Skip to content
WickArc

Decision records for quantitative research

Know the odds.
See the evidence.

Turn a market question into an Evidence Passport—a versioned decision record that preserves the claim, assumptions, contradictions, invalidation triggers, and what the evidence actually permits.

Research tools, not trade recommendations.
Estimates require eligible data and evaluation.

A backtest is a claim.

The Passport preserves the argument.

The WickArc primitive

Every idea leaves with an
Evidence Passport.

Backtests are easy to produce. Defensible decisions are hard to preserve. WickArc binds the research question to its evidence, failure conditions, and lineage—without collapsing everything into a vanity score.

Evidence PassportEP-017-V3 · SYNTHETIC
Illustrative record
Research question

Does 20-day momentum predict positive net five-session returns?

Fictional demo universe · Net of declared costs · Not a trading recommendation

Current research stateInvestigateNo eligible setup
Gate 03 / Cost sensitivityWarning

The apparent edge narrows as assumed trading costs rise.

Twenty-four illustrative scenarios vary commissions, slippage, delay, and date windows. Counts alone do not establish robustness.

Evidence attached
24 synthetic stress scenarios
What would invalidate it
Expected net return at or below zero after the declared cost policy invalidates the actionable setup.

WickArc constraintThe decision state cannot outrun its weakest required gate.

LINEAGE ATTACHEDDataset / synthetic-v12Strategy / momentum-v3Experiment / EP-017-V3Not a certification or performance score
01 / State, not scoreInvestigate. Abstain. Eligible.

The state says what the evidence permits and why.

02 / Contradiction retainedFailure evidence stays visible.

Weak folds, cost sensitivity, and missing inputs travel with the result.

03 / Lineage attachedThe argument remains retraceable.

Question, data, strategy, experiment, and evidence versions stay linked.

Inside the research workspace

A result is only the beginning.

Explore the interactive demo →
SYNTHETIC PRODUCT DEMOIllustrative data · No verified trading record
Research intelligence / Example workspace
Read-only demo
Research projects12Versioned hypotheses and strategies
Completed backtests248Historical simulations only
Walk-forward studies36Fold-level results available
Resolved forward trades0No performance claim available

Opportunity research

Research only

Fixed five-session net-positive outcome · Synthetic examples

InstrumentP(net positive)Expected net returnDecision
DEMO-A63.4%+0.18%Watchlist
DEMO-B61.8%+0.12%Watchlist
DEMO-C58.9%−0.03%Abstain
DEMO-D——No estimate

Selected experiment · Evidence

Momentum persistence / Experiment 017 / Version 3

Data & strategy versions
Pinned
Walk-forward evaluation
6 folds completed
Cost sensitivity
24 scenarios tested
Final holdout
Locked · unrevealed
Probability calibration
Review required

Backtest vs. benchmark

Synthetic indexed equity · Not forward performance

Calibration diagnostics

Synthetic temporal validation · Not an accuracy guarantee

Research assistant

Synthetic experiment explanation

Primary concern
Performance weakens as assumed trading costs rise. High-probability estimates need more unseen observations.

The final holdout remains sealed.

DEMO DATA · NOT INVESTMENT ADVICEZero resolved forward trades · No performance claim available

Actual WickArc chart and evidence components with deterministic example data. No customer records. No verified trading performance.

From question to evidence

One workspace.
More ways to challenge an idea.

Stock research, strategy testing, and evidence inspection—without treating a promising chart as proof.

01 /

Analyze a stock

Open a company, examine available price history, research scenarios, and supporting evidence.

Open stock research
02 /

Build an Evidence Passport

Bind the claim, assumptions, evidence gates, contradictions, invalidation rules, and version lineage into one decision record.

Build yours free
03 /

Challenge the backtest

When compatible data and artifacts exist, compare temporal out-of-sample periods and walk-forward folds. Test sensitivity to parameters, costs, and market regimes.

Explore validation
04 /

Define the probability

Keep every estimate attached to its exact event, horizon, evaluation evidence, and eligibility state.

Understand probability
05 /

Preserve contradiction

Keep failure evidence and invalidation triggers beside the supporting result instead of averaging them away.

Understand abstention
06 /

Retrace every version

Carry the exact data snapshot, strategy assumptions, experiment definition, and evidence together.

Follow the lineage

Capabilities depend on configured data and eligible artifacts. Automatic plain-language compilation is not currently enabled; review and experiment setup remain explicit.

The second question matters

Most tools show the backtest.
WickArc investigates whether
you should trust it.

A strong historical result is a hypothesis. Where the data and workflow support them, temporal out-of-sample evaluation, realistic assumptions, and contradictory evidence deserve a place beside it.

  1. 01Backtest
  2. 02Temporal OOS
  3. 03Stress
  4. 04Calibration
  5. 05Evidence
  6. 06Abstain if necessary
Read the research methodology →

How it works

Start with curiosity.
Finish with something inspectable.

  1. 01

    Search a stock or record an idea

    Start with a company or a specific research question.

  2. 02

    Formalize and review the experiment

    Review the proposed data, horizon, strategy, costs, and measurable outcome before creating it.

  3. 03

    Challenge the assumptions

    When supported by the configured workflow, use temporal out-of-sample periods and stress scenarios while keeping the final holdout sealed.

  4. 04

    Inspect the evidence

    Review support, limitations, and whether the result warrants further research.

Built to make uncertainty visible

Less persuasion.
More provenance.

  • Versioned research and assumptions
  • Point-in-time-aware data handling
  • Explicit out-of-sample separation
  • Costs and execution assumptions shown
  • Probabilities tied to defined events
  • Insufficient evidence stays visible

Bring a question. Expect scrutiny.

What would you like to investigate?