An event that has a "0 probability" in reality is often overestimated in the prediction market, possibly at 1%, 2%, or even 5%. The low point of probability in the prediction market depends on the expected "annualized" return of the arbitrageur. When the annualized return is too low, savvy arbitrageurs will become the counterparty of less savvy arbitrageurs—buying the Yes on impossible events at a low price. This is also a less obvious arbitrage path in the prediction market.