In cryptocurrency market analysis, you often see this phenomenon: analysts form a conclusion first, then adjust data and models in reverse to prove it. To put it bluntly, this is to make the data obedient. Truly professional data analysts should let the data speak, rather than forcing the data to perform. The reality is that, many times, models are either not flexible enough or analysts' subjective intentions are too strong. Such routines are common in market forecasting, trading signal generation, and on-chain data interpretation. If you haven't experienced the process of "repeated validation" of data (which is actually repeated adjustment), then you probably haven't truly been in this circle.