Correlation trades?

Discussion in 'Trading' started by HezBallah, Jan 4, 2012.

  1. I heard that one trade that can be made during periods of volatility is betting on the correlation of trades to lessen. Apparently in times of high volatility, the correlation of S&P stocks goes to like very close to 1. Obviously, one can bet that the correlation will decline from 1. How does one make that bet?

    On bloomberg, how do you look up the correlation of all the stocks of the S&P?
  2. I realize my question is asking for answers without providing much. The least I can do is explain a bit about why I'm asking. I'm currently a trader at a small prop firm who specializes in domestic equity. I do fairly well but I worry that my niche is small and becoming quickly saturated. I want to expand my knowledge of trading into other areas and I've been trying my best to observe everything in the domestic markets... but I haven't been able to learn much about international arb or traders in other markets involving options. I simply don't have access to them at my firm. So... I came here.

    Also, I hope to provide some interesting discussion. I hope this can go somewhere.

  3. Here's one, technical, way to do it. I'll use the Dow 30 as my example. It involves holding two assets, one the index and one a basket portfolio of all the stocks that make up that index, with the correct weightings (an exact mimic). Variances swaps are the assets used to trade this.

    If all 30 stocks in the Dow move by the exactly same amount on any given day over a period of time (e.g. all +1% one day, all -1% another day), their variance will be 'x'. If this happens then the variance of the stocks in the basket will be identical to that of the index.

    If the correlation between the stocks starts to weaken (move away from 1), then the variance of the index may not change much (rather than all up 1%, half could be up 2% and half are flat). The individual stocks are behaving differently, but the net effect (as this is an index), and therefore variance of the index, is the same.

    However, in this scenario the variance of each of the individual stocks will increase as they start to move by different increments or in alternate directions. You want to be long variance swaps in the individual stocks, and short variance swaps in the index.

    I haven't looked at that one in a good year, hopefully that's correct.
  4. All right I'm going to have to figure out what this means and I'll get back to you.
  5. Actually there is one far easier way of placing the trade you suggest. Put it on as a pair trade with a defensive and aggressive stock.

    Higher correlation between stocks causes stock betas to drift towards 1 (index beta), as they all start performing the same, which therefore means the same as the index. If a defensive stock with a historical beta of say 0.6 has a 6 month beta of 0.9, consider shorting this stock. Find an aggressive stock which has a historical beta of say 2, which has a 6 month beta of 1.2, and go long this stock. The closer together these two betas are the better, and the further apart their historical (long term) betas are the more profitable it will be. The trade *usually* works because:

    The assumption, and usual trend, is that the stock betas will not cross. I.e. the defensive beta will never be higher than the aggressive beta, despite their rapid convergence. If they cross, that means correlations are decreasing again, and since this is usually associated with exiting a crisis, it is highly unlikely that the defensive will outperform the aggressive stock in a rally. When markets rally, correlations fall, betas revert back to their mean and the pairs trade makes money. If markets continue to fall, and correlations increase, their betas will converge more and the pairs trade will lose money. This is limited by the above reasoning, but that's the bet that you take, that correlations won't increase further.

    Hope this helps.