SPY RSI2 > 90 for Four Days…Any Edge?
Monday, July 20, 2009 at 12:36 am
Friday marked the fourth consecutive day that the RSI(2) of SPY closed above 90. Is there any edge to shorting here?
The Method:
Sell short (SPY: 117.10 0.00%) at the close any time the RSI(2) of the index will stay above 90 for four consecutive closes.
Buy to cover on N day close.
100K per trade with commissions and fees not included. (Gains are not compounded.)
37 trades.
The %Profitable is not great, but perhaps the Win/Loss Ratio will be better?
Nope. Nothing to cheer about.
Profit Factor shows that this method would have barely made enough to cover commissions and slippage.
Summary:
1-6 days day out has a slight bearish edge, but it is very slight.
Instead, it may be better to look at how consecutive high RSI(2) levels might portend future strength.
Man that’s surprising. Here’s another take on short’s current “advantage”, of the double 7s variety: buy SSO if close is lowest of last 3 days, hold, sell when close is highest of last 3, hold, repeat (system is ‘always in’, although I think a time stop improves it, possibly multiple positions would as well, and maybe fourteen filters, an excessive drinking habit and an excess of “risk capital”). For 500 days it shows 125 wins, 39 losses, wins .9x losses.
100k per trade: http://screencast.com/t/ZFVPOuw05h
http://quantifiableedges.blogspot.com/
Study suggesting short-term pullback is probable.
Nice work. Strongly trending markets tend to peg oscillators. I don’t think I ever thanked you for your prior work on RSI-2. I’m still using RSI-14 to find LT divergence setups, but RSI-2 is now in the toolkit for trade maintenance.
BTW, are you trading any trend following systems (like the golden cross) or sticking to the mean reversion systems?
Yogi, I’m not trading any trend following systems, regrettably, only MR. I do have some plans in the works though.
I can’t quantify it yet, but I have a really strong gut feeling that we are going to witness the slow transition from MR back to more trending markets, over the next few years.
I think that the dislocations of the past year or so will likely change things going forward, as investor psychology has changed, much as things changed after the 1987 crash.
Like I said, no proof of this regime change, but I want to go ahead and get some systems in place should it start to happen.
I think that’s about right. I’ve found moving from one market “type” to another to be one of the more difficult tasks to master. I deal with it by slowly moving capital from one type of style to another (take profits/losses reallocate to another idea in another style). The only thing more difficult IMO is adapting trading to life changes e.g. get married, start family, career changes,etc. These have had the most profound impact on me as a discretionary trader. I would assume they would affect a systems trader also. Realizing they will have an impact is 90% of dealing with them so adjustments can be made.
Somewhat off topic I lost the link to your CCI system. Could you post it in a reply please?
Thanks
http://ibankcoin.com/woodshedderblog/2009/02/18/trade-with-the-trend-while-trading-against-it-part-two/
John, disregard the weird characters. They are remnants of our move to the new theme and will be fixed in time.
Thanks, you have a lot of interesting stuff - I’m slowly getting through it though. I agree about MR maybe fading away and I am trying to get a strategy together to cope with the morph back to follow through too.
hi, you might want to segregate the perfomance stats above certain moving averages—50,100, 200,300 or momentums etc one of the key spoilers of the RSI2 is to short when the market is in an uptrend…….in contrast going long is more profitable regardless. i don’t have these stats or the ability to test this…..just a suggestion.
dv
David, good to see you here.
RSI2 definitely works better for finding short term oversold conditions than overbought. When using RSI2 to short, I use an abnormal filter based on volatility to get me out of the short trade before things get too nasty.
I’ve run those tests before, with RSI2 above a certain moving average, and while they are helpful, they limit the sample size even more than it is already limited.