This market will resolve to which prediction is the most accurate based on the formula outlined below. For the purposes of this market, the winner of the Senate will be the party that AP declares will hold a majority in the Senate as a result of the 2026 midterms (this includes the tie-breaking vote of the vice president and independent senators who are known to caucus with one party). If there is a tie, the market that had the most accurate prediction at 8 PM on the day of the midterms will win.
Equation:
The accuracy of a market's forecasts shall be measured using a Linear Decreasing Time-Weighted Brier Score. The purpose of this scoring rule is to reward markets that make accurate forecasts earlier in the life of the market, rather than only shortly before the outcome becomes known.
Markets will be measured once per calendar day.
Each day is assigned a weight that decreases linearly over time:
August 1 receives weight 95,
August 2 receives weight 94,
...
November 2 receives weight 2,
November 3 receives weight 1.
Therefore,
A score of 0 represents a perfect forecasting record.
Lower scores indicate greater forecasting accuracy.
Earlier forecasts have greater influence on the final score than later forecasts because the daily weights decrease linearly over time.
Should there be a dispute, the rules provided will control over the answer to the question itself. While, the question is intended to provide an easy way for people to bet on their beliefs, it is not the end all be all, and all traders should look at the rules. Please don't hesitate to ask clarifying questions in the comments. In the unlikely event of significant ambiguity regarding whether the resolution criteria have been met. I reserve the right to resolve a question to a percentage that I deem fair. I will give notice before I do this.
1,000@jack yeah this is my first time using this equation so I thought I’d use a smaller number of markets. If I do this again, I’d add more markets and polling.
@AaronSimansky
I think Kalshi is pretty straight forward. But what would be a snapshot day from Poly like this with 102? Normalizing to 100 or taking the 35 Republicans (as this is the question in our Manifold market)?
@Tantalost For each day, the value at 2 am ET is entered. The likelihood of republican control of the Senate is the value that is measured.
@AaronSimansky
Thanks! To help with the calculation, where exactly is the daily "likelihood of republican control" value pulled from for each platform? Is it the raw contract price from the frontend, or buy/sell/last traded, or is it sourced via an API/normalized data point? I think the final score will be extremely close, so every single percentage point and platform design choice matters.
@Tantalost @AaronSimansky
The reason this specific data source choice is so crucial is that it completely changes the outcome in a Democratic victory scenario.
My initial visual intuition, which turned out to be wrong (maybe) , was that Polymarket would easily win on pure forecasting quality because its Democratic contract was consistently a step ahead in August and early September. Because those early days carry massive weight, that should have been uncatchable.
But, it you are measuring strictly by the raw, un-normalized Republican contract, the math might completely flip. Polymarket frequently totals 99-102% due to order book friction and spreads (like the 67% Dem / 35% Rep split in the screenshot).
By taking that raw 35% instead of a normalized baseline, Polymarket gets hit with artificial error points that have nothing to do with its prediction quality. This technical friction acts as a massive statistical drag, completely neutralizing Polymarket's early advantage and structurally handing 1st place to Manifold - simply because Manifold’s AMM framework forces a clean 100% baseline every day (Republicans Yes/No).
In a close race down to the exact decimals, the scoring rule ends up grading platform liquidity mechanics rather than pure accuracy. That is why your exact data source choice decides the whole market
(Of course, if the Republicans win the Senate, all of this is irrelevant anyway. Kalshi spent the heavily weighted month of August firmly biased toward the Republicans and bias continues. But now that we’ve completely mapped out the entire matrix of this market, I’ll leave the rest to the traders)