New Issues Create New Features

[](https://substackcdn.com/image/fetch/$s_!JV8p!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F5458daf9-bd39-4873-b62c-24c9881758a6_2050x1155.png) ### New day, new issue to solve. Today, nothing qualified for a Diamond Pick or a Power Play. Instead of forcing picks that the model isn’t confident in, I added a new section on the Consensus page called **Best Combos of the Day**. [](https://substackcdn.com/image/fetch/$s_!EA4J!,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Ffb25c690-857a-4ef9-81ab-ea385e32a979_1944x562.png)This box surfaces the strongest **model combinations based on historical win rate**, not raw consensus count. That means even if the overall consensus leans toward one team, a smaller combo of models with a better long term win rate can still be highlighted. The goal here is to make decision making clearer on days where the algorithm can’t confidently surface featured picks. Rather than “no picks,” we now still get actionable signal rooted in historical performance. For example, in **MTL vs TB**, the 2 model combo of **Dunkel + MoneyPuck** has a higher win rate than the 3 model consensus play. That combo is shown instead. I’ve enabled full tracking so we can directly compare how these combo based picks perform versus traditional consensus over time. These Best Combos will also be included in the daily emails so nothing gets lost on quieter or more fragmented slate days. This should give us better visibility, better context, and fewer forced decisions when the data isn’t aligned.