On November 22, 2025, on a Polymarket prediction market, a silent duel was unfolding.

On one side of the duel was a mysterious trader named @totofdn. On the other side was an automated arbitrage bot named sunshines.

It all started with a seemingly insignificant order. @totofdn placed a tiny sell order: 5 shares of No @ $0.34. This action instantly compressed the market's bid-ask spread to less than $0.04—exactly the magic number to trigger the platform's "order book reward."

Almost in the same second, sunshines reacted. A large sell order slammed into the order book: 100 shares of No @ $0.34. The bot arrived, strictly following its code's instructions, to earn the platform's liquidity reward.

Unbeknownst to it, this was the signal @totofdn had been waiting for.

@totofdn unhesitatingly devoured all of the bot's sell orders, pocketing 100 shares of No at an average price of $0.34. Meanwhile, the bot was left holding 100 shares of Yes at an average price of $0.66, unaware that it had fallen into a trap.

This was just the beginning. Over the next four hours, this "fake order real eat" combo was repeated multiple times. sunshines was like a runaway ATM, spitting out cash repeatedly. Four hours, dozens of repeated operations, over $1500. The bot's account was precisely drained, while @totofdn emerged unscathed, slipping away quietly.

This was a meticulously planned "cognitive siege" targeting automated scripts. It revealed a truth about on-chain arbitrage: here, automation does not equal intelligence, AI can improve efficiency but also potentially make you lose more.

The Platform Incentive's "Optimal Solution" and "Fatal Flaw"

To understand the intricacies of this battle, we must first revisit Polymarket's rules. As a decentralized prediction market, liquidity is its lifeblood. To incentivize users to provide depth to the market, Polymarket has designed a mechanism called the "Order Book Rewards Program."

The core idea of this mechanism is simple: those who provide liquidity to the market will receive rewards. Specifically, as long as users place their limit orders within the blue maximum spread line (the so-called "spread") in a designated market and meet certain share requirements, they can proportionally share the platform's provided reward pool. Rewards are distributed automatically at midnight every day, simple and straightforward. This spread is usually the current mid-price ±3ct to 4ct, with the specific width set by Polymarket in real-time.

Once any rule is quantified, it will inevitably give rise to specialized "gaming" strategies. Polymarket's order book reward quickly attracted a group of particular "miners." They don't care about the outcome of the predicted event itself; they only care about how to efficiently obtain the reward. Thus, automated arbitrage bots like sunshines emerged.

The logic of these bots is as follows:

Market Scanning: Continuously monitor all markets that meet the reward criteria.

Price Spread Analysis: Check if the current market's buy/sell price spread is less than a certain threshold (e.g., $0.04).

Order Placement: Once a market is found with a price spread that meets the liquidity reward requirement, immediately place an order within the spread that complies with the reward rules.

Reward Collection: Wait for the midnight reward distribution.

From a code perspective, this logic is flawless as it perfectly leverages the rules. The bots tirelessly "fill the spread" in various markets, contributing liquidity data to the platform and receiving rewards in return. They are the "optimal solution" to the rules, the "model citizens" in the eyes of Polymarket.

However, the issue is that these bots only analyze price spreads, shares, and rewards. They are oblivious to market sentiment, order book analysis, and lack risk control. They cannot differentiate whether the sudden appearance of a tiny order that compresses the spread to the triggering condition is a genuine trade demand or a carefully crafted trap.

When @totofdn placed a 5-share No @ $0.34 sell order, sunshines' code told it, "An opportunity has arisen! The spread has been compressed to 1¢; quickly place an order to claim the reward!" It was completely unaware that this 0.01¢ spread was false, artificially created. It only saw the "optimal solution" of the rule but failed to see the "fatal flaw" behind this solution.

Ultimately, this reward-driven bot, due to its mindless pursuit of rewards, also became prey in the hands of more sophisticated hunters.

From Physical Battle to Cognitive Battle

From MEV to Jito, and now to Polymarket's "bot hunting," the on-chain arbitrage, this smokeless war is undergoing a profound evolution.

If the early MEV (Maximal Extractable Value) war was a "physical battle" revolving around gas fees and block space, then today's on-chain game is increasingly resembling a "cognitive battle" testing strategy and psychology.

In the Wild West era of MEV, victory belongs to those who have the fastest network, the most powerful hardware, and the highest priority in block production— the "cowboys" of the era. They act like a group of trucks speeding down the highway, using sheer power and speed to front-run, sandwich, and liquidate, extracting value from ordinary users' transactions. It was a time of brute force, where the comparison was whose "muscles" were more developed.

Subsequently, MEV solutions emerged, with Jito as a representative, attempting to establish a new order for this chaotic physical battle. By auctioning block space, Jito redistributed the gains from MEV, allowing validators and stakers to also get a piece of the pie. This to some extent alleviated network congestion but also made MEV acquisition more "legitimized" and "industrialized." The war shifted from the shadows to the light, from individual heroism to an arms race between professional institutions.

The recent event on Polymarket reflects a new stage in on-chain gaming. The determining factor is no longer millisecond-level delays or sky-high gas fees but rather an understanding of the rules, insights into market players, and the application of strategies.

@totofdn did not employ any sophisticated hacking techniques or mobilize significant computational resources. His sole weapon was a deep understanding of the Polymarket reward mechanism and precise anticipation of automated script behaviors like sunshines. He won an information asymmetry war and even more, a cognitive dimension war.

The rules of the on-chain dark forest are evolving. Simple automated scripts, without the ability to dynamically adapt to the environment and engage in game theory with opponents, will increasingly struggle to survive. They are like evolutionarily incomplete species— highly efficient in a specific ecological niche (such as reward farming), but once the environment changes or when faced with more advanced predators, they are helpless.

From the physical battle of MEV to Jito's order battle, and now to Polymarket's cognitive battle, on-chain arbitrage is transitioning from an "engineer's" game to a game of "strategists" and "psychologists." In this increasingly complex dark forest, only those participants who can continuously evolve and elevate their cognitive dimension can ultimately survive.

Welcome to join the official BlockBeats community:

Telegram Subscription Group: https://t.me/theblockbeats

Telegram Discussion Group: https://t.me/BlockBeats_App

Official Twitter Account: https://twitter.com/BlockBeatsAsia