the project
no priors is a Bayesian-minded trading project. The bot doesn't start the day with an opinion about which stock will rip. It starts flat, reads the tape, and lets the posterior sharpen where the evidence is actually concentrated. Then it writes five trade cards and posts the ones that cleared the threshold.
Everything the bot does is versioned. Every factor has a weight that was earned through regression against forward returns, not intuition. Every strategy has a YAML config that must pass a backtest before it gets promoted to live. Every trade is logged to a ledger that you can read.
the person
I'm a quant-adjacent software engineer. I built this because I got tired of blog posts about markets that never showed their numbers, and of products that showed backtested hypotheticals as if they were live results. The bet is that there's a reader audience for the thing I wanted to read.
I'm not a registered investment advisor. Nothing here is advice. If you use anything I publish to trade real money, you are doing so on your own judgment and at your own risk.
the posture
Three rules keep this project honest:
- No forward-looking claims. Results are what they are; the ledger is the scoreboard.
- No model gets promoted without a backtest. If the backtest can't be reproduced, the model doesn't ship.
- Every loss gets a postmortem. If the system was wrong, the post explains why it was wrong — not why I still think it was right.
contact
Reach out at hello@nopriors.xyz. I read everything, reply to most of it.