Given that it's been over 35 years, I don't recall. But then, I don't dick-measure against other people based on it, either. I'd suggest either professional help or hiring a hooker to help you get over that.
Good idea... however a great many people have had that idea before you. You can certainly derive a predictive interval. Best to use predictive or forecast beta, rather than historical beta. Historical beta is a bad estimator of future beta. This has been known since at least Elton, Gruber and Urich 1978. Also Barr Rosenberg 1985. Almost any forecast method is better than just using historical beta. One hint, estimate forward val and forward corr separately, then recombine. Also, portfolio betas, even for a small portfolio, are much more stable and forecast-able than SN betas. Maybe move to basket pairs trading or within-portfolio stat-arb.
So a lot of hi-finance guys are wrong believing beta is how much an individual equity moves against a major index ... such as $SPX?
You're probably thinking of beta-weighted delta. Again, it doesn't stay locked against a portfolio - but it's a pretty decent rough measure of delta exposure.
Oh I've always known it doesn't stay locked but a beta of 1 doesn't become a beta of 5 overnight unless there has been a truly spectacular good or bad news.
Kevin, I need to forecast the beta of a stock over the next 30 days relative to the index. Is this the industry standard right now https://www.nber.org/papers/w26105 ? Say Im looking at NVDA and QQQ... I was thinking of decomposing the beta into correlation and vol, estimating separately and then rejoining. However, given NVDA is likely dependent on the market (QQQ), estimating NVDA vol separately before adding it back into the beta calculation does not make too much sense to me given NVDA is dependent on QQQ. I was simply going to try and forecast beta and idio vol separately to get NVDA vol. NVDA vol = sqrt(Beta^2 * QQQ vol^2 + idio vol^2) for a rel val trade.