#预测市场 Looking at the data from this AI prediction market on Kalshi, it's quite interesting. Gemini's share has jumped from 30% at the beginning of the year to 86%, with a trading volume of $14.08 million. The logic behind this shift is worth pondering.

Honestly, prediction markets are essentially voting with real money, which is more genuine than any public opinion guidance. The drop from 41% to 8% for ChatGPT reflects a significant reversal, likely indicating a market reassessment of each AI's recent performance. Gemini's dominant position is probably due to recent technological updates or performance improvements that have convinced traders to change their stance.

My experience is that participants in such prediction markets, although not all professionals, tend to provide insights that are more valuable than individual voices when aggregated. Just like when I choose to follow someone, I look not only at monthly returns but also at the size of their funds and long-term win rates—market choices made with real money often reveal more truths.

However, there's also a risk point to watch out for: whether the LM Arena leaderboard as a settlement standard is truly objective is questionable. No matter how lively the prediction market is, the final outcome depends on whether the settlement rules have loopholes. These details are equally important when following strategies—no matter how good the strategy, if the risk settings are unreasonable, it's all for nothing.