The Most Profitable Trading Bots on Polymarket Right Now

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.

Account: Bonereaper

2. 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

3. 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

> Where the edge comes from:

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

4. 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

Conclusion

The most interesting thing about these bots is that there is no single universal algorithm behind them.

Bonereaper uses a fast directional signal and rebuilds exposure through a dynamic hedge.

0xb55... tries to buy opposite outcomes during different phases of the move and convert them into complete sets.

mo-money changes its trading mode depending on the timeframe: directional logic on shorter markets and more neutral inventory management on longer ones.

almach focuses on execution quality and gradually builds positions through buy-side liquidity.

All four systems trade the same Up/Down contracts, but extract their edge from different parts of market structure: fair probability, volatility, timing, inventory balance, and execution.

This is why two transaction histories can look almost identical at the level of individual BUY fills while hiding completely different trading algorithms.

Thanks for reading! If you liked this article, follow me on X:
https://x.com/RetroValix