A few weeks ago, I started tinkering with an idea—building my own prediction market trading bot. The entire project took me several weeks from zero to a complete version. Honestly, is this kind of time investment worth it? My judgment is: yes, it is.

The reason is simple. Prediction markets indeed have insufficient price discovery at times, and although there are already bots exploring these opportunities, they are far from saturated. In other words, compared to the number of existing bots, the arbitrage space in this market is still very large.

So how does my bot work? The logic actually comes from a trading strategy I used to operate manually. Later, I wondered if I could automate this method to improve execution efficiency. Ultimately, I chose to run this bot on the BTC 15-minute up/down market.

Its core is a real-time monitoring program that automatically adapts to the current BTC 15-minute cycle. It uses WebSocket streams to fetch the best bid and ask prices in real-time, and the bot automatically places orders based on predefined logic. I backtested this strategy with historical data, starting with $1000, which grew to $1869 within a few days, achieving a backtest return of 86%.

Of course, while the backtest results look impressive, live trading will face real-world issues such as slippage, liquidity, and market volatility. Therefore, I documented the development approach, testing methods, and the actual limitations of this bot.