its hard to trade correlation without access to structureds (correlation swap, var swaps). you need to trade correlation thru volatility.
We used t/A in my office for years... We compared stocks to the Futures and bought the pull back after the mean reversion in anticipation of the stock catching up to the futures again. Or we sold the weaker ones when we saw the downtrend resume. It made a lot of money for about 6-10 people in my office for about 5 years. Our charts were essentially our way of using statistics. When you buy leaders or sell laggards charts can give you excellent insight. There are multiple ways to "baseline" your comparisons. Our setups worked great for years. The edge started to fade when the minis and the ecns allowed the NYSE stocks to be well arbed. Much of t/a is not mumbo jumbo when you set up you chart with an understanding of what the indicators do. Many anti t/a guys would be amazed what they would find when they looked at good t/a. A graphical display may give some insight until how the market really gets moved.
Thanks for your contribution to the thread. Let's promote concreteness, instead of talking in the air... What trades they did exactly? It's important to give the precise context and then make observation about whether a trade works or not.
Okay, for a two-asset case, what's the purest correlation trade to express a view on realized correlation?
Okay, for this one, how do you trade it? Thanks a lot! http://2.bp.blogspot.com/__L5g5MRRpLc/S7IyfDUNc7I/AAAAAAAAAA4/1nV2h2HAYj0/s1600/mime-attachment.jpeg
Interesting no one mentioned modern portfolio theory which is based on asset class allocation and diversification. The most important piece of the theory however that no one talks about very much is correlation amongst asset classes. Reason you buy bonds and stock to round out a portfolio is that they are negatively correlated. For example: SPY and TLT. The long term correlation is negative. There for trade the pairs trend allocating one security more than the other based on the trend of the pair ratio. If the ratio is spy/tlt and the pair ratio is trending higher allocate more to spy than tlt. If the trend is lower than allocate more to TLT. For positive correlated pairs you have either pairs that trend or pairs that do not. if they are correlated than go long one and short the other. if the pair is trending use one as a hedge to reduce volitility. If the pair is not trending and trades within a range than you can use bollinger bands to signal short and long signals for the pair. I prefer trending pairs as I can swing trade them as opposed to try and time intraday moves etc. Also trending pairs tend allows you capture bigger profits and trade less. non trending pairs tend to require a gazillion trades to capture 1-2% moves. I've never been able to generate long term profits looking for tiny profits per trade. The only winner with this strategy tends to be the broker who receives the commissions.
the guy that blew up was trading correlations futures of two european indices (Belgium and France) and the S&P500. Other ways I have seen people lose a lot of money is trying to trade eur/usd with usd/jpy, european currencies before the euro. And oil with gold. What I did myself and what works is trying to trade vol in index options hedging with other series of 'nearly similar' options (other strike-same month, other index-same country-same month) but these things are fundamentally correlated imo. (that used to be a nice game off-floor, now you would need to be a member to do that profitably imo to get the edge of sitting on the bid or ask.