Author: Jeff, IOSG. The core data of the prediction market bot panic is quite intuitive: 5% of seemingly bot-like wallets on Polymarket contribute 75% of the platform’s trading volume. 823 wallets have each made over $100,000 in net profit since January 2025, collectively extracting $131.00 million in profit from Polymarket. Among the top 20 wallets on the profit榜, 14 are classified as bots (Stacy Muur leaderboard inspection). A University of Toronto study (covering 2.4 million users, with $67.00 billion in trading volume since 2022) found that 68.8% of users are in a loss-making state, with the top 1% of users capturing 76.5% of all profits. The narrative derived from this is that the prediction market is a wealth transfer machine, and bots are its operators. The data is correct, but the framework is half-biased.
Core Viewpoint 1: The core flaw in the bot narrative is equating “trading volume concentration” with “capital plunder.” The fact that 5% of wallets on Polymarket contribute 75% of the trading volume only illustrates the distribution of account activity and does not directly imply that retail investors’ funds are being extracted by bots.
-
Data at the group level is more convincing. AI agent wallets have a positive return rate of approximately 37%, while human wallets have only 7%-13%. A 3-4 times difference at the group level is real evidence of a structural advantage; and the fact that bots account for 14 out of the top 20 on the profit榜 (Stacy Muur leaderboard inspection) belongs to the right-tail projection of this distribution and is not independent evidence.
-
The advantage of bots lies in the structural dimension, not the judgment dimension. The three types of markets dominated by bots—price latency arbitrage, automated game state of real-time sports, and cross-platform portfolio arbitrage—have in common that they do not require judgment on real-world events themselves. Once the market outcome depends on the comprehensive processing of multi-source information, the advantage of bots is systematically weakened.
-
Polymarket’s category structure has shifted from “politics 42%” to “sports 50%” in the past 12 months, and the fastest-growing category is precisely the long-cycle event market where bots do not have a structural advantage, and the platform’s overall trend toward retail investors is clear.
-
Forward-looking judgment: The proportion of bots will continue to increase as deployment costs decrease, but the scale of capital extraction by bots from humans will peak before the proportion of bots does—because the rate at which bots cannibalize each other is faster than the rate at which they cannibalize human accounts.
-
Investment strategy: Equity opportunities at the platform level (Kalshi + Polymarket accounting for 97%+ share) are basically closed; value opportunities are migrating to the L2 agent infrastructure layer (Olas/Valory model) and the venue-agnostic middle layer, and C-end bot products and the L3 data/pricing layer do not have venture-fit.
-
The scale of the track is higher than the three quantitative anchor points of the bot panic, defining the scope of this report’s discussion. First, Bernstein revised the 2026E scale of the prediction market track to $240.00 billion on April 14, 2026, and the path to $1 trillion by 2030 has become a consensus on the sell-side. Second, the combined YTD trading volume of Kalshi and Polymarket exceeded $60.00 billion in mid-April 2026, exceeding the track’s total of $51.00 billion for the entire year of 2025. Third, Robinhood launched 1,000+ Kalshi contracts, and the platform’s 1 million + customers have cumulatively traded 9 billion contracts. Robinhood’s prediction market business ARR is approximately $350.00 million, $150.00 million for the full year of 2025, and $586.00 million for 2026E, making it the company’s fastest-growing product line. The above data collectively points to one conclusion: the prediction market is no longer a single crypto-native track, and its attribute is closer to a TradFi distribution problem. The main body of the “retail investors being plundered” group assumed in the bot narrative is not crypto users, but retail investors entering the market through traditional brokerage channels. From this, we can deduce the contextual deviation of the bot panic: the track is not being automatically extracted for value, but is being injected with traffic by mainstream finance at a pace far exceeding any automated extraction speed.
-
The really important data: 37% vs. 10%. The most frequently cited data point in the bot narrative has a sample selection bias. The source data for “14 out of the top 20 on the profit榜 are bots” is premised on a small sample that has already been sorted by profit. This sample can only reflect the occupancy of bots in the right tail of the distribution and cannot be used to infer the relationship between advantages and disadvantages at the group level. Group-level data (source: Polystrat/Valory disclosure, cross-validated with multiple Polymarket on-chain analysis data): A 3-4 times difference in win rate at the group level is a true reflection of the structural advantage of bots. The statistics of 14/20 on the profit榜 should be understood as the downstream performance of this win rate distribution, rather than independent causal evidence.
-
In which markets do bots win? The extraction scale of bots is highly concentrated in the following three types of markets. The common point of these three types is that they do not require subjective judgment on real-world results, but rely on latency or pricing advantages related to the platform’s matching engine.
Representative case of price feed latency arbitrage: Wallet 0x8dxd, which made $437,600 from $313 in January 2026, only traded 15-minute BTC rise and fall contracts with a win rate of 98%. Strategy principle: Monitor Binance and Coinbase spot prices and open positions when Polymarket quotes lag behind CEX. Polymarket introduced a taker fee (peak of approximately 3% near 50% probability) for 15-minute crypto contracts on January 7, 2026, to specifically neutralize this strategy. The wallet’s cumulative win rate has fallen back to 54.7%. Conclusion: The advantage of bots in feed-based markets is real, but it is limited to a very narrow time window and is significantly compressed as the platform introduces friction costs.
Real-time sports game state automation data source: Polymarket wallet classification by the cancun2026 team (Dune query 6648075, past 7 days, as of 2026-05-11). Source of advantage: Bots react to on-field events significantly faster than retail investors using live streams (30-second delay). In addition, trading terminals such as Kreo and PolyCop open this advantage to non-programmers through copy-trade and automatic follow-up functions, so the measured bot share includes human funds routed by bots.
Cross-platform portfolio arbitrage data source: IMDEA Networks paper “Unravelling the Probabilistic Forest: Arbitrage in Prediction Markets” (AFT 2025). The study covers approximately $40.00 million in arbitrage extraction scale on Polymarket between April 2024 and April 2025, mainly consisting of two modes: one is the rebalancing of YES/NO shares within the same market; the other is cross-platform portfolio trading (buying YES on Polymarket and NO on Kalshi, and entering when the sum of the implied probabilities of the two is less than $1). This model has rigid requirements for multi-platform infrastructure and is compressed as the matching engines of each platform converge.
- The areas where human accounts win and their limitations. The category with the lowest proportion of bots is not “retail investors picking more accurately,” but “the profitability of this type of market depends on the ability to integrate multi-source real-world information,” which is a link where automation continues to be at a structural disadvantage to humans. Two independent studies confirm this judgment. Joshua Della Vedova’s (University of San Diego) on-chain behavior study points out that retail investors pick the winning results more frequently than bots; the advantage of bots lies in execution—when retail investors buy YES at $0.72, bots have already entered at $0.55, with a floating profit of $0.17 per share. A working paper from the University of Toronto/HEC Montréal/ESSEC (Akey et al., SSRN 6443103, March 18, 2026) points out that 56% of losing users’ order prices fall in the extreme range (< 10¢ or > 90¢), while only 28% of the top 0.1% of profitable users place orders in the extreme range. The typical behavior of losing users is
[IOSG]
Prediction Markets: Bots Dominate, But Value Migration Creates New Opportunities
The recent revelation that 70% of top-profit wallets on prediction markets are bots has sparked widespread concern about fairness and market integrity. However, a deeper analysis reveals a more nuanced reality that has significant implications for crypto investors. While bots do hold a structural advantage in certain market segments, the narrative of “bots plundering retail investors” oversimplifies a rapidly evolving landscape where value is migrating from platforms to infrastructure layers.
Market Impact: Beyond the Bot Panic Narrative
The prediction market sector is experiencing exponential growth, with Bernstein projecting $240 billion by 2026 and a path to $1 trillion by 2030. This growth is being driven by mainstream adoption, evidenced by Robinhood’s prediction market business reaching an estimated $586 million in 2026E, making it the company’s fastest-growing product line.
The key insight here is that prediction markets are no longer purely crypto-native phenomena but are evolving into mainstream financial products. The “retail investors” referenced in the bot panic narrative are increasingly coming through traditional brokerage channels like Robinhood rather than crypto-native users. This fundamentally changes the risk/reward calculus for investors.
The structural advantage of bots is real but narrowly focused. As the data shows, AI agent wallets achieve a 37% return rate compared to 7-13% for human wallets—a significant but not insurmountable gap. This advantage is concentrated in three specific market types:
- Price latency arbitrage (e.g., monitoring discrepancies between CEX and prediction market feeds)
- Real-time sports game state automation (reacting faster to live events)
- Cross-platform portfolio arbitrage (exploiting pricing differences between platforms)
Notably, as Polymarket introduced taker fees for crypto contracts, the profitability of latency arbitrage strategies dropped from 98% to 54.7%, demonstrating that platform defenses can neutralize certain bot advantages.
Token Price Implications: Platform Premium vs. Infrastructure Play
The article’s assertion that “equity opportunities at the platform level are basically closed” should be heeded by investors. Both Polymarket and Kalshi now account for 97%+ of the market share, leaving little room for new entrants to capture significant value at the platform level.
However, this closure at the platform level creates a value migration effect that crypto investors cannot ignore. The investment opportunity has shifted to:
- L2 Agent Infrastructure Layer: Projects like Olas/Valory model that enable sophisticated trading strategies and bot operations
- Venue-Agnostic Middle Layer: Protocols that facilitate cross-platform arbitrage and portfolio management
- Data/Pricing Layer: Services that provide the multi-source information integration where humans still maintain an advantage
This migration suggests that tokens supporting infrastructure for prediction markets may see disproportionate returns compared to platform tokens themselves. The structural shift from platform dominance to infrastructure dominance mirrors the evolution we’ve seen in other sectors like DeFi and DEXs.
Risks: Beyond the Bot Narrative
While the bot narrative may be oversimplified, several genuine risks remain for investors:
-
Regulatory Scrutiny: As prediction markets gain mainstream adoption, they will inevitably attract increased regulatory attention, particularly around market manipulation and fairness concerns.
-
Bot Cannibalization: The article correctly notes that as deployment costs decrease, the rate at which bots cannibalize each other will outpace their extraction from human accounts. This could lead to diminishing returns for bot operators and increased volatility.
-
Platform Intervention: As seen with Polymarket’s introduction of taker fees, platforms will continue to implement measures to limit bot advantages, potentially disrupting certain profitable strategies.
-
Information Asymmetry: The 3-4x gap in win rates between bots and humans creates a concerning information asymmetry that could deter retail participation if not properly addressed.
-
Market Concentration: The dominance of a few platforms (Polymarket and Kalshi controlling 97%+ market share) creates a single point of failure and reduces the diversity of innovation opportunities.
Opportunities: Where Value is Being Created
Despite the risks, several compelling opportunities have emerged:
-
Sports Prediction Markets: The article notes that Polymarket has shifted from “politics 42%” to “sports 50%” in the past 12 months. Sports betting requires multi-source information integration where humans maintain an advantage, creating a more balanced playing field.
-
Long-Cycle Event Markets: The fastest-growing category is long-cycle event markets where bots do not have structural advantages. This represents a more sustainable segment for retail participation.
-
Infrastructure Enablement: The shift toward L2 agent infrastructure and venue-agnostic middle layers creates opportunities for protocols that facilitate sophisticated trading strategies while maintaining accessibility for non-technical users.
-
Cross-Platform Arbitrage Services: As the prediction market ecosystem matures, services that facilitate arbitrage between platforms will become increasingly valuable, particularly as platform matching engines converge.
-
Data Integration Solutions: While bots excel at speed and automation, they struggle with the comprehensive processing of multi-source information. Services that help bridge this gap between human judgment and machine execution will find significant demand.
Conclusion: Nuanced View Beyond the Bot Panic
The prediction market sector is undergoing a profound transformation, moving from crypto-native experimentation to mainstream financial adoption. While bots do hold structural advantages in specific market segments, the narrative of “bots extracting all value” is overly simplistic and fails to account for the rapid growth driven by mainstream finance.
For crypto investors, the key takeaway is that value is migrating from platform-level opportunities to infrastructure layers that enable sophisticated trading strategies while addressing the structural advantages of bots. The most promising opportunities lie in the L2 agent infrastructure, venue-agnostic middle layers, and services that facilitate the integration of human judgment with machine execution.
As the prediction market sector continues to evolve, investors who look beyond the bot panic narrative and focus on the structural shifts in value creation will be best positioned to capture the significant opportunities in this rapidly growing space.