Every product needs a good name. Branding, you know. Hence, the Market Dissector™.
I wrote a good bit about these breadth indicators a while ago, and then stopped. Nothing like some real-time performance to bolster the confidence derived from some promising backtests. I will be dusting them off over the next few weeks, making some refinements, and again making them a regular feature on the blog.
Below is one of the indicators- a very simple short-term breadth measure of advancers and decliners. It buys when the decliners surpass the upper Bollinger Band and sells when the decliners fall beneath the lower Bollinger Band.
With a 70% win rate, it works great for timing the SPY. Used to trade the QQQQ, win rate is 65% and is 67.5% for IWM.
Since I coded it and started tracking the signals in March 2o10, it has netted 9.25% (with commissions) and maintained a win rate of 68.75%, with buys and sells at the close. Results improve if we delay our buys until the next open. Not bad for in-sample testing.
Here is what it looks like in action:
Green arrows are the buys; red arrows are the sells.
One element of these indicators that I want to emphasize is that they are generated with de-listed stock data. An advance/decline index that does not incorporate de-listed data may suffer from survivorship bias.
Look for the buy alerts from this indicator in the future.
If you want to incorporate delisted data into your backtests, I have found Norgate PremiumData to have the best price and an outstanding service.
I like “Market Dissector;” sounds very cool indeed :-). And don’t all great trading ideas need to have cool and/or technical sounding names?
Now, I’m no expert in technical analysis, but a 70% win rate sounds pretty impressive to me. Certainly beats the CNBC talking head win rate (hovering around maybe 5% or so?) If you kept publishing these, I think I would find it useful to, say, time capital injections into PDS (at least I would know there’s a 70% chance I’m not buying on a trending down day). If you could come up with a more intermediate time frame indicator with a 70% win rate that would be even better.
We think a lot alike Data…
I like this too, although I not quite sure what I’m looking at. I’ll study it when I have more time.
Hawaii, don’t over think it. Very simple. You are buying when the number of decliners increases to surpass a couple of standard deviations beneath the mean. You are selling when the number of decliners falls and rises a couple of standard deviations above the mean.
Nicely summarised.
If you get a chance it would be interesting to see a backtested equity curve on a log scale, or other historical performance metrics. Based on your in sample test that would indicate ~20%. Is that consist with backtests?
Thanks
I will post that soon Mike.
Since 2001, for SPY, you are looking at between 8-11% annualized, depending on whether you are buying/selling the open/close.
One thing I’m looking into are stocks that are correlated with the SPY, but with a higher beta, and then using this indicator to time them. Food for thought…
You said…
in-sample testing? why not perform walk-forward tests with the lookback and standard deviations as variables? this being tested in amibroker?
i’ll have to whip this up and take a look at it… nice, i like
btw, i’m digging your blog… and the fact that you’re using Amibroker
Thanks stank!
I will try some walk forward with the parameters you mentioned.
I use the add-to-composite function rather than the built in A/D stuff.
awesome, thanks for the heads up… i put together a means of batch processing walk-forwards and created my own reporting mechanism as Amibroker currently doesnt offer it for walk-forwards. it basically writes the trades to a text file and then i crunch the rest in php. it still needs some polishing as i built it specifically for a thesis i am writing. let me know if you’re ever interested in looking at it.
Man-made ingredients Wig http://synthetic-wigs.dolabuy.com/ Internet our wide array of supplier Man-made Hair do Hairpieces, Endless Synthetic Wigs, Low-priced Man-made Hair pieces, Easy Phony Hair pieces, Inexpensive Man made Hairpiece and the most