Explainability Product July 11, 2026 · 5 min read

Ask the dashboard “why?” and get a straight answer

A black box that happens to make money is still a black box, and the morning it stops making money, you’ll have no idea why. We built priorsum from the opposite starting point: every decision should be able to explain itself, and you should be able to poke at it until you’re satisfied.

“It bought AAPL.” Okay, but why?

Most tools respond to that with a shrug and a chart. The trade happened, here’s the P&L, have a nice day. Which is fine until the thing is running an account without you watching, and “it just did” suddenly isn’t an answer you can live with.

So priorsum keeps a receipt for every decision. Not a log line, an actual record of what was going on when it acted: which strategy was in charge, the settings it was using, and what the market looked like at that second. After a trade closes you can trace it back from the result to the reasoning that got you there. Roughly like this:

AAPL BUY 10:14:07 ET

strategythe one the selector judged best fit for this symbol, this regime, this time of day

conditionstrend and volatility read as favorable; the setup matched what this strategy is good at

confidencesignals combined into one score, then sanity-checked by a win-probability model

risk gatepassed — sizing, exposure, and daily-loss checks all cleared before the order went out

outcomeclosed later that session; the result is stitched back to this exact decision

? circle any part of the dashboard and ask what it means, in plain language

Illustrative example. Field names simplified; the real trail carries more detail.

Plain language, not a data dump

A wall of numbers isn’t transparency. It’s just a black box with better production values. So priorsum writes things down in words. Every adjustment the learning makes gets spelled out in a sentence you can read without a decoder ring. And if something on the screen has you stumped, you can literally circle it and ask, and get that part of the dashboard explained right there. Being curious shouldn’t mean reading our source code.

The bar we hold ourselves to: for any trade, any adjustment, any moment, you should be able to answer “why did it do that?” without taking our word for anything. When a decision can’t explain itself, we treat that as a bug.

Why traceability is a safety feature, not a nicety

Tying every outcome back to the decision that caused it pays off twice. You end up trusting the system, because none of what it does is hidden from you. And the system ends up able to trust itself, because that same clean trail is what lets it learn honestly, handing credit to the trades that actually earned it instead of the ones that just happened to be standing nearby. The explainability isn’t a coat of paint over the intelligence. It’s what the intelligence stands on.

See a decision explain itself

Follow any trade from outcome back to the reasoning that produced it, in language you don’t need a PhD to read.

Priorsum is an in-house paper-trading research system that models real-money mechanics end-to-end; nothing here is financial advice. The decision trace shown is an illustration of the concept, not a live screenshot.