New Trading Indicator: The McClellan RSI(2)
We’re poking around the edges of the McClellan Oscillator this week – read the previous parts of the series here, here, here, and here.
I’ve been toying with a new approach to analyzing advancing/declining numbers that I haven’t seen elsewhere – I’ll call it the McClellan RSI(2).
Basically, I’ve married the advancing/declining numbers used in the McClellan Oscillator with the concept behind blogosphere darling RSI(2). Click for an excel workbook showing the calculation.
The graph above shows the two together YTD (through 04/17) – RSI(2) in blue and the McClellan RSI(2) in red.
As I ranted in The Problem with the McClellan Oscillator, et al., despite the fact that the McClellan RSI(2) is using advancing/declining numbers instead of price, the two are still going to still track each other pretty closely.
In my own tests, I’ve found this new version to be not quite as predictive of market moves as the original RSI(2), but I’m sharing it here because I thought it was a pretty nifty idea and I’m hoping that it turns on a lightbulb in one of our many quant-oriented readers (and if it does, don’t forget who loves you =)
[Edit: click for a summary of all posts in this series on the McClellan Oscillator]
Happy Trading,
ms
To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed.
Filed under: Trading Strategies | 8 Comments
One issue I see with this – the traditional advancing/declining numbers on the NYSE (I’m assuming this is NYSE) include a lot of “non-stocks” – so you may get some junk in the numbers caused by that.
RE to D – If you had asked me that question last week I would have agreed (seems perfectly logical). But after this weeks series, and specifically this post: http://marketsci.wordpress.com/2009/04/26/the-problem-with-the-mcclellan-oscillator-et-al/, I think those non-stocks don’t cause much drift.
I don’t believe there is anything wrong nor undesirable concerning a non-price technical indicator having a correlation with price. You need this correlation because, ultimately, your trading instruments are based upon and measure price (well, you could make a case for the VIX to be somewhat of an exception…). However, you don’t want your indicator to just be one of hundreds of variations on price movement or advances/declines or volume changes). I think the elusive goal is to find a non-price technical indicator that has both a strong correlation with price and consistently differs from price in phase (not just divergence, which also includes noise) in order to provide a predictive action upon price. I’ve been looking at something similar to what you’ve described — applying RSI(2) to the $TRIN. I have come to a similar conclusion — it’s interesting but not reliable enough to use in trading.
Neat idea Michael.
I think so too. Now if someone can just figure out how to do something useful with it =)
Michael,
I have my doubts that something like a derivative (RSI2) of a derivative (MO) of market breadth -which utilizes a 39-day and a 19-day EMA in the current short-term mean-reversion determined environment- will work as leading short-term indicator for price.
If using an indicator for a mechanicl trading system (but at the end everything is price, volume and breadth over time, and every indicator nothing else but a way to simplify and quantify by a single number the effect of gread and fear on prices), I’d currently prefer Wilder’s +DI/-DI (e.g. 2-day) to spot the strength behind short-term trends in price and volume.
And Wilder’s +DI/-DI may be a helpfull as well to make a mechanical trading system adaptive by eveluating if there is any (notable) trend behind price (and it’s magnitude) and then adjusting entry and exit points accordingly (e.g. concerning the RSI2 go out/sell short at 70 assumed there is no or only a weak trend, but go out/sell short at 80+ if there is a strong trend in price and volume in order to ride the trend to its potentially ideal reversal point).
May be another idea for your ‘TO DO’ (homework) list :-)
Best,
Frank
(http://tradingtheodds.wordpress.com)
Hello Frank – take a look at the excel calc – the RSI(2) is not a derivative of the MO (i.e. has nothing to do with 19/39-day EMAs). It’s simply using advance/decline numbers in place of price to calculate the RSI(2). michael
I think it is interesting that the two areas where the corrilation breaks down (01/06/09 and 03/18/09) were turns in the market.