jev-alpha-bench

What it does

The results suggest that Jev reads the articles competently but yields no tradeable signal.

How you can use it

For a project that scores financial news, borrow this study's timing checks. A developer can rerun the saved results and compare what was known before each possible trade. Try removing company names and shuffling results too: these checks help reveal whether an AI is reading the news or benefiting from misleading test conditions.

Do not use these results to justify real trades. The tested strategies lost money after assumed trading costs. The study does not settle whether trading within minutes could work, because its news data lacks exact times. It tests one AI version and cannot rule out every small effect.