The cryptocurrency market has witnessed a surge in the adoption of automated trading solutions, with trading bots gaining prominence for their ability to analyze vast data sets and execute trades with precision.

Cointelegraph has dissected historical bot revenues and token price rollercoasters and backtested strategy returns against the buy-and-hold yardstick to decode what bots shine brightest — and when — so you can pick the perfect bot to match your style and stomach for risk.

We have examined three types of trading bots: Telegram bots trading on decentralized exchanges (DEX), non-Telegram bots trading on DEXs and on centralized exchanges (CEXs), and the recently evolving AI agent bots.

Choosing the right trading bot depends on the user’s goals, risk tolerance and experience. At a glance:

• Telegram bots are ideal for fast, opportunistic trading like token launches and memecoins.

• AI agent bots, such as ai16z or Virtuals, suit users who want hands-off automation and are comfortable with experimental strategies.

• CEX bots offer the most control and are best for structured strategies like dollar-cost averaging (DCA), grid or signal-based trading.

Bot trading strategies and performance

Trading bots are sophisticated automated systems that use algorithms to analyze cryptocurrency market data and autonomously execute trades on centralized exchanges or decentralized platforms. These bots typically operate continuously, 24 hours a day, seven days a week, requiring minimal human oversight. Their core function involves the analysis of extensive amounts of real-time and historical market data, including price fluctuations, trading volumes and order book information. 

There are numerous potential advantages to employing AI agent trading bots. Their continuous operation ensures that no trading opportunities are missed, as they can monitor markets around the clock, accommodating global market movements. Some platforms offering these bots also provide backtesting capabilities, enabling users to evaluate the potential effectiveness of different trading strategies using historical data before deploying them with real capital.

Telegram DEX bots 

Telegram bots operate via Telegram, leveraging its accessibility and real-time communication to execute trades directly on DEXs. They often focus on speed and sniping new tokens, appealing to users in fast-moving ecosystems like Solana. The recently launched protocols also included additional features that are often available in CEX trading bots, such as grid trading, DCA and limit orders.

Telegram bots such as Maestro and Unibot first appeared around 2020–2021. In 2022, many of these bots were already offering advanced features like copy trading and arbitrage.

By the end of 2023, Solana-based bots like BONKBot and Trojan Bot gained prominence for their speed in trading memecoins on DEXs. The biggest advantage of Telegram bots is their ability to trade on mobile devices without the need for a web browser extension to connect to a wallet. It hugely improves the usability of mobile trading, monitoring and integration with social networks.

The top five Telegram bots by historical trading volume across all blockchains are Trojan, BonkBot, Maestro, Banana Gun and Sol Trading Bot. The majority of the trading volumes in the past 90 days happened on Solana, where all of the top five Telegram bots operate.

The functionalities offered by the Telegram bots are very similar, with the exception that some of them (i.e., Maestro and Banana Gun) focus on multichain operations, whereas the rest focus on Solana.

The main use case for Telegram bots is to automatically identify profitable entry and exit points and execute trades quickly; it’s very difficult to track the profits or losses made by individual users from each trade. Since some of the Telegram bots, such as Banana Gun and BonkBot, offer a revenue-sharing model tied to their own tokens in the form of purchasing back their tokens with the 1% fee they charge, the token price and revenue (fees received) are used as an approximation of the performance of Telegram bots.

Looking at the total revenue in the past six months, Trojan has received the most nominal amount in fees (around $109 million), whereas Sol Trading Bot has the highest median daily revenue when normalizing the daily revenue in terms of the total revenue.

They all saw a peak around January 2025 during the memecoin season but are now facing a low-revenue period due to the broader bearish market conditions.

The two Telegram bots that share revenue through their tokens are Banana Gun and BonkBot. Looking at the price evolution in the past six months, the performance of the remaining parts is very similar, except for the significant rise in BONK’s price in November 2024. They both experienced significant price drops during the recent bearish market conditions.

Related: The whale, the hack and the psychological earthquake that hit HEX

AI agent bots 

AI agent trading bots are sophisticated automated systems that leverage artificial intelligence and machine learning (ML) algorithms to analyze cryptocurrency market data and autonomously execute trades.

The term “agent” suggests these bots possess a degree of independence and decision-making capability that extends beyond the fixed rules of traditional automated trading systems. The most well-known AI agent frameworks that exist today are Virtuals and ai16z.

Virtuals Protocol, launched in October 2024 on the Ethereum layer-2 network Base, is an AI agent generator platform designed to simplify the creation and deployment of AI agents on the blockchain. While Virtuals is not solely focused on trading, the platform enables the development of AI agents that could potentially be designed for trading purposes. For instance, Aixbt, an experimental AI agent on the platform, tracks discussions on X to identify potential market insights, suggesting a strategy that could inform trading decisions.

Since Virtuals Protocol focuses on a launchpad model where agents are tokenized individually (e.g., LUNA and AIXBT) and operate across different areas such as gaming, trading and entertainment, we’ll only look into the performance of AIXBT, the token of the trading agent with the largest market capitalization on Virtuals.

Ai16z is an AI-powered trading fund operating on the Solana blockchain. Launched in October 2024, ai16z utilizes sophisticated AI agents, powered by the Eliza framework, to autonomously analyze market data, including price movements, social media sentiment and onchain analytics, and execute trades.

The fund functions as an AI investment decentralized autonomous organization (DAO), allowing holders of its native token to participate in governance by voting on key decisions and influencing trading strategies through a “virtual marketplace of trust.” AI Marc, a virtual fund manager built using the Eliza framework, oversees the fund’s trading activities. AI16Z tokens represent ownership in the fund and grant governance rights, with the agent’s actions driving token value. 

Comparing the trading volumes from these two agents, they both reached a peak in January 2025, with AI16Z reaching $501 million and AIXBT reaching $682 million. AI16Z’s price hit its peak slightly earlier than its volume high, whereas for AIXBT, the price and volume peaks coincided around the same time.

AIXBT’s price performance is more impressive than AI16Z. At the peak, the token price was almost 4,000x the initial price in November 2024, whereas for AI16Z, this was around 111x. Even after the recent downturn and the broader market trending down, the latest price record at the end of March 2025 is still 478x the initial price for AIXBT and 6.8x for AI16Z.

DEX/CEX bots 

These platforms are web-based and operate outside Telegram. You can trade directly on DEXs through wallet connection or connect to a CEX via APIs or a simple login option as part of their integrated exchange solutions.

These web-based platforms offer a wide range of strategies and broader market access; they cater to users preferring both CEX liquidity and reliability as well as DEX’s decentralized, non-custodial nature. Some of these platforms also offer a quick switch between DEX and CEX with one click, making the discovery of price discrepancies between CEX and DEX (or CEX-DEX arbitrage) much easier.

The most common strategies available on these platforms are grid, DCA and signal bot. A DCA bot invests a fixed amount of money into a cryptocurrency at regular intervals — regardless of the asset’s price. The idea is to spread out your entry points over time, which helps reduce the impact of market volatility. This type of strategy tends to perform well during price-trending periods.

A grid bot is built for active trading — buying low and selling high in a structured way to profit from price fluctuations. A grid bot places a series of buy and sell limit orders at preset intervals above and below a set price range. This creates a “grid” of orders, and the bot profits from each completed buy-low/sell-high cycle. Grid bot works best in sideways markets with high volatility.

A signal bot executes trades based on external signals — these usually come from technical indicators, market analysis or third-party services. These signals can be relative strength index (RSI), exponential moving averages (EMA), Bollinger Bands, etc.

The following table shows the historical performance for the token pairs BTC/USDT, ETH/USDT and SOL/USDT for the three trading strategies. The parameter selection for the grid bot utilizes the 3Commas AI optimization built-in functionality to select the best parameters, whereas for DCA, the most popular classic trading strategy from their users is selected. 

For the signal bot, Dash2Trade provides strategy presets where the top strategy for each token is selected. These strategies are backtested on a proprietary system used to trade on live markets but are only available for the 120 days before Jan. 26, 2025.

Due to a lack of consistent availability of data on the platforms, three backtesting periods were used for each of the three strategies. The table below shows the simple price change during the corresponding period, which is also the return for the simple benchmark buy-and-hold strategy.

The available data suggests that performance can vary widely based on the specific bot, the trading strategy employed and the prevailing market conditions at the time the backtests were run. 

During the 120-day period from Sept. 26, 2024, to Jan. 26, 2025, when the signal bots were backtested, the market prices for Bitcoin (BTC), Ether (ETH) and Solana (SOL) were all upward trending with a buy-and-hold return of 58%, 23% and 55%, respectively. The signal bots’ strategies were performing in line with the buy-and-hold strategy (in some cases slightly worse) for BTC (58.15%), ETH (16.79%) and SOL (48.68%). 

Comparing the same 120-day period but from Dec. 4, 2024, to April 4, 2025, when grid bots were backtested, the market prices for BTC, ETH and SOL were all experiencing a downward trend, with a buy-and-hold return of -16%, -53% and -49%, respectively, which is completely different from the previous 120-day backtesting period. The grid bots’ strategies were performing much better than the buy-and-hold strategy during the downward-trending, high-volatility market conditions, giving positive returns for BTC (9.6%), ETH (10.4%) and SOL (21.88%).

For the longest 180-day backtesting period from Oct. 4, 2024, to April 4, 2025, when the DCA bots were backtested, the buy-and-hold returns for BTC, ETH and SOL were 34%, -25% and -18%, respectively. The signal bots’ strategies were performing very differently for the three tokens compared to the buy-and-hold strategy. 

For BTC, a 17.75% return is generated from the DCA bots, which is worse than the buy-and-hold strategy. However, for ETH (58.12%) and SOL (80.92%), the DCA returns are much better than the buy-and-hold returns. This might be due to the fact that ETH and SOL experienced much higher volatility during the period compared to BTC, and the DCA strategy was able to spread out entry prices to reduce exposure to bad timing.

Related: Market maker deals are quietly killing crypto projects

Trading bot performance comparison

Telegram DEX bots like Trojan and Sol Trading Bot dominated in revenue over the past six months, with Trojan earning about $109 million in fees. Sol Trading Bot stood out for consistent daily earnings relative to its size.

However, all bots saw revenue peak during the January 2025 memecoin hype and have since slowed due to bearish market sentiment. Token-linked bots (BANANA, BONK) followed a similar pattern — brief surges (notably BONK in November 2024) followed by steep drops tied to broader market trends.

AI agent bots showed explosive growth during the same period. AIXBT reached a peak price 4,000x its initial value, far outperforming AI16Z (111x). Even post-correction, AIXBT held strong at 478x vs. AI16Z’s 6.8x. Volume-wise, both peaked in January 2025, but AIXBT’s token price closely tracked its volume rise, suggesting strong speculative momentum.

CEX/DEX signals, grid and DCA bots showed the importance of market conditions, and the performance results vary quite a lot compared to the buy-and-hold strategy.

• Signal bots performed close to the buy-and-hold strategy during the uptrend market condition (backtesting period September 2024–January 2025), with marginally lower or similar returns.

• Grid bots excelled during a downtrend and high volatility environment (backtesting period December 2024–April 2025), beating the buy-and-hold strategy by wide margins, flipping negative market returns into double-digit gains.

• DCA bots over a 180-day backtesting period (October 2024–April 2025) had mixed results; they underperformed the buy-and-hold strategy for BTC but dramatically outperformed ETH and SOL, most likely due to their ability to absorb and capitalize on volatility.

Key takeaways

We have dived into the wild world of AI-powered crypto trading bots, pitting Telegram DEX bots, AI agent bots and CEX/DEX bots against each other — each a unique tool tailored to different traders and market conditions.

Telegram DEX bots are designed for ease of use, with a simple interface embedded in the Telegram app. These bots focus on trading memecoins or participating in token launches onchain. They appeal to mobile-savvy traders and memecoin enthusiasts who prioritize quick trades and social integration, with features such as copy trade and revenue-sharing through tokens.

Telegram DEX bots generated significant revenue in the past six months, peaking in January 2025’s memecoin season. But not all of them share revenue with the users. The only two who did (BANANA, BONK) faltered in the recent bearish market, with token prices dropping sharply. 

AI agent bots use natural language interfaces and AI decision-making to lower the barrier to entry for users interested in governance (e.g., AI16Z’s DAO model) or sentiment-driven strategies (e.g., AIXBT’s X analysis). Their primary strength lies in abstracting complex trading strategies through conversational interfaces.

Although AI agent bots’ token price exhibited explosive growth, the recent market downtrend has led to less trading activity and lower token prices. AI agent bots stand out as a more experimental category. They remain under development and are best suited for users who are tech-curious or seeking a hybrid between simplicity and automation.

Bots operating on DEXs or CEXs directly offer web-based platforms with diverse strategies, suiting more experienced traders who need high-speed execution, multi-exchange access, deep liquidity and complex configurability. The backtesting results show signal bots give similar returns to the bullish buy-and-hold strategy, whereas grid bots thrive in volatile downturn markets, and DCA bots outperform the buy-and-hold strategy for more volatile assets.

Magazine: Your AI ‘digital twin’ can take meetings and comfort your loved ones