Days When Interest Rates, Dollar-Yen, Stocks, and Crypto Diverge: Cross-Market Observation Notes with AI

Interest rates have risen, yet stocks are not falling. The dollar-yen is weakening, yet Nikkei futures are soft. Only Bitcoin is moving in the opposite direction.

It is easy to dismiss such days by saying the market is acting strangely. However, when you look at multiple markets, you realize that not everything reacts to the same factor in the same way.

The important thing is not to force the market into a single narrative. It is to align the movements of interest rates, currencies, stocks, futures, and crypto assets on the same timeline, and to separate what is fact from what is your own hypothesis.

I believe the best way to organize this is to use generative AI not as a machine that produces predictions, but as a tool for drafting observation notes.

First, align price movements from the same time period

When looking at cross-markets, the first thing to be careful about is the time.

The trading hours for stock spot markets, futures, currencies, and crypto assets differ. If you line up the Nikkei closing price with overnight futures prices and conclude that “stocks rose but futures fell,” you might mistake the difference in observation times for a market contradiction.

At a minimum, record the following information together:

• The observed time and time zone

• Whether it is spot or futures, and which contract month

• The start and end points of the comparison period

• Whether it is the price itself or the period return

• Whether it is a reference value outside trading hours or a value during trading

Comparing by the rate of change over the same time period makes it easier to compare markets with different price scales. When comparing the price difference between futures and spot, also note the contract month and the number of days until maturity.

Do not assume “rising interest rates mean falling stocks”

Interest rates are an important factor in viewing the market, but the direction of stock prices and currencies is not determined by interest rates alone.

For example, the stock market’s reaction can change depending on whether a rise in long-term interest rates reflects economic strength, inflation concerns, or government bond supply and demand. The dollar-yen also moves due to multiple factors, not just the Japan-US interest rate differential, including economic outlooks, policy statements, risk aversion, and position imbalances.

Therefore, rather than jumping to the conclusion that “interest rates rose, therefore stocks fell,” separate the hypotheses in between.

1. Which interest rate moved, from when to when, and by how much

2. How did the dollar-yen and stock indices react during the same time period

3. Did that change occur before or after the news announcement, or was it already priced in?

4. Which markets or assets cannot be explained by the reaction?

By writing down observed facts and interpretations of their reasons separately, you can verify later what you might have misjudged.

The correlation between crypto assets and stocks is not fixed.

Bitcoin is sometimes considered an asset separate from stocks. However, the relationship between crypto assets and stocks has changed over time. Past research has also confirmed periods where the linkage with the stock market strengthened.

This does not mean that they will always move together with stocks in the future. Correlation changes depending on the period and market environment, and high correlation alone does not mean that one moved the other.

Therefore, when looking at crypto assets, do not decide that they are a ‘substitute for stocks,’ but place them in a separate column. If they fall at the same time as stocks, first list the candidates: is it common risk aversion, or is it a factor specific to crypto assets? If you cannot judge, it is safer to leave it as ‘reason unconfirmed’.

Ask AI for organization, not prediction.

Even if you ask generative AI, ‘Will Bitcoin go up tomorrow?’ or ‘Is it time to buy Nikkei futures?’, you will not necessarily get reliable trading decisions. Instead, provide the data you have already collected and have it create a draft comparison table.

For example, ask it as follows:

‘The following are the changes in US long-term interest rates, USD/JPY, Nikkei 225, Nikkei 225 futures, and Bitcoin observed over the same period. Using only the input values, create a table separating facts, possible hypotheses, and unconfirmed points. Do not conclude causality from correlation alone. If data is missing, do not fill it in; mark it as unconfirmed.’

Do not adopt the explanation created by the AI as is; go back to the original data and verify it. Check for any mix-ups in time, units, comparison periods, prices, or rates of change. If the data before being passed to the AI is off, the conclusion will be off even if the table looks neat.

Observation note template

Keeping the following five items each time can also be used for reviewing your trades.

• Fact: Changes in price or interest rates actually observed during the same period

• Hypothesis: Factors considered to explain the price movement

• Falsification condition: Conditions for judging that the hypothesis is incorrect

• Unconfirmed: Insufficient data or factors not yet verified

• Action: Pass, reduce holdings, wait for conditions, etc.

For example, if interest rates rose and stocks rose at the same time, instead of just writing ‘stocks are strong despite rising interest rates,’ check which index rose, whether the rising stocks were widespread, and whether only futures were leading. Compare USD/JPY and Bitcoin in the same time frame, and it is important not to force the deletion of things that do not fit the explanation.

Record your judgments rather than explaining the market

Macroeconomic news provides clues for considering stocks, futures, currencies, and crypto assets. However, not all markets move for the same reasons, and there is no guarantee that past correlations will continue.

Using AI can reduce the time spent organizing information from multiple markets and writing down hypotheses and falsification conditions. Instead of leaving the trading conclusions to the AI, record the observations upon which you based your decisions. This accumulation of records will help you verify the market later and improve your trading rules.

※This article deals with methods for organizing market information and does not recommend investment in any specific financial product.