Tokyo 21
● System Live
Protocol 01: Integrity

The Verification Pipeline for Institutional Grade Alpha

At Jade Quant Systems, a strategy is not "live" until it has survived an exhaustive battery of stress tests, bias checks, and execution simulations. We do not chase backtest beauty; we build for market reality.

Foundational Backtesting Integrity

The majority of algorithmic failures stem from data snooping and look-ahead bias. Our primary objective is to eliminate these variables before a single dollar is deployed.

"A backtest is a historical conjecture; a live system is a physical confrontation with liquidity." — Engineering Lead, Tokyo 21

In-Sample vs. OOS

We strictly enforce a 70/30 split between in-sample training and out-of-sample (OOS) testing. Strategies that lack consistency across unseen data are immediately discarded regardless of potential returns.

Walk-Forward Analysis

Static optimizations are prone to overfitting. We utilize rolling walk-forward verification to ensure the quant systems adapt to changing market regimes rather than just hunting historical ghosts.

Significance Testing

Using Monte Carlo simulations, we run 10,000+ permutations of trade sequences. If the strategy’s success can be explained by random luck (p-value > 0.05), it is flagged for rejection.

Slippage Modeling

All trading analytics are run against variable spread and realistic slippage models. We assume conservative fill rates to protect against the "idealized world" fallacy.

Quantitative infrastructure at Jade Quant Tokyo office

Infrastructure Strength

Low-latency arrays and proprietary signal isolation.

Deployment Gateways

Every strategy must clear these four checkpoints sequentially. Failure at any stage restarts the logic development process.

G-01

Semantic Audit

A manual review of signal logic to ensure the underlying market thesis is economically sound and not a result of mathematical over-optimization.

Requirement: Thesis Coherence
G-02

Fat-Tail Stress Test

Injecting synthetic black-swan events and correlation shocks to observe strategy behavior during liquidity crises and regime shifts.

Requirement: MDD Threshold < 15%
G-03

Paper Correlation

Execution in a simulated "Forward-Live" environment using real-time feeds to verify that broker-side latency matches the backtest assumptions.

Duration: 4-Week Incubation
G-04

Phased Deployment

Gradual capital scaling starting at 10% of target size. Monitoring for Alpha decay or execution drift before reaching full capacity.

Combating Survivorship Bias

In the field of trading analytics, what you don't see is more important than what you do. We maintain a "Graveyard of Systems" database—a collection of every failed hypothesis, overfitted model, and rejected strategy. This data pool allows us to cross-reference new strategies against known failure modes, ensuring we aren't unknowingly repeating the same errors.

Our data sourcing involves high-resolution tick data, cleaned and verified through multiple providers (IEX, LSEG, and proprietary aggregators). We account for dividends, splits, and delistings to ensure our quant systems are built on a ground-truth representation of historical volatility.

Backtest Cleanliness

We provide full transparency on data cleaning scripts and handling of outliers, ensuring the statistical mean isn't skewed by non-replicable anomalies.

Execution Verification

Pre-trade impact models are compared against post-trade analysis (TCA) every 24 hours to calibrate our slippage assumptions.

Ultimately, verification at Jade Quant Systems is a culture of skepticism. We approach every innovative finding with the intent to disprove it. If a strategy survives our adversarial validation process, it possesses the robustness required for institutional deployment.

Explore our rigorous methodology in action.

HQ: Tokyo 21, JP
Verification Standards Updated: March 2026