vs coin flip: Brier minus 0.250 (−0.005 ± 0.027, 95% margin) — not distinguishable from 50/50. vs Market: gap to Market’s last pre-start price on the same games; positive means worse.
Cumulative average as games resolved, oldest first. Lower is better; the dashed grey line is the Market on the same games.
When GPT said a number, how often did it happen? Dots on the dotted diagonal are perfectly calibrated; whiskers are 95% intervals, hollow dots have fewer than 5 forecasts. On average it is off by 11 percentage points. Compare with everyone →
Forecast is the probability of YES as the question was worded. A tick means the forecast was on the right side of 50%.
“Alcaraz is one of the very best players in the world and typically a heavy favorite over Yibing Wu on hard courts, especially in a Slam setting. Wu has talent and can threaten with first-strike tennis, but his results and durability have generally been much less reliable. Without current-day form/injury updates, a baseline estimate strongly favors Alcaraz. I assign Wu roughly an 8% upset chance, with a small implicit allowance for variance or a compromised Alcaraz performance.”
“Arsenal are the clearly stronger side and have started 3-0-0, while Sunderland are 1-1-1. The strongest evidence is the sportsbook consensus implying about a 14.0% Sunderland home-win chance. Home advantage gives Sunderland some upset potential, and market odds include margin so I shade very slightly upward from the raw consensus, but not much. Without reliable lineup/injury information, the best estimate is that Sunderland win roughly 1 time in 6-7.”
“Manchester City are at home, have started 2-0-0, and are typically a much stronger side than Coventry. Coventry are 0-0-2, suggesting an early-season quality gap consistent with City being heavy favorites. The strongest concrete evidence is the sportsbook consensus around 85.9% for a home win; absent any specific injury or lineup information, I anchor close to that while allowing a little uncertainty for early-season variance and the possibility of a draw.”
“Main evidence is the sportsbook consensus, implying about an 82% Chelsea home win chance. That fits the general prior that Chelsea are substantially stronger than Hull, especially at Stamford Bridge. The small sample records (Chelsea 2-0-1, Hull 2-1-0) add little and do not outweigh the class/home advantage signal. I trim slightly below the market-implied number to allow for bookmaker margin uncertainty and unknowns like early-season variance, but Chelsea remain a clear favorite.”
“Sportsbook consensus implies Shelton around a 19.7% chance, and that is strong base-rate evidence. Independently, Alcaraz is generally the superior player on hard courts and especially dangerous in high-leverage matches, while Shelton’s big serve and lefty power give him upset potential but still leave him a clear underdog. With no reliable additional information on injuries or unusual conditions, I stay close to the market and give Shelton a 20% chance.”
“Atlanta is 81-55 at home versus San Francisco 56-81 on the road, a very large team-strength gap. Home-field advantage further boosts the Braves. The listed starters also lean Atlanta: Bryce Elder is at least an established MLB starter, while Yunior Marte is more uncertain as a starter-quality option. With no market odds available and no reliable lineup/injury info, I anchor mainly on team records, venue, and probable starters. I assign San Francisco roughly a 23% chance to win, with a small residual chance of postponement/cancellation effectively folded into the estimate.”
Average cost per forecast on these games: $0.0034 (122 forecasts).