Thursday’s Breadth Report
Thursday, March 18, 2010 at 10:53 pm
Breadth expansion slowed today but showed no real divergence from price.
The raw advancers – decliners measure is giving another buy signal as the decliners line (red) has closed above the upper Bollinger Band (purple). We are currently long already from the same signal 3 days prior, but the signal is valid, nonetheless.
The 52 week new highs new lows indicator pulled back today. This type of pullback is an anecdotal precursor to a more significant pullback in price.
As breadth expansion has slowed with all measures near their historical high ranges, I am considering initiating a small short position on the open. This may be early, and breadth is not showing any major divergences from price.
You discretionary bastard
My timing was impeccable though!
Hi Wood,
Some time ago you said you would be working on a short system. Any developments on this?
Yeah, Le Fly is busting my chops about getting it done. We are working on it slowly but surely. At this point, I can’t reveal my partner, but will say he is top notch!
Wood,
A couple of years back I did study on NYSE breadth. I found that using actual levels is more profitable than using Bollinger bands. Also, given that number of stocks on NYSE doesn’t change much and number of unchanged stocks is small, it is sufficient to use either $NYADV or $NYDEC. It makes charts less cluttered and calculations easier. More importantly, 3 day SMA gave better results that single day reading. A system that bought $SPX when 3 day SMA of $NYADV was below 1000 and sold when $NYADV moved above 1600 gave 74.5% winning rate with 28 bars in the winning trade and 12 bars in the losing trade. A system which was based on single day reading had the same 75% win rate but was inferior in profit factor and payoff ratio, thus bearing higher risks. This system went long $SPX when $NYADV was below 500 or short above 2700 and held $SPX for 2 days. Both the 3 day SMA and single day systems were more profitable (although not as frequent) than their Bollinger band based counterparts.
Ruschem, thanks for your comment. I will see if I can’t run those parameters and test it out.
My thought in using my own universe for the raw advance decline readings was that possibly there are already so many eyes on the NYSE that some of the edge may be eroded.
I started working with your last suggestion regarding the RSI and the lookbacks, and didn’t have much success with it. It does need some more work to confirm whether it would improve results.
Thanks again for your very helpful ideas!
I completely agree that using “the universe” is more representative and more accurate. One big problem with the NYSE breadth statistics is that each provider gives different readings and the discrepancy between providers could be huge. I am 100% for calculating my own indicator. As to what to include, after a few back and forth goings I settled on $NYADV that I calculate myself. Using Russell 3000, NYSE, and SPX components for the breadth, I didn’t find a significant difference in timing of the signals. Entry and exit strategy has a much bigger impact on results than the set of stocks used for breadth. I didn’t use a bigger universe mostly because it consumes too much computer time, which I don’t like. Another disadvantage of using price and volume in the universe is that every day you may have a different set of stocks in the universe and thus you will not compare apples to apples in your breadth. Indexes and especially the entire exchange are more consistent in this respect. One way to eliminate this potential source of inconstancy is to use % advancers or decliners instead of the actual number.
Sorry, I forgot to mention in the previous comment that the 3 day SMA system can be further refined using 30 min chart instead of the daily chart, in which case a 39 bar SMA of $NYADV should be used.
Ruschem,
What kind of stop loss did you use with your system for daily bars. 28 day is pretty long period,also any other filter condition.
Bill
Bill,
I don’t use any filters with this system. I tested 3-7% stop loss and 3-7% trailing stop. In addition, I tested different time based exits (2 to 60 day holding period) and profit target exit. Based on backtesting, I didn’t find any advantage in using any of these preset exits. Opposite, with preset stops my results were worse overall but profit target at 5-7% worked rather well for $SPX. Anyway, I use 3 day SMA of $NYADV as an alert rather than the signal. I will enter trade only if the market action warrants and place mental stop at slightly below the nearby strong support.
Woody,
How do you get NYSE advancer / decliner. Do we need to add all the ticker to a LIST and then use add composite from SYMBOLS menu.
thanks
Bill, check your mail. I sent you the code.
is that tradestation for the charting? if so, i would like to know how to code “unvierse”. i thought there was no way to roll your own breadth indicator in TS since you can only access data from symbols that are added or through the annoying ADE system(which i think would be too demanding for a large # of stock breadth…) help appreciated! thank you!
jojo- I use AmiBroker with a PremiumData.net data feed.
I used Tradestation’s brokerage, but once you get used to AmiBroker, there is no reason to use Tradestation for backtesting. AmiBroker is vastly superior, imo.
Woody,
thanks for sending the code.
do you add all the tickers for NYSE in the list before you run your scan. IF yes how do you add all the tickers.
thanks
Just load up the .afl, and then use “Apply To” —> Choose Filter.
From there, you choose “Scan” with an “N Last Days” = 1.
Basically, you can decide using the filter options how many stocks to scan. Or, depending on your data provider, (some do it differently) you can just scan the NYSE stocks.
I have mine running over all Major Exchange Listed stocks, excluding all OTCBB.
Hope this helps.