FAQs How is it different than prediction markets? We use prediction markets as one factor in our forecasts, but prediction markets are not forecasts. They are bets. We apply a scientific method to predict the future using a structured process. We believe our tools are useful for prediction market betters. But very different things, fundamentally. How accurate is it? We'll see. In our testing, surprisingly accurate. But this is a live, public test. So we'll have to see how it does. We are intentionally launching now with a range of 2026-settle events, including elections, as a test. Should I gamble with GameOut forecasts? Probably not. Why does GameOut use multiple models? Models are selected largely based on the goal of trying to get a diverse set of perspectives. Training data varies model-to-model and offers one way to get different perspectives. We also utilize each model in multiple ways. The theory goes: the sum of all AI might be more predictive than any one AI. How many tokens does GameOut burn per prediction? Around 4 million across models. What if GameOut is wrong? It will be honest about it, and we will use incorrect forecasts to adjust and improve our process automatically. How do I suggest a forecast topic? Reply to our threads on X, or when you create an account, you can suggest one in the app itself. What are GameOut's future plans? We believe forecasting is important, and everyone should have access to forecasting tools. We will validate our technology first and refine it. When we are confident it is reliable, we will open it up to the public to run forecasts yourself on any question you have. Will GameOut be free once users can use it? Probably not. Because tokens are expensive, we doubt we will be able to offer it for free. We also worry that if it is free, it will be flooded with unserious requests. We will keep it cheap and affordable and post here once it is available for use. Why is it different than just asking an LLM? We tried that too, and realized that LLMs cannot forecast. They can only guess. Forecasting requires a significant amount of looping and a custom logic. We built our own harness to make this possible. Additionally, a single LLM is biased to its own training data. Multiple models can provide perspectives that one will never be able to.

Is it effective? Maybe. We think so. But, we have to prove it. To do that, we have run a bunch of forecasts which we have published in public. We will run more in the coming days. Our initial forecasts will focus on near-term/intermediate events, primarily (think elections, court decisions, FDA approvals, corporate events like IPOs). Every call goes up before the result is known. Then we'll post what happened, including the ones GameOut gets wrong. gameout.ai/our-forecasts