Oh. Okay. Yeah, you can obviously replicate a variance swap (in theory) on your own with a bundle of options. But I wonder whether bid/offer spread + transaction cost might make it infeasible for small trade sizes. I don't know. Maybe.
I am not endorsing this method, but what some people do is use volatility cones: http://www.m-x.ca/f_publications_en/cone_vol_en.pdf In addition, my intuition says that five days is probably not enough. Volatility and therefore option prices are directly proportional to the square root of time. Five days seems too small when you take its square root to give you enough expentancy at 70% rate of accurary [although in the markets of the last three months, five days of vol seems like fifty days of vol not that long ago...] But if you are good at predicting future realized volatility at a 70% clip, why not trade the underlying and forget vol? That would make you very rich if you used say 1/2 kelly to put on positions so that you would guarantee not blowing out on a bad streak. nitro
Nitro, thanks. I use a similar method and model in trading FOREX and have had good results. I wanted to see if I can do the same in HV and apply it to options.
Good for you neural but eeeuh, how does this work? Say you can predict future realized vol 100% and just trade the underlying. How do you profit?
Index vol prediction IS predicting direction, but I can't imagine how a successful vol prediction would be terribly useful in FX or any other [than vol] market if trading spot. Var-swap replication is done daily by Citadel, Susq, DES, Jane St, etc. To the OP -- why not trade VIX futures and/or options?
I was thinking the same as atticus. If you can predict how far the market will travel, ratio writes may pan out better than strangles. Set the long strike inside the range and the short strikes outside the range. Make sure you collect a credit when putting the position on. That should give you a very wide profit range and more room to be wrong. Just a thot. Peace and gtty, Lar
Noone is suggesting the S&P 500 3-month Variance Futures traded on CFE ! I was trying to apply a similar result with neural networks but I have a question: if you are really able to forecast HV there is a good chanche to forecast IV and the index itself since the correlation is not so bad. So it seems to me too good to be true indeed...