As I have been using RSI(2) extensively to guide my discretionary trading, it seems only logical that I back test it. For this I am using Tradestation. Being a new user of the platform, it is taking some time to learn how to use it properly. Results will be forthcoming.
In the mean time, two other traders, Dogwood and Crazy Eddie have run tests of the RSI(2) strategy, and their results are significant. However, I encourage you all NOT to visit their blogs and want to emphasize that you should NOT trade this strategy as I would really like to get my automated trading set up before you, the internet stranger, ruins the edge for me.
Dogwood’s back testing is here.
Crazy Eddie’s back testing is here.
Both traders have run subsequent tests, so be sure to check out their home pages for the latest updates.
To read the original post on the RSI(2) strategy, which sort of kicks off the methodology, read this: Improve Your Timing With RSI(2).
Late Addition: Bill Rempel has contributed in the comments section some research he has completed on RSI(2). It is very interesting…. Read it here: About that 2-Day RSI.
Shed, you will become my backtest minion for the ibc machine, yes? I have 20 odd years of data now.
Those results are amazing. Also striking is that the strategy w/ rsi(2) is extremely similar to my going 25%, 50%, or 100% in on leveraged etfs when IBC machine says buy.
Boone, I would be honored. I believe I have access to 40+ years of Dow data and 20+ of S&P.
You remember when we first spoke about the machine that I mentioned we might want to try adding an RSI(2) component.
Here’s my post from July 2007 on the subject of 2-day RSI and the S&P 500 index, dating from the 1950s to the present (2007 at the time of test).
Enjoy!
http://www.billakanodoodahs.com/2007/07/about-that-2-day-rsi/
Thanks Bill. I added your work to the original post. Very interesting…Have you drawn any conclusions from from the implications?
Yeah, I was kinda miffed that y’all were “measuring dicks” on how much data y’all had on RSI(2), when I went back 57 years almost nine months ago. LOL!
I think the indicator’s been stable enough for long enough that I would feel comfortable allocating SOME money to a short-term system based on it, if I were trading that style.
I would also feel comfortable using it as a cue for discretionary entries in another system based on longer-term technicals or fundamentals.
What I took from the multi-decade analysis is that the markets change, and some indicators change as well, so if I built something around RSI(2), I would probably take a look under the hood every year or so to make sure it was still working the way I expected.
would it not be somewhat dangerous to rely on one indicator RSI(2) to place trades since it seems one indicator may only work under certain conditions (for example: works better in bull markets than in bears)..I understand that even if you use several indicators (Macd, Rsi, Sto, etc,) you’re still taking chances but the odds might be more in your favor to place a winning trade. In any event, I look forward to your further comments. And Danny’s ibc machine (using many indicators) sure looks interesting. Is this the holy grail perhaps? Dank you for sharing.
Bill, lol at the measuring dicks…although my comment re: data, was pertaining to backtesting a strategy that Danny has been developing. Although you should know better than anyone since I subscribe to your posts that I read most everything you write. However,If I read that post, I had forgotten it.
Anyway, good point about checking under the hood, although it seems that it would be pretty obvious when it stops working, as your edge would disappear.
Buylo, I would be willing to commit a large portion of my capital to trade this strategy as it is really very simple. Markets will always move in over-reactions. These over-reactions will produce over-bought and over-sold conditions. As this strategy has you buying weakness, and possibly, selling strength, it seems that these conditions will not disappear, and therefore the strategy should continue working. However, Bill’s research shows that things do indeed change over time.
I just backtested my garden hose.
Fly, that would make sense if you said you “backflow tested” it. City slickers…lol
Shedder, thanks for the links over the last couple days. Much appreciated.
Buylo, one problem with RSI(2), if you are using it to trade a market index, is that trending markets will produce very few trades, i.e. August to November 2007.
I wouldn’t trade RSI(2) exclusively, but I think it would work well paired with a trend following system. The former works in chop and bear markets when long-only trend following systems fall apart, and the trend system works in trending markets when the RSI(2) system produces few or any trades for months.
The advantage, of course, is that when trend signals begin disappearing (November 2007 to March 2008), RSI(2) signals start cropping up on a regular basis, and vice versa.
I think combining the two has great potential.
Good comments Dog.
I wonder if Bullish Jim doesn’t have an opinion here. Maybe I’ll ping him…
Good luck with the system dev. Can’t wait to see your results. I’m typically not a fan of indicators but I can’t dismiss quantitative results. I think I am going to take a look at layering this over the trend-following stuff I’ve been working on over at ibdindex as well.
Woodshedder,
I have to say the preliminary results of my own tests on this are pretty astonishing. I have a machine running through about a million trades in my three test portfolios and will probably get something up over the weekend but the expectancy curves I typically look at are about as perfect as you could hope for to reveal a strong correlation between the indicator and expectancy. Thanks for all the links.
Best,
Bhh
Bhh- thanks, I’m looking forward to reading it.
Unfortunately work has blocked .blogspot addresses so I am unable to read you during the day hours….I’ll check back at your site this evening.