Research prototype · August 2026
投资之前，先核实。
Before money leaves your hands, make sure the evidence is strong enough. Verity gives you a structured second look—organizing what is known, exposing what is missing, and showing when to pause.
The problem is not a lack of advice
A bank recommends a fund. A friend recommends gold. A creator recommends a stock. An AI assistant suggests an investment.
Before you send money:
What has actually been verified?
Knowing that investment risk exists is not the same as knowing when not to act.
Start with what happened
Check before money moves
Paste a recommendation, add a link, or attach a PDF or screenshot. The prototype classifies evidence and applies a structured readiness gate; it does not provide personal financial advice.
Illustrative example
Four important items remain unverified.
Who legally receives and controls the investor’s money?
Where is the gold held and who independently verifies ownership?
How and when can the investor recover principal?
What economic activity produces the claimed return?
What Verity checks
Verity is an evidence-verification and investment decision-checking layer—not a stock picker, trading platform, robo-adviser or guarantee against loss.
Who receives the money?
Who is legally responsible?
Where are the assets held?
Where does the promised return come from?
How can you get your money back?
What do you pay?
Who earns money if this is sold?
What happens if it performs badly?
Does this create excessive exposure?
What is not disclosed?
Which claims can be independently supported?
How Verity thinks
What happens if this is wrong?
Confidence is not verified evidence.
Trust falls as uncertainty and inconsistency rise.
We do not have enough evidence yet.
Move high-stakes uncertainty to family or qualified professionals.
Verity core engine
PDF · screenshot · fund document · message · URL
Who · what · returns · fees · custody · exit
Verified · unverified · missing · contradictory
Downside · liquidity · sensitivity · conflicts
READY · CHECK MORE · STOP
Ask questions · family review · professional review
Why use Verity if I already use AI?
General AI tools are excellent at answering broad questions. Verity is built around a specific sequence:
Verity is not trying to provide another prediction. It asks: Do I have enough verified information to act?
More advice is not always better.
Stop. Verify this first.
Research engine
Verity is the consumer application of one continuous research progression—not a separate chatbot project.
Can a quantitative model run?
Can the model’s result be trusted?
When do volatility, correlation or instability make it less reliable?
Can the system reduce, abstain or stop?
Can these ideas help ordinary people pause when evidence is insufficient?
Academic pillars
Does a small change in assumptions dramatically change the conclusion?
Is this meaningful evidence or weak, noisy information?
Can the system explain exactly what it found and why?
Has the situation moved outside the conditions where the conclusion was valid?
Where the research began
Research moved from whether a model could produce useful signals to a harder question: When should the model stop trusting its own output?
Work includes look-ahead bias, point-in-time data, transaction costs, sensitivity, regime change, correlation shifts, drawdown, model failure, abstention and human escalation.
Research mentorship and practical thinking in derivatives, risk pricing, position control, shutdown logic and software systems inform the independently tested principles inside Verity. No proprietary formula or performance claim is asserted.
What we changed after being wrong
Reliability comes from documenting mistakes, revising the system and making the next test stricter.
Several candidates had histories too short for a valid comparison.
DECISIONRemoved them from the main 3.1 research universe.
A larger sample is not automatically a better sample.A volatility-scaling implementation risked using future information.
DECISIONChanged the calculation to use only information available through t−1.
A strong result is meaningless if the test leaks future information.The project originally emphasized reliability in quantitative systems.
DECISIONApplied the same decision-safety question to consumer investments.
Verity extends the research; it does not abandon it.Built to learn
Verity is designed to improve through documented failure, testing and revision.
About the researcher
Karina is a high school student exploring mathematics, statistics, computer science, AI decision systems and quantitative finance. Her interest in finance began in elementary school and later expanded into programming and quantitative research.
Her research gradually shifted from asking whether a model could produce an answer to asking whether the available evidence was strong enough to justify action. Verity is the consumer-facing application of that continuing research.
Research beta · August 2026
We are building an early research prototype for people who want a clearer second look before making an investment decision.