VERIFIABLE QUANT RESEARCH

Turn investment ideas into
verified strategies.

从投资想法到可验证、可复现的量化策略。Vine 不只生成答案，还展示结论为什么值得相信。

01Describe your idea用自然语言定义目标

→

02Test and validate确定性计算与稳健性检验

→

03Make an informed decision查看证据、风险与失效条件

ChatGPT / ordinary copilots“Here is a strategy that sounds reasonable.”

Vine QuantPilot“Here is the evidence, uncertainty and audit trail.”

01 · PORTFOLIO X-RAY

Your portfolio, understood before optimized.

预置模拟账户 · 资产权重、风险贡献与跨资产相关性

Total assets$1,000,000Demo portfolio

Annualized return11.2%Historical estimate

Annualized volatility17.8%Above target range

Maximum drawdown−31.4%Exceeds 20% objective

Current allocation

Market value weights

4 assets

100%invested

SPY$340k34%

GLD$350k35%

BTC$220k22%

SCHD$90k9%

Risk contribution

Contribution to total portfolio variance

Concentrated

BTC61%

SPY24%

GLD11%

SCHD4%

22% BTC weight drives 61% of estimated risk.
Capital allocation looks diversified; risk allocation is not.

Cross-asset correlation

Monthly returns · full illustrative sample

Methodology ↗

SPYGLDBTCSCHD

SPY1.000.080.340.91

GLD0.081.000.120.06

BTC0.340.121.000.31

SCHD0.910.060.311.00

02 · COPILOT

Describe the outcome. Confirm the research contract.

LLM 负责理解意图；在计算前把目标转成明确、可审计的 Strategy DSL。

QuantPilotResearch copilot · not an investment adviser

● ONLINE

Tell me what you want to improve and which risks you cannot accept. I will structure the request before running any calculation.

我希望保持长期增长，但把历史最大回撤控制在 20% 以内。可以调整当前资产权重，也可以加入 SGOV。

✦ STRUCTURED BY COPILOT

Research specification

Ready to confirm

ObjectiveImprove risk-adjusted return

ConstraintHistorical max drawdown ≤ 20%

UniverseSPY · GLD · BTC · SCHD · SGOV

RebalanceQuarterly

Transaction cost5 bps per trade

Evaluation2015–2026 · Walk-forward

Strategy DSL v1.0

universe: [SPY, GLD, BTC, SCHD, SGOV]
objective: maximize_risk_adjusted_return
constraints:
  max_historical_drawdown: 0.20
  max_asset_weight: 0.50
rebalance: quarterly
transaction_cost_bps: 5
validation: walk_forward

Why confirm first? This freezes the research question so results can be reproduced and audited.

03 · DETERMINISTIC RESEARCH

See the research, not a loading spinner.

每个数值步骤由确定性工具执行。此 Demo 的市场结果为预计算示例，工作流状态为交互演示。

PRECOMPUTED DEMO RESULTS

Research readyConfirm the specification above to start

Market DataLoad adjusted price histories

WAITING

Candidate GeneratorBuild 428 constrained portfolios

WAITING

Walk-forward EngineSeparate selection from evaluation

WAITING

Cost & Parameter Tests5–20 bps and weight sensitivity

WAITING

Stress & ValidatorScenarios, leakage and fragility checks

WAITING

Research ID will be assigned after confirmation.

Growth of $100

Illustrative historical comparison · 2015–2026

CurrentCandidate

Candidate allocation

Lower concentration, added liquidity sleeve

Max DD met

SPY30%

GLD28%

BTC10%

SCHD12%

SGOV20%

The candidate sacrifices some historical return to reduce drawdown and volatility. It is not presented as a return-maximizing portfolio.

What changed?

Current portfolio vs. selected candidate

Illustrative · not actual performance

MetricCurrentCandidateInterpretation

Annualized return11.2%10.6%−0.6 pp

Annualized volatility17.8%12.9%−4.9 pp

Maximum drawdown−31.4%−18.7%Within objective

Sharpe ratio0.630.82+0.19

Worst calendar year−22.1%−12.8%+9.3 pp

04 · QUANT VALIDATOR

A score is not enough. Show the evidence.

可信度不是未来收益预测；它总结研究过程的质量、稳健性证据与已知风险。

STRATEGY CONFIDENCE SCORE

78/ 100

Evidence is reasonably strong

4 checks passed · 2 require attention

WeakModerateStrong

Not a prediction of future returns

✓

Data leakage

No future-available fields detected. Signals use lagged, point-in-time inputs.

PASS

✓

Out-of-sample behavior

Risk-adjusted performance remained positive across held-out windows.

PASS

Parameter sensitivity

Results weaken when BTC weight rises above 15%; allocation is not fully stable.

MEDIUM

✓

Transaction costs

Conclusion remains directionally stable from 5 to 20 bps per trade.

PASS

↯

Market-regime dependence

Candidate weakens during rapid inflation shocks and simultaneous equity/bond losses.

RISK

Overfitting risk

Low parameter count helps, but one selected sample cannot eliminate selection bias.

LOW–MED

Reproducibility record

Everything needed to rerun this research

Research IDQP-260714-A17F

Data snapshotVMD-2026.06.30

Strategy specDSL v1.0 · SHA 8c2e

Research engineVine Quant Core 0.8.2

Validation policyQV-CROSS-ASSET-04

Generated2026-07-14 12:08 UTC

05 · STRESS LAB

Ask “what breaks?” before “what earns?”

选择历史/假设情景，或拖动资产冲击。损失估计为一阶权重近似，仅用于产品演示。

Scenario library

Select a preset shock

2020 market shock

Adjust shocks to test a custom scenario

LINEAR ESTIMATE

Estimated candidate loss−19.8%Current portfolio estimate: −26.3%

Current−26.3%

Candidate−19.8%

This simplified first-order model excludes correlation breaks, nonlinear option payoffs, slippage and market closure risk.

06 · INVESTMENT MEMO

Decision-ready, with conditions attached.

自动整合研究目标、结果、风险、适用条件和下一步；任何交易动作停留在人工确认草稿。

VINE QUANTPILOT · RESEARCH MEMORANDUM

Cross-Asset Drawdown Control

Research ID QP-260714-A17F · Generated 14 Jul 2026

78Confidence

Decision summary

The candidate portfolio historically reduced maximum drawdown from −31.4% to −18.7%, while lowering annualized return by 0.6 percentage points. The drawdown objective was met in the illustrative sample, but inflation-shock dependence and BTC-weight sensitivity remain material risks.

Objective

Maintain long-term growth while keeping historical maximum drawdown at or below 20%. Existing assets may be reweighted and SGOV may be added.

Current portfolio issue

BTC represents 22% of capital but approximately 61% of estimated variance. SPY and SCHD also share substantial equity-factor exposure.

Candidate allocation

30% SPY · 28% GLD · 10% BTC · 12% SCHD · 20% SGOV. Quarterly rebalance with 5 bps assumed cost.

Key evidence

Walk-forward evaluation, transaction-cost range of 5–20 bps, parameter perturbation, leakage checks and five stress scenarios.

Risks & failure conditions

Fast inflation shock causes cross-asset diversification to weaken.
Result is sensitive to BTC weights above 15%.
Future correlations and liquidity may differ from the sample.

Next step

Review tax and execution constraints, validate with the user's actual account data, then paper-track the candidate before any rebalance decision.

Illustrative demo results only. Not actual performance, a recommendation, or a forecast. The user remains responsible for any investment decision.

Turn investment ideas intoverified strategies.

Your portfolio, understood before optimized.

Current allocation

Risk contribution

Cross-asset correlation

Describe the outcome. Confirm the research contract.

Research specification

See the research, not a loading spinner.

Growth of $100

Candidate allocation

What changed?

A score is not enough. Show the evidence.

Evidence is reasonably strong

Reproducibility record

Ask “what breaks?” before “what earns?”

Scenario library

2020 market shock

Decision-ready, with conditions attached.

Cross-Asset Drawdown Control

Objective

Current portfolio issue

Candidate allocation

Key evidence

Risks & failure conditions

Next step

Follow the evidence,not the hype.

Join the research list

Turn investment ideas into
verified strategies.

Follow the evidence,
not the hype.