Let's clarify these misconceptions because they seem to repeat themselves over and over. 1) The Power Law Itself Doesn't Indicate Bubbles It's essential to emphasize this point repeatedly: According to the power law model, there are no inherent bubbles. Any apparent deviations (bubbles and corrections) effectively cancel each other out over the long term, leaving the underlying trend intact. Bubbles Exist at a Deeper Level and Can Be Analyzed While the power law smooths out extremes, bubbles do manifest in the data. By subtracting the overall power law trend from the price series, you can isolate and study these deviations—their structure, patterns, and regularities. This data-driven approach is far superior to speculative guesses about future peaks. 2) Characterizing Bubbles: Exponential Decay Method Several techniques exist for modeling these bubbles. One reliable method observes that the decay from peak highs often follows an exponential pattern. If this decay persists, it suggests the next bubble could deviate by about 80% above the power law trend. For instance, if the power law projects a reference level of around 130K by year-end (a common point for cycle tops), this implies a potential peak near 200K. Alternatively, you can visualize this as a "decay channel" that bounds the tops over time. 3) Quantile Regression for Modeling Deviations Another approach uses quantile regression to model deviations from the power law. I've discussed the pros and cons of this method in detail in one of my articles (link in the comments). Its advantage is that it doesn't require an explicit decay assumption for the tops. However, a key drawback is that it fits power laws directly to the peaks, which can overestimate possible deviations. This is why models like @TheRealPlanC 's tend to predict higher tops compared to mine. 4) Hybrid Quantile Model with Decay To address limitations, you can enhance the quantile method by incorporating an explicit decay component, as outlined in my article. This hybrid combines the strengths of the decay channel (realistic bounding of peaks) and quantile regression (flexible handling of data distributions), yielding more balanced estimates. 5) Critique of the Bitbo Approach (Power Law Fit to Tops or the Power Law Corridor) The Bitbo method simply fits a power law directly through the historical tops, an idea first proposed by @hcburger1 to estimate potential ranges (he called this the Power Law Corridor). The bottom of the corridor by the way is valid and supported by solid stats. But not the top. While I've discussed this with him extensively, it's not the optimal approach—it's akin to quantile regression but inferior, as it assumes a simple power law governs the tops. In reality, only the lower 50th percentile of data adheres strictly to a power law; the tops exhibit more complex behavior. This leads to unreliable and overstated estimates. I hope this breakdown clarifies the distinctions between these methods. Overall, remember that predictions for bubble deviations are inherently less predictable than the core power law trend itself. The power law remains robust and strengthens with more data, serving as the foundational reference. Cycle-specific deviations, while useful, are secondary and not central to the theory—they rely on the power law for context and carry higher uncertainty.