To answer this question, researchers analyzed 17.1 billion order book entries, 14.3 million trades, 147,000 wallets, and more than $84 billion in trading volume. The results turned out to be much more interesting than expected.

Anyone who has even a passing interest in blockchain has surely heard stories about “smart wallets.” Some addresses regularly buy tokens just before a price surge, while others exit the market with surprising precision right before a drop. This has even led to the emergence of dozens of copy-trading services that allow users to simply replicate the trades of successful traders.

But are there really wallets that consistently make better decisions? Or are stories like these just lucky coincidences?

A group of researchers led by Daojing Zhai attempted to answer this question and published their work in the journal *Frontiers in Blockchain*.

One of the most comprehensive analyses of trader behavior

The authors used the decentralized exchange Hyperliquid as their data source, where each transaction is linked to a public wallet address. This made it possible to track the activity history of each market participant.

The scope of the study is impressive:

• 17.1 billion order book messages;
• 14.3 million aggressive trades;
• 147,113 unique wallets;
• trade volume exceeding $84.3 billion.

Instead of the usual analysis of prices and volumes, the researchers decided to focus on the market participants themselves.

What did they want to check?

Most forecasting models use price, volume, order book depth, and technical indicators. However, the authors hypothesized that a trader’s individual trading history could be another source of information.

Simply put, if the same trading account consistently makes better decisions over a long period of time, can this information be used to analyze the market more accurately?

To test this, the researchers created an awareness rating for each wallet. After each aggressive buy or sell, they analyzed how the price changed in the following seconds. If, following transactions by a specific address, the market tended to move in that direction, the rating for that wallet increased.

The authors then examined whether this advantage persisted over time.

The Most Unexpected Result

The researchers then developed two predictive models.

The first analyzed only traditional market data: the order book, trading volumes, and trade flow.

In the second update, they added another parameter—information about which specific wallets are currently conducting transactions.

After incorporating information about the traders' identities, the accuracy of short-term price movement forecasts improved by approximately 13% compared to a model that used only market data.

In other words, the information about who exactly executed the trade did indeed contain an additional signal that could not be derived solely from price, volume, or the order book.

Interestingly, this finding complements the study on MEV bots that we discussed earlier. While researchers previously demonstrated that not every trade reflects the actions of a real investor, a new conclusion has now emerged: to understand the market, it is important to consider not only the trade itself but also who executed it. Read more: Why Most MEV Bots Operate at a Loss…

Does that mean copy trading really works?

At first glance, the study’s results might seem to be solid confirmation of the concept of copy trading. If some wallets do indeed make more informed decisions, does that mean all we need to do is find them and simply copy their trades?

The authors of the study did not examine the profitability of copy trading nor did they compile a list of “ideal” wallets. Their goal was different—to determine whether the trading history of individual market participants contains additional information about future price movements.

The answer turned out to be yes.

Some wallets did indeed exhibit more informed behavior, and information about their actions helped to more accurately predict short-term market movements.

But there is a huge difference between understanding the market and making a profit.

Even if a particular trader consistently makes the right decisions, that doesn’t mean those decisions can be replicated without incurring losses. Commissions, slippage, order execution speed, and changes in market conditions can completely alter the final outcome.

Why Is This Important?

In recent years, dozens of copy-trading services and platforms for tracking large wallets have emerged.

Until now, most of them have been based on a simple idea:

"If this trader has been proven right time and again, it might be worth paying attention to his next moves."

A new study has shown for the first time that this idea may indeed have a scientific basis.

However, this does not mean that copy trading automatically becomes a profitable strategy.

The market is constantly changing. A trader who was successful yesterday may make mistakes tomorrow. Moreover, by the time information about their trade becomes known to other market participants, the market situation has often already changed.

Therefore, the study’s findings should be viewed not as a guide to action, but as a new understanding of where market advantage might actually come from.

This is precisely the main conclusion of this study.

The authors show that market analysis is gradually moving beyond charts, volume, and the order book.

Sources

• Daojing Zhai. “Who Trades Matters: Identity Information Improves Cryptocurrency Price Prediction.” Frontiers in Blockchain.
• Hyperliquid (data used in the study).