Critique of Bitcoin Pricing Model by Raoul Pal

Raoul Pal, CEO of Co-Founders-Co-Founder-Real Vision at hedge fund 21st Capital, has drawn attention to a popular Bitcoin pricing model, labeling it as a “textbook case” of issues like data illiteracy and overfitting.

This model presents a close relationship between Bitcoin and Global M2, which measures the global money supply. Typically, M2 data is advanced to forecast Bitcoin’s price movements over a period of 10 to 12 days. Pal argues that macro liquidity conditions can impact the cryptocurrency cycle, potentially providing insights into current market trends through financial expansion.

Expert Criticism of M2 and Bitcoin Correlation

However, Sina, a data scientist with experience in teaching data analysis at various levels, argues that the model does not hold up under examination. In a video posted on June 24, he expressed, “This is a terrible mistake of not understanding overfitting.”

Sina highlights that while a correlation between Bitcoin and Global M2 appears to exist, the data is manipulated to align with past patterns. “If you’re playing with data, you can find great matches between various data sets,” he noted.

A significant issue he pointed out is that the Global M2 data is flawed. This discrepancy arises from mixing data from different central banks, often influenced by exchange rates. The combination of fast-reporting countries like the United States with others that have delays can create a misleading picture of daily global liquidity changes. “It may seem to work on a daily basis, but it blends various reporting speeds,” Sina stated. “That’s not a true signal.”

Moreover, Sina contends that the model falters when viewed beyond a selectively narrow chart. While Pal showcases closely aligned top and bottom figures for Bitcoin and Global M2, Sina demonstrated how small adjustments in timing or scale yield dramatically different outcomes. “Let’s try the 80-day lead. It doesn’t look good. 108? Now we have peaks, so let’s zoom in again and pretend it works,” he quipped. “This isn’t modeling; it’s just tinkering.”

He illustrated how changes to the model reveal a lack of systematic rigor, shifting from a 12-week lead to 10 weeks, or even 108 days. “Without a solid model, predicting the future becomes impossible,” he remarked. “This is a classic case of overfitting—it forces data to conform to historical trends, sacrificing broader applicability.”

To clarify his point, Sina compared the model to trying to fit a curve to an erratic sine wave. A good model captures essential patterns while ignoring extraneous noise. Conversely, an overfit model attempts to account for every small variation, resulting in decreased predictive accuracy with new data. “The overfitting might look pleasing, but it merely models noise, which isn’t reliable,” he noted.

Sina also questioned whether Bitcoin serves as a leading indicator of liquidity. “When I reviewed the last cycle, Bitcoin peaked first, and liquidity followed about 145 days later,” he mentioned. This observation challenges the causal relationships suggested by the Global M2 model and raises concerns about the model’s overall utility.

In closing, Sina emphasized caution regarding overfitting, stating, “They might seem to align, but are coerced to fit historical data. I’m uncertain about this model’s predictive validity.”

As of the latest update, Bitcoin was trading at $106,952.