Quantitative analysts claim that the global M2 cannot predict Bitcoin prices.

According to Gate News bot, NewsBTC reported that Sina, co-founder of the hedge fund 21st Capital, publicly refuted the popular Bitcoin price model promoted by Real Vision CEO Raoul Pal, calling it a typical case of data illiteracy and overfitting.

The model predicts the future price trends of Bitcoin by deriving the close correlation between Bitcoin and global M2 (an indicator of global money supply) by advancing M2 data by a certain number of weeks (usually 10 to 12 weeks). Raoul Pal once used this chart to argue that macro liquidity conditions drive the cycles of cryptocurrencies, and current market behavior can be predicted through monetary expansion.

But Xina stated during the undergraduate and graduate courses on data analysis that this model does not hold up to scrutiny. "This is a serious failure to understand overfitting," he said in a video released on the X platform on June 24. "What I've seen wouldn't even pass the first month of a first-year data analysis course."

Sina pointed out that the apparent correlation between Bitcoin and global M2 is merely because the data has been "distorted" to fit historical patterns. "If I could manipulate the data at will, I could definitely find good matches between various data blocks," he warned, and it is precisely this flexibility that allows analysts to create the illusion of accurate predictions.

He explained that the main issue lies in the flaws of the global M2 data itself. This data is derived by multiplying the M2 data of various central banks by exchange rates—mixing fast-reporting economies like the United States with countries that have data delays of weeks or even months. This creates a misleading impression of daily fluctuations in global liquidity. "It looks like it's changing every day, but in reality, it mixes frequent and infrequent updates," said Sina. "This is not a real signal."

More importantly, Sina believes that once the perspective of a specific chart is narrowed, the model fails. While Raul Pal and others have shown examples of Bitcoin closely aligning with the global M2 at tops and bottoms, Sina has demonstrated how subtle adjustments in lead time or scale can produce entirely different results. "Let's try a lead time of 80 days. It doesn't look good. How about 108 days? Ah, now the top is aligned—so let's zoom in again and pretend it's valid," he said sarcastically. "This isn't modeling. This is playing."

He emphasized that each adjustment of the model—from a 12-week forecasting lead time to 10 weeks, and then to 108 days—revealed its lack of systematic foundation. "Without a proper model, you cannot predict the future," Sina said. "This is a typical case of overfitting. You force the data to match historical behavior, but lose general applicability."

To explain this concept, Sina compared it to fitting a curve through a noisy sine wave. A well-structured model can capture the core patterns and ignore the noise. In contrast, an overfitted model tries to match every tiny fluctuation, resulting in poor predictive performance when new data arrives. "Overfitting looks better, but it simulates noise. And noise does not repeat," he said.

Sina also questioned whether Bitcoin would really lead liquidity rather than follow it. "If you look back at the last cycle, Bitcoin peaked first. Liquidity peaked 145 days later," he said. This overturns the causal relationship implied by the global M2 model and calls into question its entire premise as a predictive tool.

His conclusion was very straightforward: "You have to be very careful about overfitting. It looks like a match, but it is forcing a fit to historical data. You have no idea about the predictive accuracy of this thing."