Autonomous equity trading · self-learning, self-explaining.

Like a 401k but x100.

Priorsum scans the market, decides on its own, manages every position, and gets a little better after each outcome, then tells you exactly what it did and why, in plain language.

Open the dashboard → See how it works

You supervise a system that runs itself, trading a paper account, learning in the open.

The actual dashboard

This is what you watch while it decides.

The Now cockpit and the live decision stream, side by side, every call, the confidence behind it, and the plain-English reason, updating as the engine works.

Every decision, logged and explained, exactly as it happens.

What it is

One decision loop, running the whole day, on its own.

Configure a strategy once. From the opening bell, Priorsum handles the rest, no dashboards to babysit, no signals to place by hand.

01 · autonomous

Runs itself

Market scanning, signal generation, order submission, position management, and exits, all handled end-to-end. You set the intent; it does the work.

02 · adaptive

Learns from outcomes

Every closed trade feeds back into the system. Signals that helped get more weight; strategies that stop working get picked less. It tunes toward what's actually working now.

03 · transparent

Explains every move

A live feed of every decision as it happens, and a plain-English reason for each one, including why it didn't trade. Nothing is a black box.

How it learns

Two feedback loops, running in parallel.

The system improves on two clocks at once, a fast reflex after every trade, and a slower, more deliberate shift once there's enough evidence to trust it.

● fast loop

After each closed trade

The moment a trade closes, the signals that contributed to the win, or the loss, are nudged up or down. Small, immediate, self-correcting.

trade closes→ score signals→ nudge weights
● slow loop

After a meaningful batch

Once a statistically significant run of trades is in, strategy parameters mutate toward the configurations that actually performed better, never on a single lucky trade.

enough trades→ evaluate→ mutate params
You can follow along

Built to be watched, not just trusted.

You're a supervisor and a spectator of a system that runs on its own. So the dashboard's whole job is to let you follow the story and form appropriate trust.

  • ✓

    Every decision, as it happens

    A rolling feed of what it looked at, what it decided, and the confidence behind it, in real time.

  • ✓

    Every adaptation, explained

    When a weight moves or a parameter mutates, it says so in plain language, and why.

  • ✓

    Told when it matters

    Calm by default. It gets loud only for things actually worth your attention.

Correctness first

Modeled like it's real money, because someday it might be.

Priorsum trades an in-house paper account today, funded with simulated cash. But every order path models real-market mechanics end-to-end, and the safety rails are hard limits the learning loops can never touch.

Paper, funded, and honest

Each account starts with simulated cash and fills against live market data, real mechanics, no real capital at risk.

Hard risk floors

Position size, total exposure, daily loss limits, and stop-losses are enforced at decision time. Nothing the system learns can widen them.

Learning stays in its lane

The feedback loops tune signals and strategy, they can never change risk limits, place an unreviewed order, or reach the stop ladder.

Safe by construction

Protective exits run independently of the decision engine, so the safety net keeps working even when everything else is paused.

Common questions

The honest answers.

No hedging. Here's what Priorsum is, what it isn't, and where the lines are drawn.

Is Priorsum trading real money? +

No. Priorsum is an in-house paper-trading research system. Each account starts with $100,000 in simulated cash and fills in-memory against live market data. No real capital is ever at risk, and no orders are sent to a broker.

What does it actually do on its own? +

Everything in the loop. From the opening bell it scans the market, generates signals, decides, sizes the position, submits the order, manages the trade, and exits, without you placing anything by hand.

How does it get better over time? +

Two learning loops running in parallel. A fast loop re-weights the signals behind a trade the moment it closes; a slow loop mutates strategy parameters after a statistically meaningful batch of trades, never on a single lucky one.

Can the AI remove or widen the safety limits? +

No. Position size, total exposure, the daily loss limit, and stop-losses are hard floors enforced at decision time. The learning loops can tune signals and strategy, but can never touch those limits, place an unreviewed order, or reach the stop ladder.

Is it profitable? +

It's a paper research system, not a return you can withdraw. Over one clean week (June 25 to July 2, 2026) it finished +$281.96 on 107 trades at a 47.7% win rate, shown honestly on the proof page, losing days included. That's one window of one account, and not a promise about any future period.

How is it different from a screener or a signal service? +

A screener or alert tool tells you what's happening and leaves the trading to you. Priorsum decides, trades, manages the position, and learns from the outcome on its own, then explains each move afterward in plain language.

Watch & enjoy

Come watch it trade.

Sign in with Google to open your dashboard and follow the system decision by decision.

Open the dashboard → Read the blog