Ah, I was just now messing around with a volatility of volatility strategy in QM, basically holding SVXY when the volatility of SVXY is low or declining. Also, could you explain Trend Clarity sometime? https://quantmage.app/grimoire/cc6c07b2545dfe94500ae6025f1d9beb
Wow, you already started experimenting with the new indicator! :) The "What's New" section https://changes.quantmage.app/shedding-light-on-the-clarity-of-the-trend?id=1114 gives you a short introduction, but in the end the original paper will be the right reference. I feel that it probably works more reliably with a longer window than the 20d you used. I'll share more once I get any more insights on it.
Boy does this resonate with me. I did exactly the same kind of thing, fitting predictive relationships to a back sample. Various VIX durations. Very nice edge in sample.
But it turned that predictability was zero for even one day in the future, much less a week.
Yeah, in hindsight, I should have been more cautious with what I can assume for the time series these data providers provide. At first, it was like "let's revisit this assumption later", but then after seeing many backtests producing expected results (when compared against outcomes from other tools), I had totally forgot about it until this happened. ^^;
Ah, I was just now messing around with a volatility of volatility strategy in QM, basically holding SVXY when the volatility of SVXY is low or declining. Also, could you explain Trend Clarity sometime? https://quantmage.app/grimoire/cc6c07b2545dfe94500ae6025f1d9beb
Wow, you already started experimenting with the new indicator! :) The "What's New" section https://changes.quantmage.app/shedding-light-on-the-clarity-of-the-trend?id=1114 gives you a short introduction, but in the end the original paper will be the right reference. I feel that it probably works more reliably with a longer window than the 20d you used. I'll share more once I get any more insights on it.
Boy does this resonate with me. I did exactly the same kind of thing, fitting predictive relationships to a back sample. Various VIX durations. Very nice edge in sample.
But it turned that predictability was zero for even one day in the future, much less a week.
-Hydromod
Yeah, in hindsight, I should have been more cautious with what I can assume for the time series these data providers provide. At first, it was like "let's revisit this assumption later", but then after seeing many backtests producing expected results (when compared against outcomes from other tools), I had totally forgot about it until this happened. ^^;