Yes, they did. And not with backtesting, but projecting a decade or more out. Note that the first fully predicted boom-crash was the late 1990s one, because it takes some significant computing horsepower to build really good models, crunch that wasn't really available and cost-effective prior to the 1980s.
I saw models in the late 1980s that predicted not only the timing and magnitude of the 1990s stock boom, but the crash as well. Actually, they put the crash in the late 2001 timeframe IIRC, but I give'em an "A" for effort; in the intervening years they may have done a model update that revised the actual downside date. The further out the projection, the bigger the error margin on magnitude and timing. The models are occasionally checkpointed with reality to reduce the error margin; models that have drifted more than a certain amount at the checkpoint need to be rebuilt, which is expensive. Among the many reasons that cash flow into the stock market from institutions liberalized a few months ago is that several big models checkpointed okay, which basically confirmed that we were past the bottom as the models predicted and that the models (which almost universally stated we should be on a long upside) were still reasonably valid predictors.
I was skeptical when I saw some of the reports generated by these models in the mid- to late-1980s, but over the years I've become a believer because those models delivered despite my skepticism. I've since involved myself in the actual science and theory of large-scale predictive models of complex dynamic systems. Most big financial institutions and hedge funds have close guarded predictive models of the markets and economy. Some are better than others, but I can state for a fact that most are vastly better than most people generally believe. Also, the factors that go into even simple models are unbelievably diverse. All the most important factors are largely unrelated to finance and the stock market, yet they are remarkably good at predicting finance and markets.
I design predictor technologies for hedge funds among other clients. Good predictors are heavy duty trade secrets and are worth quite a bit of bank to those that can produce something just a few percent better than the institution down the street.
So I suspect that you'd say that Sornette's analysis is a one trick pony, suitable mostly for "predicting" the couple years of charts prior to when it was published. Which is, as someone noted above, useless BS.
hmmm ...
I do not have access to the models of which you speak. But if they are as you say, then this is most impressive.
We shall soon enough see. I've been quite happy with the very dark, pro-Gold, anti-stock, advisors I've been following for the last three years, such as James Dines, Elliott Wave and bits and pieces such as Sornette's analysis above.
But I pulled out of Gold earlier this week, and currently am only invested in my health, my profession and my children. I think I will stay on the sidelines for a bit until I see which way this is playing out.