Which answer formats do forecasting bots silently drop? A cross-tool benchmark

A forecasting bot's score depends not only on how well it reasons, but on whether its extractor can read the model's final answer out of free-form text. If the extractor returns nothing, that run is dropped from the ensemble; if it reads a number that was never a forecast -- a base rate, a figur...

Source: [Dev.to](https://dev.to/munimv2/which-answer-formats-do-forecasting-bots-silently-drop-a-cross-tool-benchmark-5gjb)