Iterative Prediction Markets

Forecasting Tomorrow's Forecasts

Long-term prediction markets, burdened by the extended wait for a distant resolution, become less appealing compared to short-term markets. When both start with the same probabilities, the latter not only promise faster resolutions but also yield more favorable interest rates.

For example, imagine a prediction market asking whether faster-than-light travel will be discovered before 2030. It’s a safe bet that the answer will be no, but how much would you be willing to bet if the market started at 90% - NO, 10% - YES? With almost 6.5 years until resolution at the time of this writing, buying NO at 90% today would mean a meager yield of around 1.5% annually. Probably not the most ideal investment. We can address this challenge by breaking down the long-term prediction into a series of short-term predictions, each building upon the last.

In this iterative approach, instead of placing a direct bet on the 2030 outcome, participants would engage in annual markets, each predicting the sentiment or outcome of the subsequent year's market. Think of it as a ladder of forecasts, where each rung represents a year and its associated prediction.

Let's break it down: In 2024, a market could be created to predict the sentiment of the 2025 market regarding faster-than-light travel. Participants would place bets on whether the 2025 market, 30 days after its opening, would lean more towards YES or NO. The 2025 market, in turn, would predict the sentiment of the 2026 market, and so on, until we reach the final 2030 market.

This approach offers several advantages. Foremost, it effectively tackles the time value of money issue inherent in long-term prediction markets. By breaking down a distant forecast into shorter-term predictions, participants are not deterred by the diminished returns of long-term investments. This, in turn, incentivizes greater participation, leading to more vibrant and active markets. Additionally, the iterative nature of these markets captures the evolving sentiment and knowledge about the subject. As new scientific discoveries emerge or technological advancements occur, these short-term markets can swiftly adjust, more readily reflecting the most current beliefs and information.

While the annual iterative approach offers a significant improvement over traditional long-term markets, there's potential to enhance this even further. Imagine if, instead of yearly predictions, we had markets every 30 days (or even more frequently as needed). This would allow for a more frequent recalibration of sentiments, allowing for a much finer granularity in tracking shifts in opinion and knowledge. Such frequent markets would capture real-time reactions to new information, research breakthroughs, or any other pertinent events.

Manually creating and managing these short-term markets would be cumbersome and inefficient, but this is where software automation can help. Prediction market platforms could automatically derive these short-term markets leading up to the terminal market with each short-term market forecasting the sentiment for the subsequent market.

This automation would not only streamline the process but also ensure consistency in market creation. This approach might more effectively address the time value of money problem and provide better insight into current sentiments.