A systematic crypto fund operating on minute-scale data across spot and derivatives. Twenty-one production strategies, multiple uncorrelated signal families, validated across six years of market history.
系统化数字资产投资机构，立足分钟级行情数据，覆盖现货与永续合约市场。21 项生产策略，多类正交信号体系，历经六年完整市场周期验证。
Cross-venue spreads capture institutional capital flow — a leading signal across major digital assets. 跨交易所价差，捕捉机构资金流向，领先主流资产行情。
Identify structural trend reversals, filter false breakouts, raise hit rate. 识别趋势结构反转，过滤假突破，提升胜率。
Adapt to market state — activate only in suitable regimes, controlling drawdowns. 自适应市场状态，只在有效环境下开仓，控制回撤。
Every signal uses only past data. Walk-forward validation. No look-ahead, no curve fitting beyond the optimization window. 所有信号严格基于历史数据，前向验证；杜绝未来函数与跨窗口曲线拟合。
Commission, slippage, and funding modeled per-trade. Next-bar-open execution. Backtest realism that exceeds typical retail and academic standards. 逐笔建模佣金、滑点与资金费率；下一根 K 线开盘成交。回测真实度显著高于零售与学术常见标准。
Multiple orthogonal signal families across five assets and five timeframes. When one cluster is wrong, others typically aren't. 多类正交信号体系，覆盖五种资产、五个时间周期；单类失效时，其他通常仍然有效。
Our 21-strategy production portfolio. In-sample backtest Sharpe 3.05 (2022–2025), validated out-of-sample on 2020–2021 (Sharpe 2.07) and 2026 YTD live OOS (Sharpe 2.87). 21 项策略组合。样本内回测夏普 3.05（2022–2025）；样本外覆盖 2020–2021（夏普 2.07）与 2026 年迄今真实样本外（夏普 2.87）。
Targets below are anchored on actual 2026 out-of-sample results through May 7 (127 days, true OOS), with conservative haircuts applied. Numbers are presented as target live / in-sample backtest. 以下目标基于 2026 年迄今 127 天真实样本外表现（截至 5 月 7 日），经保守折扣后给出。数据成对呈现：实盘目标 / 样本内回测。
Sharpe retention well above the industry >0.5 threshold for genuine edge in two independent out-of-sample windows places the portfolio in the top decile of systematic strategies. 在两段独立的样本外区间，夏普保持率均显著高于 0.5 的行业真实阿尔法阈值——这一表现跻身系统化策略前十分位。
Backtest-to-live divergence held within 1–2 percentage points. Engineering-grade deployment. 回测与实盘偏差控制在 1–2 个百分点，工程级部署。
Minute-resolution OHLCV across multiple market regimes. Coinbase, Binance, Binance Perp. Funding rate history modeled for directional cost. 横跨多个市场周期的分钟级 OHLCV 数据。来源涵盖 Coinbase、Binance、Binance 永续，包含完整资金费率历史。
Under 2x nominal leverage, actual net risk exposure is only 0.67x — multi-strategy internal hedging substantially reduces portfolio volatility. 名义 2 倍杠杆下，实际净风险敞口仅 0.67 倍，多策略内部对冲，大幅降低波动。
No discretionary overrides. No narrative trades. Every position is a model output. Drawdown limits are pre-committed, not reactive. 不做主观干预，不做叙事交易。每笔仓位皆为模型输出；回撤上限事先约定，而非事后被动响应。
We publish selected research notes — drawdown post-mortems, regime studies, structural-edge writeups — to qualified subscribers. The first issue ships with the live launch. 面向合格订阅者，定期发布精选研究笔记——涵盖回撤复盘、机制研究、结构性阿尔法解析等主题。首期随基金正式上线发布。
Subscribe to notes 订阅研究笔记For institutional allocators, family offices, and qualified investors. Email to schedule due diligence — responses within five business days. 面向机构配置者、家族办公室及合格投资者。请邮件预约尽调，五个工作日内回复。