Under the hood Learning July 11, 2026 · 5 min read

Two loops, no hands: how priorsum gets better while it trades

A lot of trading tools really only have one gear. They run whatever rule you gave them until you go back in and change it yourself. Priorsum isn’t built that way. It’s learning in two different ways at the same time, all day, and you don’t have to babysit either of them. Here’s what’s actually going on.

Why one loop backs you into a corner

Any single feedback loop makes you pick your poison. Tune it to react fast and it starts chasing noise, so one bad trade sends it into a spiral. Slow it down and it’s cheerfully trading last month’s market a week after that market packed up and left. We didn’t want to choose, so we didn’t. Priorsum runs both, and the two keep each other honest.

Loop 1

The fast loop

Fires after every closed trade

The instant a position closes, priorsum looks back at which signals actually earned the result and adjusts how much it trusts each one. Small corrections, made constantly. The signals that keep being right slowly earn a louder vote.

Loop 2

The slow loop

Fires after a meaningful batch

Every so often, once there are enough trades to mean something, it steps back and asks a harder question: would a different setup have done better? If the numbers make the case, it drifts that way. If they don’t, it leaves things alone.

The fast loop keeps priorsum quick on its feet. The slow loop keeps the fast one from reading too much into a good afternoon. Reflex versus second opinion, basically, and the second opinion takes some convincing, a hot streak on its own won’t do it.

The part we’re going to be cagey about

How exactly a signal earns or loses trust, when a batch of trades finally counts as “enough,” what makes a new setup convincing enough to actually adopt, those are the details we keep to ourselves. The part worth knowing is the mindset behind them: both loops are built to distrust themselves. Anything new has to show it’s a real edge and not a lucky coincidence before it gets to change how the account trades.

One rule sits over both loops, and neither gets to touch it: the risk limits don’t move. Position size, total exposure, the daily-loss cap, the stops, all fixed. Nothing the system learns, on either clock, can loosen them. It’s free to get smarter about what it trades. It never gets a say in how much.

What you actually see

Best part: none of this happens behind a curtain. Every change the system makes turns up in the dashboard as a plain sentence, which signal got promoted, which setting shifted, and what made it worth doing. You don’t have to take it on faith that priorsum is learning. You can sit there and watch it.

Watch both loops turn

See priorsum decide, trade, and adapt in real time, with every adjustment explained as it happens.

Priorsum is an in-house paper-trading research system that models real-money mechanics end-to-end; nothing here is financial advice. Learning behavior is bounded by hard, non-negotiable risk limits by design.