Who is the trading weapon for bull and bears? In-depth analysis of the winners and losses of TG, AI agents and DEX/CEX trading robots

Author: ELAINE HU , Cointelegraph

Compiled: Tim, PANews

The crypto market has witnessed the massive adoption of automated trading solutions, with trading robots becoming increasingly prominent due to their ability to analyze massive amounts of data and execute transactions with precise accuracy.

This article deeply analyzes the returns performance of historical trading robots and the violent fluctuations in token prices. By backtesting and comparing strategic returns with the "buy and hold" benchmark, it decodes which robots perform the most outstandingly in a specific market cycle and their outstanding opportunities, so as to help you accurately match the best trading robots based on your own investment style and risk tolerance.

This paper studies three types of trading robots: Telegram-based DEX trading robots; non-Telegram robots traded on DEX and centralized exchanges; and the recently evolving AI proxy robots.

When choosing a suitable trading robot, you need to consider the user's goals, risk tolerance and experience.

in short:

Telegram robots are very suitable for opportunity-based trading scenarios such as token issuance and meme coins that require quick response and grasp the opportunity.

Artificial intelligence proxy robots (such as ai16z or Virtuals) are suitable for users who want to implement fully automated operations and are willing to try experimental strategies.

Robots in centralized exchanges provide maximum control and are particularly suitable for executing structured strategies such as fixed investment (DCA), grid trading or signal-based trading.

Robot trading strategies and performance

Trading robots are complex automation systems that analyze and encrypt market data through algorithmic analysis and encrypt transactions independently on centralized exchanges or decentralized platforms. This type of system usually operates continuously 24 hours a day, 7 days a week, with almost no manual intervention. Their core function is to analyze massive real-time and historical market data, including key indicators such as price fluctuations, transaction volume and order book information.

Using AI proxy trading robots has many potential advantages. Their continuous operation ensures that no trading opportunities are missed, as these robots can monitor the market 24/7 and adapt to dynamic changes in the global market. Some platforms that provide such robots also have backtesting functions, and users can use historical data to evaluate the potential effectiveness of different trading strategies before investing real funds in actual transactions.

Telegram DEX robot

Telegram trading robots operate on the Telegram platform, taking advantage of its convenience and real-time communication advantages, can directly execute transactions on decentralized exchanges. This type of robot is usually focused on increasing transaction speed and rushing tokens, especially attracting users in ecosystems such as Solana. The latest protocol also integrates common features of centralized exchange trading robots, such as grid trading, cost average method (DCA), and limit orders.

Telegram robots, such as Maestro and Unibot, originally appeared around 2020-2021. By 2022, many of these robots have begun to provide advanced features such as order trading and arbitrage.

By the end of 2023, Solana chain-based trading robots such as BONKBot and Trojan Bot will attract much attention for their ability to quickly trade Meme coins on decentralized exchanges. The biggest advantage of this type of Telegram robot is that users can complete transactions directly on mobile devices without connecting to their wallets through browser plug-ins. This feature greatly optimizes the ease of use of mobile transactions, while strengthening market monitoring functions and seamless integration with social networks.

Calculated by historical transaction volume, the top five Telegram robots in cross-blockchain platforms are Trojan, BonkBot, Maestro, Banana Gun and Sol Trading Bot. In the past 90 days, the vast majority of transaction volume has occurred on the Solana blockchain, and all of these five top Telegram robots operate on the Solana chain.

 DEX trading robot battle source: Dune Analytics 

Telegram robots offer very similar features, the difference is that some of them (such as Maestro and Banana Gun) focus on multi-chain operation, while the rest focus on Solana chains.

The main purpose of Telegram robots is to automatically identify profitable transaction entry and exit points and execute transactions quickly; however, because the individual user profits and losses generated by each transaction are difficult to track, some Telegram robots (such as Banana Gun and BonkBot) use the revenue sharing model bound to their tokens, that is, use the 1% fee charged to repurchase their own tokens, and use the token price and protocol income (the fees charged) as approximate indicators to measure the performance of the trading robot market.

 Telegram robots’ daily revenue (USD) Source: Dune Dashboard 

 Daily income to total income Source: Dune Dashboard

Telegram DEX robot revenue in the past 6 months

Trojan has the highest nominal expense amount (approximately $109 million), and Sol Trading Bot performs best on median daily income after standardizing daily income by total revenue.

They both peaked during the Meme coin boom around January 2025, but are currently facing a sharp drop in earnings due to a wider market downturn.

 Daily token price fluctuation percentage Source: Dune Analytics

BANANA and BONK token price performance

Price trends over the past six months show that the rest of the two Telegram robots that share income through tokens, except for the significant increase in BONK tokens in November 2024, have performed very similarly. In the recent bearish market environment, both prices have experienced a sharp decline.

AI Agent Robot

AI proxy trading robots are highly complex automated systems that use artificial intelligence (AI) and machine learning (ML) algorithms to analyze cryptocurrency market data and execute transactions independently.

The term "agent" shows that this type of robot has a certain degree of autonomy and decision-making ability, which goes beyond the operating methods based on fixed rules in traditional automated trading systems. The most well-known AI agent frameworks are Virtuals and ai16z.

AI Agent Robot Comparison

Virtuals Protocol was launched on Ethereum Layer 2 network Base in October 2024. It is an AI proxy generation platform designed to simplify the creation and deployment process of AI proxy on the blockchain. Although the protocol is not designed for transactions, its platform supports the development of AI agents that may be used for transaction purposes. For example, Aixbt, an experimental AI agent on the platform, can track social media X's discussion hotspots, identify potential market signals by analyzing user conversations, thereby generating strategic recommendations that can assist transaction decisions.

Since the Virtuals protocol adopts a LaunchPad model where various agents (such as LUNA and AIXBT) are issued as independent tokens and operate in multiple different fields such as gaming, trading and entertainment, we will focus on analyzing the market performance of AIXBT, which has the largest market value on the platform.

 AIXBT Price Historical Source: CoinMarketCap

Ai16z is an artificial intelligence-powered trading fund running on the Solana blockchain. The fund was launched in October 2024. Through complex AI agents built on the Eliza framework, it can independently analyze market information such as market price fluctuations, social media sentiment, and on-chain data, and automatically execute transaction decisions.

The fund operates in the form of a decentralized autonomous organization (DAO), allowing its native token holders to participate in key decision-making governance by voting and affect trading strategies through a "virtual trust market." AI Marc, a virtual fund manager built by the Eliza framework, manages fund trading activities. AI16Z tokens represent fund ownership and grant governance rights, and the operations of its agents will directly affect the value of the token.

 AI16Z price historical source: CoinMarketCap

By comparing the transaction volume data of these two agents, it can be found that they both peaked in January 2025, with AI16Z trading volume of $501 million, while AIXBT hit a higher record with a transaction volume of $682 million. It is worth noting that the price peak of AI16Z appears slightly earlier than its trading volume peak, while the price and trading volume peak of AIXBT are basically synchronized during the same period, and the two show a closer correlation in the time dimension.

 Comparison of prices and transaction volumes of AI16Z and AIXBT. Source: CoinMarketCap

AI16Z and AIXBT price performance

AIXBT's price performance is more eye-catching than AI16Z. During the peak period in November 2024, its token price soared by nearly 4,000 times compared with the initial issue price, while AI16Z only reached an increase of about 111 times during the same period. Even though it has experienced a decline recently and the overall market is in a downward trend, the latest price record as of the end of March 2025 shows that AIXBT still maintains a 478-fold increase in the initial price, while AI16Z maintains a 6.8-fold increase.

DEX/CEX robot

These platforms are built on the web and run independently of Telegram. Users can either directly trade on DEX through wallet connection, or connect to CEX through API interface or simple login. These functions are all part of the platform-integrated exchange solution.

These network-based platforms provide a wide range of policy choices and a broader market access channel, and take into account users' demand for CEX liquidity and reliability in functional design, as well as the unmanaged and decentralized features of DEX. Some platforms also support quick switching between DEX and CEX through one-click operation, making it more convenient to capture the price difference between centralized exchanges and decentralized exchanges (i.e. CEX-DEX arbitrage opportunities).

The most common strategies on these platforms include grid trading, average costing method, and signaling robots. The average costing robot puts a fixed amount of money into cryptocurrencies at fixed time intervals (regardless of how asset prices fluctuate). The core idea of ​​this strategy is to disperse entry points through time dimensions, thereby reducing the impact of market volatility. This type of strategy often performs well in markets where prices show a clear trend.

Grid trading robots are designed for active trading and make profits from price fluctuations through structured low buying and high selling strategies. The robot will place a series of limit-price buy orders and sell orders at the set interval above and below the preset price range, thereby creating an order "grid". Every completed low buy and high sell cycle will generate returns, and this strategy performs best in a sideways and volatile market.

Signal trading robots perform transactions based on external signals, which usually come from technical indicators, market analysis or third-party services. Common signal indicators include the relative strength index (RSI), exponential moving average (EMA), Bollinger Bands, etc.

Comparison of different DEX/CEX strategies

The following table shows the historical performance of the three token pairs BTC/USDT, ETH/USDT and SOL/USDT under three trading strategies. Among them, the parameter selection of grid trading robots uses the built-in AI optimization function of the 3Commas platform to automatically select the best parameters; while for the average cost method, the most popular classic trading strategy among users of the platform is adopted.

For signal robots, Dash2Trade provides policy preset functions, and the system will automatically select the optimal strategy for each token. These strategies are based on the independently developed trading system for historical backtesting and verification, and have been applied to real-time market trading, but the validity period of the strategy is only limited to trading cycles within 120 days before January 26, 2025.

Due to the lack of consistently available data on each platform, we adopted three backtest cycles for each strategy. The following table shows the simple price movements in the corresponding period, and this value also represents the rate of return of a simple benchmark buy and hold strategy.

DEX/CEX robot strategy rise and fall

The available data show that there may be significant differences in the performance of trading robots, which depends on the specific trading robots used, the trading strategies used, and the market conditions at the time of backtesting.

 Bitcoin and Ethereum price data source: CoinMarketCap

During the 120-day backtest period from September 26, 2024 to January 26, 2025, Bitcoin, Ethereum and Solana all showed an upward trend, with their buy-holding returns of 58%, 23% and 55% respectively. During this period, the strategic performance of the signal robot was basically the same as the buying and holding strategy (some currencies were slightly inferior), and the specific performance was: the return rate of Bitcoin strategy was 58.15%, the Ethereum strategy was 16.79%, and the Solana strategy was 48.68%.

During the same 120-day cycle from December 4, 2024 to April 4, 2025 (during the period, the grid robot strategy was backtested), the market prices of Bitcoin, Ethereum and Solana all showed a downward trend, with the returns of their buy-holding strategies being -16%, -53% and -49% respectively. This is completely different from the market environment of the previous 120-day backtest cycle. In a market with obvious downward trend and severe market volatility, the performance of the grid robot strategy is significantly better than the buy-holding strategy, creating positive returns for BTC (9.6%), ETH (10.4%) and SOL (21.88%).

 Bitcoin and Solana price data source: CoinMarketCap

During the longest 180-day backtest period from October 4, 2024 to April 4, 2025, when backtesting the fixed investment robot, the buy-holding strategy returns of Bitcoin, Ethereum and Solana were 34%, -25%, and -18%, respectively. In sharp contrast, the signal robot's strategy performs very differently to the buying and holding strategy for these three tokens.

For Bitcoin, the use of DCA robots achieved a return of 17.75%, which performed weaker than a one-time buy and hold strategy. However, DCA returns of Ethereum (ETH, 58.12%) and Solana (SOL, 80.92%) were significantly outperforming one-time investments. This difference may be due to the fact that ETH and SOL have significantly higher volatility than BTC during the statistical cycle, while DCA strategy effectively reduces timing risks by investing funds in batches and spreading the entry prices.

Trading robot performance comparison

DEX robots such as Trojan and Sol Trading Bot have dominated revenue performance over the past six months. Among them, Trojan earned about US$109 million through handling fees, while Sol Trading Bot performed outstandingly due to its stable daily income relative to its scale.

However, revenue from all trading robots peaked during the Meme coin boom in January 2025, followed by a slowdown due to market sentiment. Robots pegged tokens (such as BANANA and BONK) also showed similar trends, and after a brief surge (especially BONK's outstanding performance in November 2024), they all fell sharply due to overall market trends.

The market for artificial intelligence agent robots showed explosive growth during the same period. Among them, the price of AIXBT tokens reaches 4,000 times the issue price, far exceeding AI16Z (111 times). Even after experiencing a market pullback, AIXBT remained at 478 times the issue price, while AI16Z fell to 6.8 times. In terms of trading volume, both peaked in January 2025, but the price and trading volume of AIXBT tokens continued to grow, indicating that its speculative buying momentum was strong.

CEX/DEX signals, grid trading and fixed investment robots show the importance of market conditions, and there are significant differences in the returns of these strategies compared with buy-holding strategies.

In the upward market environment (backtest period is September 2024 to January 2025), signal robots perform close to buy-holding strategies, but their returns are slightly lower or comparable.

Grid Robots performed well in downtrends and high volatility environments (backtest period December 2024 to April 2025), defeating the buy and hold strategy with a significant advantage, reversing the market negative returns to double-digit positive returns.

During the 180-day backtest period (October 2024 to April 2025), the performance of DCA robots was significantly differentiated: although the fixed investment strategy of Bitcoin was underperformed by the simple holding strategy, the fixed investment of Ethereum and SOL was significantly outperformed, which is likely to be due to the ability of this type of strategy to absorb market volatility and volatility arbitrage advantages.

Key Points

We explored the wild world of AI-powered cryptocurrency trading robots: Comparison of decentralized exchange robots based on Telegram, AI agent trading robots, and DEX/CEX robots. Each robot is a unique tool tailored to different traders and market conditions.

Telegram DEX robot is designed with ease of use as its core design, and its simple operating interface is directly embedded in Telegram applications. This type of robot focuses on Meme currency trading and on-chain token issuance, mainly attracting investors and Meme currency enthusiasts who are proficient in mobile operations and pursue fast trading. Its special features include social order transactions and token dividend incentive mechanisms, deeply integrating social attributes to meet users' needs for instant transactions and social collaboration.

Telegram DEX robots have generated considerable revenue in the past 6 months and peaked in the Meme quarter in January 2025. However, among these projects, only two tokens BANANA and BONK share the income with users. In the recent bear market environment, the prices of both tokens have experienced a sharp decline, with BONK token plummeting 90% in three months, and the market value of BANANA tokens also shrunk to less than one-third of the peak period. Data shows that although the overall transaction volume of the DEX robot track remains active, the sustainability of the token economic model is facing severe tests.

AI agent robots lower the threshold for users to participate in governance (such as the DAO model of AI16Z) or implement emotion-driven strategies (such as the X analysis of AIXBT) through natural language interfaces and AI decision-making capabilities. Its core advantage lies in the ability to abstractly present complex trading strategies through a dialogue interface.

Although the token price of AI proxy robots has shown explosive growth, the recent market downward trend has led to a decline in trading activity and the token price has declined. As a more experimental field, AI agent robots are still in development and are best suited for users who are curious about technology or pursue solutions combining simplicity and automation.

Trading robots running on DEX and CEX provide diversified strategies through web platforms, making them more suitable for experienced traders who require high-speed execution, multi-exchange access, high liquidity and complex configuration capabilities. Backtest data shows that the return of signal robots is comparable to that of buying and holding strategies in the bull market, and grid robots perform outstandingly in the market for declining volatility. For high volatility assets, the returns of fixed investment robots can surpass the buying and holding strategies.