Crypto without humans: How much of the market do bots control?

There was a time when a trader could gain an edge by seeing the news before everyone else and clicking a button faster. Today, a significant share of trading takes place at speeds beyond human capabilities, while in some crypto markets more than half of tracked volume is already linked to arbitrage and other automated strategies—so how far has automation gone?

The story began long before Bitcoin. In the late 2000s, Wall Street entered a technological arms race in which milliseconds became worth hundreds of millions of dollars and servers were placed as close to exchanges as possible.

The crypto market inherited this model almost ready-made but took it further: blockchain allows software not only to analyze the market but also to execute transactions independently.

View post on X Loading...

The tweet was deleted by the author.

But we saved everything 🙂.

Three milliseconds for $300 million

In 2010, Spread Networks completed construction of a nearly straight fiber-optic line between Chicago and New York. The roughly $300 million project cut round-trip signal latency from about 16 milliseconds to 13 milliseconds.

For a human, the difference is imperceptible. For a high-frequency trading firm, it was enough to spot a price change on one venue before a competitor and take advantage of it elsewhere.

The race continued: firms placed servers next to exchanges, optimized their code, and switched to microwave connections. As a result, an increasing number of trades were executed at speeds at which humans could no longer compete directly with algorithms.

In 2010, the U.S. Securities and Exchange Commission estimated that high-frequency trading firms accounted for more than 50% of U.S. equity trading volume. During the Flash Crash that same year, just 17 such participants accounted for 40–50% of dollar trading volume.

High-frequency trading was not identified as the sole cause of the crash, but the episode demonstrated a new reality: within minutes, the market could enter a chain of interactions between algorithms that humans were unable to control in real time.

Crypto was built for machines

Bitcoin emerged just as automated trading was gaining momentum. Electronic order books, application programming interfaces and algorithmic strategies became part of the crypto industry almost from the beginning.

Machines gained additional advantages here. Trading runs 24/7, the same asset is traded across dozens of venues, and price discrepancies between them constantly create opportunities for arbitrage algorithms.

A bot can monitor multiple markets simultaneously and execute both sides of a trade faster than a human.

Determining the overall share of such activity is impossible. Centralized exchanges do not disclose whether an order came from a person or an algorithm. On blockchains, however, automation is much easier to observe.

In August 2026, ClearTrace analyzed $7.13 billion in tracked weekly DEX volume on Base. MEV — extracting additional profit from transaction ordering — and arbitrage accounted for 55.96% of the volume.

View post on X Loading...

The tweet was deleted by the author.

But we saved everything 🙂.

Another 9.29% of volume could not initially be attributed at all. After further analysis, researchers identified some contracts, including MEV and arbitrage infrastructure.

That left $590 million, or 8.28% of tracked volume, unattributed.

The 55.96% figure cannot be extrapolated to the entire crypto market because it covers only ClearTrace-tracked Base DEX volume over a single week. But it provides a clear indication of how machine-driven some parts of trading have become when blockchain data makes them visible from the inside.

When a bot sees your trade first

In DeFi, algorithms gained an ability that traditional high-frequency traders did not have: they could see a user's transaction before it was executed and try to profit by placing their transaction before or after it.

This is how MEV works — additional profit that can be extracted by controlling or exploiting transaction ordering. The simplest example is arbitrage. If Ethereum trades at different prices on two DEXs, a bot buys it where it is cheaper and almost simultaneously sells it where it is more expensive.

Another example is liquidations. Programs continuously monitor borrowers' positions and race to liquidate them when their collateral becomes insufficient.

But automation can also work against users. In a sandwich attack, a bot detects a large purchase, buys the asset before it, waits for the user's transaction to push up the price and then sells at a higher price.

View post on X Loading...

The tweet was deleted by the author.

But we saved everything 🙂.

This is precisely why removing bots from DeFi would have consequences. Arbitrage eliminates price discrepancies between venues, while automated liquidators help maintain the stability of lending protocols. The same machines that can worsen the execution of an individual trade also help keep the broader system functioning.

The next stage: AI agents

A bot and an AI agent are not yet the same thing. Most trading algorithms simply execute logic predefined by humans: find a price discrepancy, check fees, and execute the trade.

An AI agent can go further. It can analyze multiple data sources, choose a strategy, and carry out a sequence of actions without requiring a separate human command at every step.

In a 2026 study, researchers examined 306 major crypto AI agents. About 13% primarily specialized in trading and analytics, while projects associated with them accounted for 21.9% of the total market capitalization of the sample—about $1.89 billion.

But it is too early to talk about machines taking over the market. Many projects described as agents still rely on conventional application programming interfaces and require human oversight. As a result, the true scale of fully autonomous trading remains difficult to measure.

View post on X Loading...

The tweet was deleted by the author.

But we saved everything 🙂.

Experiments, nevertheless, point to where the market may be heading. In one study, an autonomous AI trader consistently became the market's top performer, while human participants' wealth declined in its presence.

At the same time, the market itself did not become significantly more efficient.

This is the next stage of automation. In high-frequency trading, humans handed machines execution speed. With crypto bots, they handed over continuous market monitoring. With AI agents, they are gradually beginning to hand over the decision of what to do as well.

There is no single figure showing what share of the crypto market bots control today. Centralized platforms do not allow human and machine trades to be separated, while Base's 55.96% figure applies only to one specific segment.

Yet automation is no longer just another trading strategy. It has become part of the market's infrastructure.

Humans still set objectives, determine risk, and provide capital. But the direct competition for execution increasingly takes place between programs and at speeds where humans have long ceased to be serious competitors.

A crypto market without humans remains an exaggeration. But a market where humans set the goal while machines increasingly decide how and when to achieve it already exists.