This is a dispersion question. I will also use a current real life example to make it more appropriate. SPX Jan 2019 100% moneyness vol = ~ 12.5% CBOE implied correlation Index (Jan 19 expiry) 28.92 (I want to mention that this is only the implied correlation for largest 50 stocks in SPX). Formula for calculating index implied variance. Currently the market is implying only a .28 correlation going forward which is relatively low. This being said a good dispersion trade would be put on if implied correlation is greater than my forecasted correlation. A reverse dispersion trade if my forecast correlation is likely to be higher. As a hypothetical trade, AAPL and MSFT make up ~ 8% of SPX. Their 60 day correlation is .75. If I believed this would continue, would a reverse dispersion trade be a reasonable trade? Here is the theoretical portfolio implied variance(assuming equal weight in AAPL and MSFT). Annual vol is converted into variance/standard deviations. AAPL Jan 19 100% moneyness = ~22% MSFT Jan 19 100% moneyness = ~23% AAPL variance = 9.87% MSFT variance = 10.34% var = sum(.5^2*.0987, .5^2*.1034)+ 2*sum(.5*.5*sqrt(MSFT_var)*sqrt(AAPL_var)*.75) var = .0625. Converting this to annual vol sqrt(.0625)*sqrt(179/365) = .175 This gives us a portfolio implied vol of 17.5%. In fact, even when we drop the correlation to .10 the portfolio vol is still 16%..... Let me know what you guys think and if I have made any mistakes with the math. Talk soon. Happy trading.
I think you can't replicate the index with just 2 stocks. Run a PCA and find out what are the main factors driving returns then you can think about how many stocks u need.
You can trade APPL and MSFT options against SPX and I guess you can call it a dispersion trade. But it will be a very high variance trade even if your view is correct -- the two halves of your trade don't match very well, dispersion/correlation will not be isolated. APPL and MSFT only make up about 7% of SPX (7.4%). You might try adding AMZN and GOOG and trading it against the NDX, where the first four fims make up more than 40%: > head(getIdxSymsWeights('ndx',byFirm=TRUE)[,1:3]) FullName Symbol Weight 1 Apple Inc AAPL 0.11150777 2 Amazon Inc AMZN 0.10474791 3 Microsoft Corp MSFT 0.09827803 4 Alphabet Inc GOOG:GOOGL 0.09578808 5 Facebook Inc FB 0.04982900 6 Intel Corp INTC 0.02638947 Stepping back a little for a bigger picture view; your math (at first glance anyway) looks correct, but dispersion is not the same trade it was in 1995. Its not even the same trade it was in 2005. It's gotten much more complex. While it is not as crowded as it was, that is because the edge it no longer there. The successful private traders I know have one thing in common -- they know one market or even one trade really, really, really well. You were doing pretty well in assault on earnings trades, I would concentrate on that until you have it down to a science and put dispersion on the back burner for now. I also think there is more edge in earnings trades for individual traders because earnings response is so dependent on hard-to-analyze idiosyncratic factors.
LOL Is getIdxSymsWeights a personal function? It's not in quantmod or PerformanceAnalytics. Thanks for the advice Kev. Dispersion trading is very interesting, I have read 2 decent papers on it published in early 2000's. But your saying that's somewhat mute now. Any good papers you have on it from more recent dates? Lastly any specific stat books/online courses you have taken that got you really thinking and could somewhat be specific to analyzing idiosyncratic risks?
Your pairwise correlation term is going to dance all over the place. When Gold moves by, say, 3 sigma, your pairwise correlation value will go up. So too if, say, corn move 3 sigma? What you;re not trading that stuff? Doesn't matter, when one market moves, however disparate, they all tend to move - and, in stocks, together. This is a major calculation problem. QUOTE="TheBigShort, post: 4699422, member: 504861"]This is a dispersion question. I will also use a current real life example to make it more appropriate. SPX Jan 2019 100% moneyness vol = ~ 12.5% CBOE implied correlation Index (Jan 19 expiry) 28.92 (I want to mention that this is only the implied correlation for largest 50 stocks in SPX). Formula for calculating index implied variance. View attachment 189259 Currently the market is implying only a .28 correlation going forward which is relatively low. This being said a good dispersion trade would be put on if implied correlation is greater than my forecasted correlation. A reverse dispersion trade if my forecast correlation is likely to be higher. As a hypothetical trade, AAPL and MSFT make up ~ 8% of SPX. Their 60 day correlation is .75. If I believed this would continue, would a reverse dispersion trade be a reasonable trade? Here is the theoretical portfolio implied variance(assuming equal weight in AAPL and MSFT). Annual vol is converted into variance/standard deviations. AAPL Jan 19 100% moneyness = ~22% MSFT Jan 19 100% moneyness = ~23% AAPL variance = 9.87% MSFT variance = 10.34% var = sum(.5^2*.0987, .5^2*.1034)+ 2*sum(.5*.5*sqrt(MSFT_var)*sqrt(AAPL_var)*.75) var = .0625. Converting this to annual vol sqrt(.0625)*sqrt(179/365) = .175 This gives us a portfolio implied vol of 17.5%. In fact, even when we drop the correlation to .10 the portfolio vol is still 16%..... Let me know what you guys think and if I have made any mistakes with the math. Talk soon. Happy trading.[/QUOTE]
I'll send you a couple dealer research papers. Dispersion is a tricky business, since besides the actual correlation (which is different at different maturities, as correlation risk premium grows with time), there is a matter of skew and weighting. Also, you can try playing games like partial dispersion (which is more like relative value in gamma) and/or exotic dispersion via structured products. YYMV
Realistically, you have to be a big player to trade dispersion anyway. Nobody is going to quote you a top-50 package unless you’re a client (cause it’s a pain) and assembling that position on your own is very annoying. Then there is the return on capital - dispersion is a 2nd order effect so you need access to institutional level of leverage (though may portfolio margin can do it for you somewhat)