Polymarket 上目前最賺錢的交易機械人

Using Claude, I analyzed every trade made by some of the most profitable trading systems on Polymarket and reconstructed their logic inside each market to understand how they consistently grow their capital and have already reached the 6 and 7-figure range.

Below, I will break down the strategy behind each trading bot, where its edge comes from, and why these systems continue to generate positive PnL while many of their competitors are no longer profitable.

1. Bonereaper (Directional Rotation with Dynamic Hedge) Bonereaper operates as a directional system. The algorithm forms a view on the current direction of the underlying asset, builds a position in the corresponding outcome, and then continues recalculating fair probability as the market moves. If the signal changes, the bot usually does not close the old position. Instead, it starts accumulating the opposite outcome.

Its cycle looks roughly like this:

Model detects an edge in Up -> Bot starts buying Up -> The underlying asset reverses -> Fair probability shifts in favor of Down -> Bot starts buying Down -> The original Up remains part of the total position, while the opposite side acts as a dynamic hedge.

This is especially visible in 5m markets. Bonereaper accumulated both sides in roughly 55% of these markets, while the median delay between the first entry on one side and the first entry on the opposite side was only around 26 seconds. At the same time, positions usually remained unbalanced: the smaller side was only around 45% of the larger side on a median basis.

The combined average price of Up and Down in two-sided 5m positions was around $1.08. This means it is not complete-set arbitrage. The bot pays for the hedge to reduce directional exposure after the signal changes.

Where the edge comes from:

The main objective of the algorithm is to recalculate the probability of Up and Down faster than Polymarket after the price of the underlying asset changes. To do this, the model takes into account external spot prices, order book depth, short-term momentum, movement speed, distance from the opening price, and the time remaining until resolution.

If fair probability has already changed while the contract price has not yet adjusted, a short-term pricing gap appears.

Bonereaper tries to capture that gap with the first side of the position. If the market continues moving in the expected direction, the bot can increase exposure. If the signal breaks, the opposite outcome limits the impact of the original entry.

Instead of a single binary position, the result is a position that is constantly being rebuilt.

1. 0xb55… (Asynchronous Complete-Set Accumulator) The second bot has a completely different structure. Both sides of the position appear much more systematically. It buys both Up and Down in roughly 75% of all markets. On 15m markets, this rises to 79%, on 1h to 87%, and on 4h to 100%.

But this is not classic arbitrage where Up and Down are purchased at the same time. On 5m markets, the median gap between the first entry on one side and the first entry on the opposite side was around 52 seconds. On 15m markets, it was already close to three minutes. This means the algorithm uses price movement across time.

For example:

Up temporarily becomes cheap -> Bot accumulates Up -> Some time later, the underlying asset reverses -> Down becomes cheaper -> Bot starts accumulating Down -> Part of the two positions becomes a complete set

The economics then look like this:

Complete-set cost = VWAP(Up) + VWAP(Down)

If the final cost is below $1, the matched part of the position has a positive structural edge regardless of which outcome finishes higher at settlement.

This effect becomes especially visible on longer timeframes. On 1h markets, the median combined price of both sides was around $0.93, while on 4h markets it was around $0.91. This looks similar to temporal arbitrage.

Where the edge comes from:

Up and Down are constantly repriced as the underlying asset moves. The algorithm does not need to see the following prices at the same time:

Up = $0.42 Down = $0.55

It can buy Up now and buy Down several minutes later, after the entire pricing structure has changed. The longer a market remains open, the more opportunities the bot has to acquire both sides during different phases of the move.

This is why 1h and 4h markets are particularly suitable for this model. During a single window, the underlying asset can move up and down several times, creating attractive levels first for one outcome and then for the other. Part of the inventory becomes neutral, while the unmatched remainder keeps directional exposure.

Account: 0xb55fa1296e6ec55d0ce53d93b9237389f11764d4

1. mo-money (Multi-Timeframe Hybrid Bot) What makes mo-money interesting is that it does not use the same model across every timeframe. On shorter markets, it looks primarily directional. On longer markets, its behavior gradually shifts toward two-sided inventory management.

On 5m markets, both sides appeared in only around 32% of markets. On 1h markets, this increased to 67%, and on 4h markets to roughly 85%. Both the hedge frequency and the structure of the position change.

In two-sided 5m markets, the smaller side was only around 43% of the larger side on a median basis, while the average cost of the complete set was around $1.07. This means the bot keeps a clear directional bias on shorter timeframes.

On 1h markets, the sizes of Up and Down in two-sided positions were almost identical, while the median sum of their VWAPs was around $0.91. In practice, the same bot switches between two different modes:

• On shorter timeframes

Identify the current bias -> Buy the primary outcome -> Add to the position while the signal remains valid -> Use the opposite side only when necessary -> Keep a noticeable directional remainder

• On longer timeframes:

Start with one side -> Wait for a new price range -> Accumulate the opposite outcome -> Match a large portion of the contract quantities -> End up with nearly neutral inventory
Different timeframes have different market microstructure.

On 5m markets, there is very little time. It is more efficient for the algorithm to react quickly to a short-term signal than to wait for the perfect entry on the second side.

On 1h or 4h markets, the situation changes. The underlying asset has enough time to move through several local swings, allowing the bot to use volatility to improve the average price of both sides.

Account: mo-money

1. almach (Buy-Side Inventory Builder) almach uses another variation of a buy-only model. Its algorithm continuously places buy orders at different price levels and allows the market to gradually fill its inventory.

The logic behind almach becomes especially clear on longer timeframes. On 5m markets, both sides were present in only around 23% of markets. On 1h markets, this increased to 72%. On 4h markets, it reached 96%.

When both sides were built, the inventory was also significantly more balanced than Bonereaper’s.

On 1h markets, the smaller side was around 96% of the larger side on a median basis, while the combined average price of Up and Down was around $0.97. On 4h markets, the balance was around 92%, with a combined cost of roughly $0.99.

This strongly resembles a system designed not to buy a complete set immediately, but to build it gradually through passive BUY orders during different phases of the market.

How the algorithm works:

Place BUY liquidity on one or both sides -> Receive fills when price reaches the target levels -> Start accumulating the second side during the next move -> Match equal quantities of Up and Down into a neutral component -> Keep unmatched inventory only where the price justifies it

Where the edge comes from:

The main advantage here is execution quality. Instead of crossing the current ask and immediately paying for the full size of the position, the algorithm waits for the market to move toward its own levels.

For example, the bot gradually gets:

Up at an average entry of $0.46 Down at an average entry of $0.51

The total cost of the matched portion is $0.97

Those $0.97 prices did not need to exist in the order book at the same time. One side could have been filled much earlier than the other. The longer the market remains open, the more time the system has to wait for attractive prices on both sides. This explains why the share of two-sided positions for almach increases sharply on 1h and 4h markets.

Account: almach

Not every trading bot makes money in the same way. Even when several bots trade the same Polymarket Up/Down markets, their strategies can be completely different.

Let’s break down four interesting trading approaches:

🔹 Bonereaper — Directional Trading + Hedging

This bot looks for signals that suggest the market will move up or down. It takes a position based on that signal, then uses hedging to manage risk and adjust its exposure as the market changes.

🔹 0xb55… — Buying Opposite Outcomes

This strategy buys opposite outcomes at different prices or stages of a market move. The goal may be to combine them into a complete set, where one outcome pays $1 in total, potentially creating a profit if the combined purchase cost is below $1.

🔹 mo-money — Adaptive Trading

This approach changes its strategy depending on the market timeframe. Shorter markets may favor directional trading, while longer markets may call for more balanced inventory management.

🔹 almach — Execution-Focused Trading

Instead of buying a large position all at once, this approach gradually builds positions through buy orders. The goal is to manage entry prices, liquidity, and execution costs more carefully.

🚀 What Does This Mean for My Trading Bot?
My project explores the world of automated Polymarket trading:

💻 GitHub:

Benj4minCup / Polymarket-Trading-Bot-python

polymarket trading bot polymarket twap bot polymarket arbitrage bot polymarket bot polymarket trading bot polymarket twap bot polymarket arbitrage bot polymarket bot polymarket trading bot polymarket twap bot polymarket arbitrage bot polymarket bot polymarket trading bot polymarket twap bot polymarket arbitrage bot polymarket bot polymarket trading

Polymarket Trading bot system

An open-source and Strong Strategy collection of Polymarket trading bot and Polymarket arbitrage bot and Polymarket TWAP trading bot in Python for high-performance automated trading on polymarket crypto 5min and 15min markets.

This repository is primarily intended for educational and research purposes. It includes strategy concepts, implementation approaches, and selected performance screenshots to help developers understand how different automated trading strategies can be designed and tested.

The repository does not provide a complete production-ready trading bot source code. Instead, it provides strategy descriptions and research materials that you can use as a foundation for developing your own system.

If you are interested in building a Polymarket Trading Bot, you can follow my tutorials and use the concepts in this repository to develop your own implementation.

For users who prefer a ready-to-deploy solution or require custom strategy development, commercial bot development and customization are also available.

Features

…

🎥 YouTube:

📩 Telegram: https://telegram.me/BenjaminCup

The key lesson is that a trading bot is more than a collection of BUY orders. Its real strategy depends on why it enters a position, when it exits, how it manages risk, how it balances inventory, and how it handles execution.

Two bots can produce similar transaction histories while following completely different strategies.

The real challenge is understanding the logic behind every trade and building a system that can manage risk in changing market conditions.

Trading involves risk. These strategies do not guarantee profits, and a complete set is profitable only when its total cost and fees are below its eventual payout.

Polymarket #TradingBot #AlgorithmicTrading #CryptoTrading #AutomatedTrading #Python #TradingStrategies #QuantTrading #PredictionMarkets #MarketMaking #RiskManagement #OpenSource #AITrading