanks again,Elliot
Here's a timing indicator based on the strength of the Nasdaq relative to the S&P500. http://www.visalert.com/sample/apr01%20testing%20confirms%20...I haven't tried to replicate the results, but I'd be interested if someone out there does.TLF
Best Of |
Favorites & Replies |
Start a New Board |
My Fool |