Would Jev work for filtering evidence in a prediction market bot?

I have a side project that trades on prediction markets. It pulls in news, social posts and on-chain data, then uses an LLM to estimate the odds of each outcome.

The weak spot is the middle step. Most of what comes in is noise, and asking an LLM to read everything and write back a probability is slow and hard to trust.

Here's how I'd try Jev:

1. A Noul per item: "does this actually change the odds of this market?" Anything below a threshold gets dropped.

2. A Score for how strong the evidence is, from rumor to official source.

3. The code combines the scores with the current market price and decides whether there's an edge. Jev never places a bet.

Has anyone used Jev on a pipeline like this? I'm also curious if the probabilities hold up on fast-moving news, or if I should send the close calls to a bigger model.