🎁 福利专区全网大模型免费应用 + 新用户福利 + 注册活动入口,低成本玩转 AI

量化工具箱 pengpengyi92/dsh-quant

DeepSeek Harnessv0.90.0
pengpengyi928 天前有提交已提供中文文档扫描:含敏感能力

🐳 Dsh-Quant:全能插件,AI 原生量化操作系统

53.2/100
综合分
57
GitHub 星标
363
npm 周下载
—/5
用户评分

数据截至 2026/10/27 · 周下载取自 npm 全网(含 CI / 镜像),非本站安装数;综合分由星标、活跃度与兼容性检查合成。

安装命令 · 需先装 dsh CLI 引擎
dsh plugin --profile web add dsh-quant
npm 源 · dsh-quant 的发布者账号已校验与本仓库一致

状态速览

判定依据:实装验证 · 静态安装检查 · 自动扫描 · 仓库维护状态
信任档位:仅索引
是什么
dsh 原生插件 · tool
装得上吗
静态安装检查通过
未做真实安装
L4 · 真实安装等级说明 →
安全吗
本站尚未对该插件做风险分级
暂未覆盖,不等同于无风险
还在维护吗
活跃:最近一次提交在 8 天前
档位由哪些信号合成

本站实装验证(真实安装 L4 覆盖全站;L5 为运行时冒烟或实机截图,覆盖范围小于 L4) · 验证所用 dsh 版本 · 静态安装检查 · 风险分级 · 仓库维护状态。 下方各区块是它的证据明细。

验证判据与等级说明 →
社区实装还没有人打卡 · 装过的第一个就是你
安装兼容性检查✓ 自动检查通过
展开查看检查明细 ↓

以下结论由程序自动检查 npm 包、engines 声明与入口文件得出,未做人工实机验证——能装不等于用着没问题。

✓npm 包dsh-quant @ 0.90.0
✓Node 引擎要求 >=20.0.0 · 基线 Node 22.19 满足
✓dsh CLI 依赖未声明 dsh 版本约束
✓入口文件main/exports/bin 已声明

验证方式:npm registry 存在性 + package.json 静态校验 · 最后检查 2026/10/30 09:04:14

安全扫描自动 · 静态分析源码
v2.7.0
🔵 扫描:含敏感能力尚无人工评估
展开查看扫描说明与证据 ↓

静态扫描发现插件具备触达本地系统的能力:执行外部命令、读写/删除本地文件、读取剪贴板或密钥类环境变量(至少其一)。这是能力清单,不是危险判定——具体能力项与证据见下方,安装前请自行判断是否需要。

读写/删除本地文件读取剪贴板或密钥环境变量发起外部网络请求
查看扫描证据(共 26 处,展示前 26 条)
  • 读取本地文件
    cli/main.mjs:57
    const md = readFileSync(file, 'utf8')
  • 读取本地文件
    cli/main.mjs:65
    return readdirSync(dir)
  • 读取本地文件
    cli/main.mjs:147
    md = readFileSync(file, 'utf8')
  • 读取本地文件
    cli/main.mjs:175
    md = readFileSync(file, 'utf8')
  • 读取本地文件
    cli/main.mjs:200
    const lines = readFileSync(join(HISTORY_DIR, file), 'utf8').split('\n')
  • 读取本地文件
    cli/main.mjs:233
    md = readFileSync(file, 'utf8')
  • 发起网络请求
    lib/dsh-data/market.js:88
    res = await fetch(url, { signal });
  • 发起网络请求
    lib/dsh-data/market.js:186
    res = await fetch(url, { signal, headers });
  • 读取密钥环境变量
    lib/dsh-community/github.js:64
    if (process.env.GITHUB_TOKEN)
  • 读取密钥环境变量
    lib/dsh-community/github.js:65
    headers.Authorization = `Bearer ${process.env.GITHUB_TOKEN}`;
  • 发起网络请求
    lib/dsh-community/github.js:67
    fetch(base, { signal, headers }),
  • 发起网络请求
    lib/dsh-community/github.js:68
    fetch(`${base}/pulls?state=open&per_page=100`, { signal, headers }),
  • 发起网络请求
    lib/dsh-community/github.js:69
    fetch(`${base}/releases/latest`, { signal, headers }),
  • 请求外部主机registry.npmjs.org
    lib/dsh-community/npm.js:40
    fetch(`https://registry.npmjs.org/${enc}`, { signal }),
  • 发起网络请求
    lib/dsh-community/npm.js:40
    fetch(`https://registry.npmjs.org/${enc}`, { signal }),
  • 请求外部主机api.npmjs.org
    lib/dsh-community/npm.js:41
    fetch(`https://api.npmjs.org/downloads/point/last-week/${enc}`, { signal }),
  • 发起网络请求
    lib/dsh-community/npm.js:41
    fetch(`https://api.npmjs.org/downloads/point/last-week/${enc}`, { signal }),
  • 请求外部主机api.npmjs.org
    lib/dsh-community/npm.js:42
    fetch(`https://api.npmjs.org/downloads/point/last-month/${enc}`, { signal }),
  • 发起网络请求
    lib/dsh-community/npm.js:42
    fetch(`https://api.npmjs.org/downloads/point/last-month/${enc}`, { signal }),
  • 读取本地文件
    cli/tui.mjs:67
    return readdirSync(dir)
  • 读取本地文件
    cli/tui.mjs:73
    const md = readFileSync(file, 'utf8')
  • 读取本地文件
    cli/tui.mjs:179
    const md = readFileSync(join(DIR, f), 'utf8').toLowerCase()
  • 读取本地文件
    cli/tui.mjs:228
    pagerLines = renderMd(readFileSync(join(DIR, firm), 'utf8')).split('\n')
  • 写入本地文件
    demos/gen-ui-demo-data.ts:51
    writeFileSync(new URL('./ui-demo-data.json', import.meta.url), JSON.stringify(demoData))
  • 读取本地文件
    demos/gen-ui-demo-data.ts:53
    const html = readFileSync(new URL('./ui-demo.html', import.meta.url), 'utf8')
  • 写入本地文件
    demos/gen-ui-demo-data.ts:55
    writeFileSync(new URL('./ui-demo-standalone.html', import.meta.url), selfContained)
扫描于 2026/11/3(0 天前) · 范围 入口闭包 + 全包扫描 · 来源 npm 发布包
涉及外部主机:registry.npmjs.org、api.npmjs.org

静态扫描只反映源码中出现的调用,不构成安全背书;人工评估与自动扫描相互独立、互不覆盖。

依赖的 DSH / Cordis 模块
@deepseek-ai/cordis@deepseek-ai/dsh-llm@deepseek-ai/dsh-tools@deepseek-ai/dsh-system-prompt@deepseek-ai/cordis-plugin-loader@deepseek-ai/cordis-plugin-include
给这个插件打分还没有人投票↓
用户评分
还没有人投票,来当第一个

界面预览

3 张 · 点击查看大图

功能亮点

  • AI原生量化操作系统
v0.91.02026-09-18
QRT(Qube Research & Technologies)按 DD_STANDARD 重写成九部分深度尽调(5 个并行研究智能体、约 130 次检索、标题级证据)。 · 资管规模链条补全:10 亿美元(2016 年分拆)→ 160 亿美元(2023-02)→ 200 亿美元(2024)→ 300 亿美元(2025-07)→ 420 亿美元(2026-02)→ 500 亿美元(2026-07) · 2025 年业绩:旗舰基金 +30% / 整体 +22%(双来源) · 中国路径更正:QFII(立方科研)+ WFOE PFM(宽立资本)——不是 QDLP;Dao 基金 98% 重新基准化为 2022 年 11 月以来累计 · 删除「Olivier Jaillon = CEO」这一前提(零共现);Companies House 09867306 待查 · 2026-07 人工选股团队给「纯量化」标签加了一条限定 · QRT Labs(牛津/剑桥/帝国理工)= 人才开放,技术封闭 · 监管记录清白:日本冒名 = 受害方,DJT SEC 投诉 = 未采取行动,汇丰做空报告故障 = 技术错误 · 新增 14 项待核实清单;公告快照:ann/2026-09-18_qrt_deep_dd_launch.md
v0.90.02026-08-23
· HIGHFLYER_SPECIAL——双引擎之王 梁文锋的量化公司如何资助 AGI:量化引擎(700 亿+ 闭源,2025 年 +56.6%)、AGI 引擎(DeepSeek 投前 450 亿美元,大基金领投,马/曾/丁/刘出资超 500 亿)、萤火超算血脉(3FS 已开源)、母静子显的开放模式(High-Flyer → DeepSeek → dsh → dsh-quant)、全球 AI 界最富创始人(360 亿美元)、2026-08 对冲基金回撤,以及与 UBIQUANT_SPECIAL 的双开源对照。 94 份档案 + 50 份研究报告。 公告:Discussion #104
v0.89.02026-08-23
· UBIQUANT_SPECIAL——九坤的 AI 转向 WorldQuant 系创始人,约 800 亿 AUM,碰上范式翻覆(2018/2021 冻结、3 年锁定、自购),AI 转向(「北冥」超算、「80% 策略用上 AI」),2026 年开源的 IQuest-Coder-V1(40B 代码模型,胜过 GPT-5.1,流式训练),智知创新研究院 + 上海交大实验室,2022 年 -39% 的杠杆教训,以及「第二个 DeepSeek 时刻」的论点。 94 份档案 + 49 份研究报告。 公告:讨论 #103
数据来自仓库的 GitHub Release,可能与插件实际版本不一致。

README

由 DeepSeek 最新模型翻译生成
🐳 dsh-quant — 万物皆插件的量化操作系统 📣 公告归档:2026-09-01 X 开源发布文案 🌐 站点:https://dsh-quant-site.pages.dev · ✅ 已收录进 awesome-dsh-plugin(通过 dsh-market 一键安装) AI 原生、DSH 原生的量化工具箱,覆盖量化的每个环节 — 59 个工具 · 6 大领域(数据 / 因子 / ML / 风控 / 执行 / 生态)· 一条端到端 PDAT→PET 研究流水线。方法开源,机密留在内部。 🧩 核心理念:万物皆插件(量化版) dsh 的理念是万物皆插件;dsh-quant 把它带进量化 —— 把内部五团队范式 (PDAT → PAAT → PCPT → PRT → PET)开源成五个可插拔模块: data plugin dsh-data market data / sources / quality ← plug in Binance or your own data alpha plugin dsh-alpha indicators / factors / eval ← write your own alpha (internal alpha stays private)
展开完整文档(README 剩余 1 节)
🐳 dsh-quant — 万物皆插件的量化操作系统

📣 公告归档:2026-09-01 X 开源发布文案

🌐 站点:https://dsh-quant-site.pages.dev · ✅ 已收录进 awesome-dsh-plugin(通过 dsh-market 一键安装)

AI 原生、DSH 原生的量化工具箱,覆盖量化的每个环节 — 59 个工具 · 6 大领域(数据 / 因子 / ML / 风控 / 执行 / 生态)· 一条端到端 PDAT→PET 研究流水线。方法开源,机密留在内部。

🧩 核心理念:万物皆插件(量化版)

dsh 的理念是万物皆插件;dsh-quant 把它带进量化 —— 把内部五团队范式
(PDAT → PAAT → PCPT → PRT → PET)开源成五个可插拔模块:

data plugin dsh-data market data / sources / quality ← plug in Binance or your own data
alpha plugin dsh-alpha indicators / factors / eval ← write your own alpha (internal alpha stays private)
model plugin dsh-ml backtests / ML/DL/RL framework ← train your own models (internal research stays private)
risk plugin dsh-risk VaR / drawdown / options / bonds ← set your own risk limits
exec plugin dsh-execution sim execution / fund / report ← build your own trading system (paper or live)

- 开源的是范式:模块怎么组合、契约怎么定义(空值对齐 / 无未来函数 / 手算校验)、结果怎么验证 ——而不是内部机密
- 内容由你填:产品力 = UI + 策略 + 数据接口 + DL/RL 模型 + 交易系统搭建,全部自己组装,全部插件化
- 无限自我进化:往框架里填你的模块 → 跑模拟盘/实盘 → 把成果回馈生态 —— 这就是 dsh-quant

插件征集:Issue #27(五个模块 × 众多插件) —— PDAT 插件、PET 插件,你能想到的都可以 🐋

🤖 AI 原生是刻意设计(设计声明)

dsh-quant 的主要使用者是智能体(模型),不是人——这是从第一天起就定下的刻意选择:

- 工具 schema 注入系统提示词——每份契约(参数 / 输出 / 对齐规则)都从模型视角写
- 等长 null 对齐——输出与输入等长;开头窗口位置为 null,模型按索引对齐,永远不用补齐
- 规范 JSON 与渲染分离——机器读结构,人读文字
- 全部 isConcurrencySafe——纯函数,无共享状态;智能体可并行调用全部 59 个工具,互不干扰
- 技能层——skill/quant-research 让模型自己加载工作流

完整声明见 Issue #14「AI-native is deliberate」。

🐍 为什么几乎没有 Python?

常见疑问:量化项目没有 .py 文件?答案:0 个 Python 文件,21 个 TypeScript 源文件,零运行时依赖——这是刻意的:

- dsh-quant 是 dsh 插件,跑在 harness 的 Node 运行时里:与智能体同进程,可由 Loader 组合,注册可回滚(HMR 安全)——TypeScript 是唯一自然的选择
- 需要 Python 的地方(akshare / tushare / baostock),走通道知识库(quant_data_guide,15 个通道):dsh 自带 shell / subprocess 能力,由智能体决定是否拉起 Python——dsh-quant 不带运行时
- 所有数值方法(指标 / 回测 / 期权 / 债券)都是纯函数,附手算基线:零依赖、可离线验证,npm test 在任何地方都能通过

目标只有一个:打造最好用的 AI 原生量化仓库 🐋

59 个 quant_ 工具 · 6 个领域 · 298 个测试 · 零运行时依赖。完整定位见置顶 Issue #9。

🤖 给来访智能体

落到这个仓库的智能体:先看 AGENTS.md(根目录,Copilot / Claude Code / dsh 会自动读取),再看闭环示例 plugin/CASE_STUDY.md 或 Discussion #38,以及一眼看懂契约的 mcp/AGENT_GUIDE.md。

兼容版本发布状态

0.91.1 候选版本包含 DSH 0.2.1-rc.2 兼容性,同时保留 0.1.0-rc.6 / 0.2.0-rc.2 / 0.2.1-alpha.3 的覆盖。本候选版本于 2026-10-18 准备时,npm 上仍是 0.90.0。把修复当作已发布之前,请先查交付记录和 npm view dsh-quant version。原始兼容性贡献:@hufeide(#136);用户反馈:@ziladan(#139 / #140)。

v0.91.1 标签固定在已评审的源码上。首次带认证的发布尝试因 npm E404 失败,所以该标签不能作为 npm 可用的证据。见发布尝试。
维护者可以手动运行 release acceptance,用 source=candidate 复现 Linux 包;发布之后也可以用 source=registry 验证四个干净的真实 Loader 消费方。这个工作流从不发布。

快速安装(dsh 用户)

npm i dsh-quant

在 cordis.yml 里加一行:

- name: 'dsh-quant'

59 个工具自动注册——指标 / 回测 / 因子 / 风险 / 基金模拟 / 生态指标开箱即用。一个 quant_research_pipeline 跑完整个 PDAT→PET 链路。ML/DL 知识见 docs/ML_GUIDE.md;可执行示例见 npx tsx demos/ml-workflow.ts。

🚀 产品体验:三分钟跑通完整量化流水线

安装完就能体验完整的 PDAT→PET 流程(BTC 公开数据 + 简单策略 + 回测 + 模拟盘):

data(quant_market_fetch) → quality(quant_data_quality) → factors(quant_factor_evaluate)
→ backtest(quant_backtest) → metrics(quant_metrics) → risk(quant_risk)
→ drawdown(quant_drawdown) → paper sim(quant_execute_sim) → fund sim(quant_fund)
→ report(quant_report)

一行搞定:quant_research_pipeline(symbol=BTCUSDT, limit=120) 一次调用返回全部结果。

然后往每个模块里插你自己的插件(数据源 / alpha / 模型 / 风险 / 执行——一切都是插件,提案见 Issue #27)。

带讲解的五步上手:docs/ONBOARDING.md ·智能体速览指南:mcp/AGENT_GUIDE.md

🖥️ UI 工作台(dsh-quant-ui)

dsh-quant UI

dsh-quant-ui:K 线 + MA 叠加 + 成交标记、净值曲线、基金 NAV / 管理费 / 业绩费卡片、指标选择器——还有一只游动的 Q 版鲸鱼 🐋(连点标题 3 次)。

在线演示:https://dsh-quant-ui.pages.dev

⌨️ 命令行(dsh-quant 终端)

零依赖、可读的终端界面(纯 Node + ANSI,和 P-Research 命令行同一套理念)。不用浏览器就能浏览研究专栏和实时行情:

node cli/main.mjs repo # 59 tools · 6 domains
node cli/main.mjs history # 53 firm archives index
node cli/main.mjs history citadel # one firm's archive (rendered)
node cli/main.mjs history --reports # ANALYSIS / TIMELINE / LINEAGE / BANK_LINEAGE
node cli/main.mjs history --search 高频 # cross-archive search
node cli/main.mjs kline BTCUSDT --limit 20 # colored OHLC table + stats
node cli/main.mjs browse # interactive TUI: arrow-key firm browser

执行 npm install -g . 之后,命令可简写为 dsh-quant repo、dsh-quant history citadel 等。

工具

| 工具 | 参数 | 规范输出 | 首个有效索引 |
| quant_data_compare | dataType(如 "financials"/"daily bars") | { dataType, channels: [{ name, cost, covers, bestFor }] }(覆盖优先) | — |
| quant_data_advice | dataType + budget(free/low/institutional)+ purpose(research/backtest/official) | { recommendations: [{ rank, name, reason }] }(决策树排序) | — |
| quant_series_stats | values: number[] | { count, mean, std, min, max, median, skew, kurtosis, autocorr1, annualizedVol, totalReturnPct } | —(取数后的第一步) |
| quant_var_backtest | returns + varSeries + confidence=0.95 | { failures, expected, lrStat, pValue, passed, periods }(Kupiec POF 检验) | —(VaR 模型的基准真值) |
| quant_option | spot + strike + timeToMaturity + riskFreeRate + type + volatility/price 二选一 | { price, impliedVolatility, delta, gamma, vega, theta, rho, … } | —(灵感来自 Optiver:BS 定价 + 五个希腊字母 + IV 求解) |
| quant_volatility | close: number[] + annualization=252 | { annualized, perPeriod, n, logReturns(aligned) } | —(已实现波动率;RV 与 IV 研究的入口) |
| quant_bond | couponRate + periodsToMaturity + paymentsPerYear? + ytm/price 二选一 | { price, yieldToMaturity, macaulayDuration, modifiedDuration, convexity, dv01, … } | —(FICC 环节:定价/久期/凸性/DV01,教科书式贴现) |
| quant_drawdown | equity: number[] | { underwater(aligned), maxDrawdownPct, currentDrawdownPct, periods(peak/trough/recovery/depth/duration), ongoing } | —(回撤区间分析) |
| quant_resample | candles + period(week=7 bars/month=30 bars) | { candles }(OHLCV 聚合,7×24 市场) | — |
| quant_report | strategy/metrics/risk/factor/fund(各模块输出) | { report }(Markdown 研究报告) | —(研发结论汇总) |
| quant_repo_stats | owner + repo | { stars, forks, watchers, openIssues, openPullRequests, topics, latestRelease, … }(公开 GitHub API,无需凭证) | —(生态数据) |
| quant_npm_stats | pkg | { latest, weeklyDownloads, monthlyDownloads, description, … }(npm registry + downloads API) | —(生态数据) |
| quant_oss_pulse | stars + downloadsWeekly? + starsPrevious? + openIssues? + openPullRequests? + daysSinceRelease? | { score(0-100), grade(A-D), components, suggestions, summary } | —(开源影响力评分;缺失输入按中性 50 计分) |
| quant_stress_test | weights + betas + assetVolsPct + correlation=0.6 | { weights, scenarioLossesPct, worstScenario, maxLossPct, portfolioVolPct, notes } | —(崩盘/流动性/波动情景下的组合损失) |
| quant_risk | returns(小数序列)+ benchmarkReturns? + confidence=0.95 | { var95, cvar95, downsideDeviation, maxDrawdownPct, beta, alpha, informationRatio, trackingError, periods } | —(核心风险模块) |
| quant_fund | equityCurve + initialCapital=1e8 + managementFeeRate=0.02 + performanceFeeRate=0.2 | { initialCapital, finalNavNet, finalAum, peakNav, peakAum, gross/netReturnPct, fees, navNet } | —(量化对冲基金模拟:净值从 1.00 起步,每日计提管理费,业绩费 20%,按高水位线收取) |
| quant_metrics | equityCurve + trades? | { totalReturnPct, maxDrawdownPct, sharpe, annualizedVol, calmar, sortino, winRate, profitFactor, avgPeriodReturnPct, tradeMetrics }(必备三项:收益 / 回撤 / 夏普) | —(METRIC_CATALOG 供 UI 选择器使用) |
| quant_chart | kind(candles/series/annotations)+ 对应数据 | 结构化图表数据(dsh-chart 协议:candles+overlays+markers / 多序列 / 标注视图) | —(UI 路由数据层) |
| quant_execute_sim | close + orders[{index, side, quantity?/valueFraction?}] + initialCash? + feeRate? + slippageBps? + latencyBars? | { fills, equityCurve, finalEquity, totalReturnPct, totalFee, totalSlippageCost, tradeCount, unfilledCount, cash, position } | —(执行框架,不接实盘交易) |
| quant_trading_cost | quantity + price + commissionRate=0.001 + spreadBps=5 + annualVolPct=30 + dailyAdv? + participationRate=0.01 | { totalCostBps, commissionBps, slippageBps, impactBps, notional, notes } | —(佣金 + 滑点 + 市场冲击) |
| quant_trade_quality | fills(来自 quant_execute_sim)+ unfilledOrders? + holdingPeriodBars? | { orders, fills, fillRate, totalSlippageCost, avgSlippageBps, avgHoldingBars, buys, sells, avgFillValue, notes } | —(执行质量:从模拟到实盘预期) |
| quant_research_pipeline | symbol? + interval? + limit? + provider? + candles? + 策略 / 基金参数 | { candles, quality, stats, metrics, risk, drawdown, fund, factor, report, charts } | —(一次调用完成 PDAT→PET 研究) |
| quant_factor_evaluate | factorValues + forwardReturns(factor[i] 预测 ret[i+1])+ quantiles=5 + window=20 + decayHorizons=5 | { ic, rankIc, icDecay, icir, icSeries, quantileReturns, longShort, turnover, autocorr1, n }(alphalens 指标集 + RankIC/IC 衰减) | — |
| quant_factor_neutralize | factorValues + groups? + styleFactors? + method? | { values(standardized), method, groupCount, styleCount, rSquared } | —(分组 z-score / OLS 残差中性化) |
| quant_walk_forward | returns + features[][] + trainWindow + testWindow + step? | { predictions(null-aligned), oosIc, oosRankIc, oosCount, windows, trainR2Mean } | —(滚动训练 / 样本外,无未来函数) |
| quant_rebalance_schedule | driftPerPeriod + costPerRebalance + maxFrequency=60 | { frequencies, totalCosts, bestFrequency, bestCost, costBreakdown, notes } | —(漂移 vs 成本:最优再平衡频率) |
| quant_parameter_sensitivity | baseValue + range=0.2 + steps=9 + metricValues? | { paramName, values, metricValues, baseValue, robustness, bestValue, bestMetric, worstMetric, notes } | —(网格稳健性:平台区还是针尖) |
| quant_linear_model | X(samples×features) + y + lambda? + predictX? + yTest? | { intercept, weights, lambda, trainR2, n, predictions?, testR2?, testIc? } | —(独立 OLS/Ridge 拟合与预测) |
| quant_factor_correlation | factors(等长)+ factorNames? + threshold=0.7 | { factorNames, correlationMatrix, highCorrelationPairs, meanAbsCorrelation, effectiveFactorCount, notes } | —(因子冗余:合成前先去重) |
| quant_factor_combine | factors: number[][](等长)+ weights? | { signal(rank 0..1), effectiveWeights, factorCount } | —(z-score 加权 + 截面排名) |
| quant_ic_decay | factor + returns(等长)+ maxHorizon=10 | { horizons, icByHorizon, halfLife, bestHorizon, peakIc, peakHorizon, signalType, notes } | —(IC 衰减:信号保质期 → 调仓频率) |
| quant_layered_backtest | factor + returns(时间×资产矩阵)+ layers=5 + horizon=5 + feeRate=0.001 | { layers, topEquity, bottomEquity, longShortEquity, topReturnPct, bottomReturnPct, longShortReturnPct, rebalances, layerMeanReturnPct, notes } | —(分位分层回测:因子 → 策略草图) |
| quant_series_quality | values: number[],jumpThreshold=0.2 | { count, missingCount, zOutliers, jumps, longestConstantRun, healthy } | —(序列级质量) |
| quant_data_annotate | values: number[],jumpThreshold=0.2 | { count, annotations: [{index, label, severity, detail}], summary } | —(点级标注,致敬 Scale AI) |
| quant_data_quality | candles(quant_market_fetch 输出) | { count, highBelowLow, nonPositive, timeNotIncreasing, timeGaps, extremeMoves, healthy } | —(分析前健康检查) |
| quant_deflated_sharpe | observedSharpe + numPeriods + numTrials=1 + skewness? + kurtosis? | { observedSharpe, minSignificantSharpe, deflatedSharpe, significant, pValue, notes } | —(Bailey & López de Prado 过拟合校正 Sharpe) |
| quant_data_pit | values: (number\|null)[] + channels? | { healthScore, pit{pass, lookAheadIndices, notes}, survivorship{continuous, gaps, tailTruncated}, channels[] } | —(AI 基础设施质量:时点 / 幸存者 / 通道可靠性) |
| quant_channel_guide | channel(如 "akshare")+ check? + hasCredentials? | { channel, displayName, steps[], prerequisites[], example, fallback, readiness? } | —(面向智能体的通道接入指南 + 就绪检查) |
| quant_data_guide | query(频道名/数据类型,例如 "tushare"/"financials")或 channel(精确名称) | { query, results: [{ name, url, cost, dataTypes, setup, tutorialUrls, bestFor, … }] } | —(内置 15 频道数据知识库:A 股/美股/债券 + dsh 生态数据插件) |
| quant_market_fetch | symbol: string(例如 BTCUSDT / sh600000 / AAPL)、interval: 1m…1M、limit: 1-1000、provider: binance/okx/bybit/sina/tencent/yahoo | { symbol, interval, provider, candles: [{openTime, open, high, low, close, volume}] } | — |
| quant_sma | values: number[]、window: integer | { values: (number\|null)[], window } | 索引 window-1 |
| quant_ema | values: number[]、window: integer | { values: (number\|null)[], window } | 索引 window-1(种子 = 首窗均值,alpha = 2/(w+1)) |
| quant_rsi | values: number[]、window: integer = 14 | { values: (number\|null)[], window } | 索引 window(Wilder 平滑) |
| quant_macd | values: number[]、fast=12、slow=26、signal=9 | { macd, signal, histogram }(等长) | macd:slow-1;signal/histogram:slow+signal-2 |
| quant_bollinger | values: number[]、window=20、multiplier=2 | { upper, middle, lower, window, multiplier } | 索引 window-1(总体标准差) |
| quant_atr | high/low/close: number[]、window=14 | { values: (number\|null)[], window } | 索引 window(Wilder 平滑) |
| quant_kdj | high/low/close: number[]、window=9 | { k, d, j }(等长) | 索引 window-1(RSV 法,K/D 以 50 为种子) |
| quant_williams_r | high/low/close: number[]、window=14 | { values: (number\|null)[], window } | 索引 window-1(范围 -100..0) |
| quant_cci | high/low/close: number[]、window=20 | { values: (number\|null)[], window } | 索引 window-1(±100 超买/超卖) |
| quant_obv | close/volume: number[] | { values: number[] } | 全程(首值为 0,无 null) |
| quant_adx | high/low/close: number[]、window=14 | { adx, plusDi, minusDi, window } | ±DI:索引 window;ADX:索引 2window-1 |
| quant_roc | values: number[]、window=12 | { values: (number\|null)[], window } | 索引 window |
| quant_backtest | close: number[]、fast=10、slow=30、feeRate=0.001、stopLoss?、takeProfit? | { totalReturnPct, maxDrawdownPct, sharpe, position, equityCurve, trades(with exitReason) } | 首次确认交叉后一根 K 线开首笔交易 |
| quant_backtest_bollinger | close: number[]、window=20、multiplier=2、feeRate=0.001、stopLoss?、takeProfit? | 同上(上轨突破买入,中轨下穿卖出) | 首次确认突破后一根 K 线 |
| quant_backtest_rsi | close: number[]、rsiWindow=14、buyBelow=30、sellAbove=70、feeRate=0.001、stopLoss?、takeProfit? | 同上(RSI 上穿 buyBelow 买入,下穿 sellAbove 卖出) | 首次确认信号后一根 K 线 |
| quant_backtest_portfolio | assets: [{name, close}]、weights?、rebalanceEvery?、feeRate=0.001 | { totalReturnPct, maxDrawdownPct, sharpe, equityCurve, assetNames, finalWeights, rebalances } | —(多资产组合) |
| quant_portfolio_optimize | returns(时间×资产矩阵)+ method=maxSharpe\|minVar\|riskParity + iterations? | { method, weights, annualReturnPct, annualVolPct, sharpe, assetSharpe, concentration } | —(权重优化器;把结果喂给 quant_backtest_portfolio) |
| quant_attribution | returns(时间×资产)+ weights(和为 1)+ factorExposures? [time][asset][factor] | { totalReturnPct, assetContributionsPct, assetContribShares, factorContributionsPct, residualPct, factorR2, notes } | —(组合归因:收益从哪来) |
| quant_backtest_grid | close: number[]、fastMin=3、fastMax=10、slowMin=10、slowMax=30、feeRate=0.001 | { results(sorted by return desc), best, fastRange, slowRange, feeRate } | —(网格搜索;跳过 fast >= slow 的组合) |

典型调用链(模型视角)

quant_market_fetch(symbol: BTCUSDT, interval: 1d, limit: 100)
→ take close → quant_sma / quant_ema / quant_rsi / quant_macd / … → quant_backtest

已实测跑通:真实 Binance 日线 → 指标 → 回测(fast 5 / slow 20),端到端。

回测约定

- 双均线交叉:fast SMA 上穿 slow SMA 时全仓买入,下穿时清仓;信号在第 i 根 K 线确认,在第 i+1 根收盘成交(无未来函数)。
- 买卖两边都按名义金额收费(feeRate 为单边费率)。
- 尾部持仓未平:最后一笔交易的 exitIndex/exitPrice/returnPct 为 null。
- position 和 equityCurve 与输入等长;净值归一化(从 1 开始); Sharpe 按日频年化(√365)。

对齐约定

- 所有输出与输入等长;开头未进入窗口的位置为 null —— 模型按下标对齐,无需补齐。
- 空序列或 window > series length 是合法结果(全 null),不是错误。
- 非有限数(NaN/Infinity)在注册表的无损 JSON 参数快照层(模型的 JSON 边界)就被拒绝,永远到不了 execute。
- 约束(window ≥ 1 且为整数、macd fast < slow、atr 数组等长、multiplier > 0)在 execute 里手工校验;抛出的错误经注册表转成 isError 结果。

契约(defineTool)

- 参数使用统一 schema DSL,由 defineTool 在 execute 之前校验(类型 / 必填 / 整数)。
- execute 只返回规范 JSON 值;output.render 生成面向模型的文字说明。
- 每个工具都是 isConcurrencySafe: true —— 纯函数,无共享状态,无副作用,可并行调度。
- 注册是可逆副作用:ctx.tools.register 返回一个 disposer; fiber 销毁时自动注销。

模型体验

模型看到什么

每个工具的名称、描述、JSON schema 都会注入到 system-prompt 组装流程中(ctx.systemPrompt.tools())。描述里写明对齐规则(哪些头部位置为 null),模型无需猜测。

Token 影响

每个工具占用一个固定 schema 块;调用结果按渲染后的内容计费。null 对齐设计避免了模型反复请求补齐。

KV 缓存影响

schema 前缀稳定(只要工具集与顺序不变就复用);结果追加在可复用前缀之后。

发布历史(NEWS)

| 版本 | 日期 | 说明 |
| 0.91.0 | 2026-09-11 | QRT Deep-mode 全面尽调——九节重写(AUM 链条至 $50bn、2025 旗舰 +30%、QFII+WFOE 而非 QDLP、Dao 自 2022-11 累计 98%、移除 Jaillon 前提),5 份研究档案 |
| 0.90.0 | 2026-08-23 | HIGHFLYER_SPECIAL——双引擎之王(量化基金 AGI、DeepSeek $45B、Wenfeng 全球 AI 首富),共 50 份报告 |
| 0.89.0 | 2026-08-23 | UBIQUANT_SPECIAL——中国 AI 转型深度剖析(WorldQuant 血统、IQuest-Coder 40B 开源、范式倾覆),共 49 份报告 |
| 0.88.0 | 2026-08-22 | quant_trading_cost + quant_rebalance_schedule——成本闸门 + 漂移/成本优化器(59 个工具,215 个单元测试) |
| 0.87.0 | 2026-08-22 | 验证工具 ×4——factor_correlation / deflated_sharpe / stress_test / parameter_sensitivity(57 个工具,210 个单元测试) |
| 0.86.0 | 2026-08-22 | 桥接 ×3——quant_layered_backtest / quant_trade_quality / quant_attribution(53 个工具,200 个单元测试) |
| 0.85.0 | 2026-08-22 | quant_ic_decay + quant_portfolio_optimize(50 个工具,193 个单元测试) |
| 0.84.0 | 2026-08-22 | AI 基础设施数据模块 ×4——quant_data_pit / quant_channel_guide / CLI 质量 / 链式循环(48 个工具,186 个单元测试) |
| 0.83.0 | 2026-08-22 | quant_factor_neutralize 修复 + 5 个基线(46 个工具,179 个单元测试) |
| 0.82.0 | 2026-08-22 | TYO_QUANT——东京日元利率中心普查(约 9 家公司、$30M 人才战、Capula 据点),共 48 份报告 |
| 0.81.0 | 2026-08-20 | CHI_QUANT——芝加哥做市城市普查(约 14 家公司、交易所基因、UChicago 输送管道、Citadel 出走),共 47 份报告 |
| 0.80.0 | 2026-08-20 | QUANT_PEOPLE_CN + QUANT_PEOPLE_GLOBAL——第 101 次发布的量化从业人数估算(中国约 30-50k、四城约 25-38k、全球约 80-120k),共 46 份报告 |
| 0.79.0 | 2026-08-20 | QUANT_WORLD_MAP——第 100 次发布特辑:全球量化世界地图(五城轴线、9 条路径、人才树、四城普查综述),共 44 份报告 |
| 0.78.0 | 2026-08-20 | NYC_FOREIGN_QUANT——纽约对冲基金宇宙普查(约 28 家公司、12 个总部、CT 郊区暗核、NY-LDN 双城),共 43 份报告 |
| 0.77.0 | 2026-08-20 | LDN_FOREIGN_QUANT——伦敦全球量化中心普查(约 30 家公司、12 个总部、四城对比),共 42 份报告 |
| 0.76.0 | 2026-08-20 | SG_FOREIGN_QUANT——新加坡外资量化普查(约 20 家公司、加密/做市/家族办公室边缘、HK 双城对比),共 41 份报告 |
| 0.75.0 | 2026-08-20 | HK_FOREIGN_QUANT — 香港外资量化普查(约 26 家公司、5 个类别、总部与分部对比、2025-26 扩张潮),共 40 份报告 |
| 0.74.0 | 2026-08-20 | QRT_SPECIAL — 数据之王深度剖析(Credit Suisse MBO、10 年 420 亿美元、Dao China 10 倍/98%),共 39 份报告 |
| 0.73.0 | 2026-08-20 | TWOSIGMA_SPECIAL — 机器学习先驱深度剖析(DE Shaw 孵化旗舰、数据优先、双创始人治理危机),共 38 份报告 |
| 0.72.0 | 2026-08-20 | DESHAW_SPECIAL — 摇篮之王深度剖析(计算金融起源、DE Shaw Mafia、Anton 超级计算机),共 37 份报告 |
| 0.71.0 | 2026-08-20 | RENAISSANCE_SPECIAL — 黑箱之王深度剖析(Simons 三次转身、Medallion 30 年 66%、利润超 1000 亿美元),共 36 份报告 |
| 0.70.0 | 2026-08-20 | WORLDQUANT_SPECIAL — alpha 工厂深度剖析(BRAIN 众包、1 亿个 alpha、IQC、101 Alphas),共 35 份报告 |
| 0.69.0 | 2026-08-20 | SIG_SPECIAL — 扑克之母深度剖析(概率操作系统、ByteDance 15000 倍、人才树之根),共 34 份报告 |
| 0.68.0 | 2026-08-20 | CITADEL_SPECIAL — 规模之王深度剖析(基金 + 做市双引擎、峰值年 160 亿美元、迈阿密总部),共 33 份报告 |
| 0.67.0 | 2026-08-19 | XTX_SPECIAL — 人均生产力之王深度剖析(人均 1400 万英镑、六大秘诀),共 32 份报告 |
| 0.66.0 | 2026-08-19 | SHOWDOWN_CN_GLOBAL — 中美六维对决(外加透明度反转),共 31 份报告 |
| 0.65.0 | 2026-08-19 | LISTED_QUANT — 上市量化普查(Virtu/Flow/Man + Knight 死亡链条),共 30 份报告 |
| 0.64.0 | 2026-08-19 | CAPITAL_MODEL — 外资资本结构普查(自营/募资/混合),共 29 份报告 |
| 0.63.0 | 2026-08-19 | POD_PLATFORM — pod 商店收官之作(5 个视角 + dsh 同构),共 28 份报告 |
| 0.62.0 | 2026-08-19 | BALYASNY_SPECIAL — 第六家公司深度剖析(Schonfeld 血统 + 2018 年腰斩 + 重建),共 27 份报告 |
| 0.61.0 | 2026-08-19 | MILLENNIUM_SPECIAL — 第五家公司深度剖析(pod 联邦 + 中国人才之根),共 26 份报告 |
| 0.60.0 | 2026-08-19 | POINT72_SPECIAL — 第四家公司深度剖析(SAC 重生 + Cubist + 14 个办公室),共 25 份报告 |
| 0.59.0 | 2026-08-19 | OPTIVER_SPECIAL — 第三家公司深度剖析(荷兰名字 + Ready Trader Go + 工具血统),共 24 份报告 |
| 0.58.0 | 2026-08-19 | JANE_STREET_SPECIAL — 第二家公司深度剖析(SIG 三人组 + OCaml 文化),共 23 份报告 |
| 0.57.0 | 2026-08-19 | IMC_SPECIAL — 首家公司深度剖析特辑(办公室编年史 + Prosperity),共 22 份报告 |
| 0.56.0 | 2026-08-19 | QUANT_VENDORS_CN — 中国的卖铲人层(Kafang/RQAlpha/jqdatasdk),共 21 份报告 |
| 0.55.0 | 2026-08-19 | FOREIGN_CN_MAP_V2 — 完全核实的外资在华地图(7 家 PFM、2024-2026 第二波),共 20 份报告 |
| 0.54.0 | 2026-08-19 | 上海引力 + 外资在华地图 — SHANGHAI_GRAVITY + FOREIGN_CN_MAP,共 19 份报告 |
| 0.53.0 | 2026-08-19 | 量化地图 ×2 — QUANT_MAP_CN + QUANT_MAP_GLOBAL(以城市为中心),共 17 份报告 |
| 0.52.0 | 2026-08-19 | Office 图谱 ×2 — OFFICE_CN + OFFICE_GLOBAL,共 15 份报告 |
| 0.51.0 | 2026-08-19 | 签名百科 ×2 — SIGNATURES_CN + SIGNATURES_GLOBAL,共 13 份报告 |
| 0.50.0 | 2026-08-19 | 年代纪事 ×2 — AGE_CN(2004-2022)+ AGE_GLOBAL(1783-2018),共 11 份报告 |
| 0.49.0 | 2026-08-19 | D 级研究报告 ×4 — REGULATION / TALENT_MAP / STAR_PRODUCTS / QUANT_AI(共 9 份报告) |
| 0.48.0 | 2026-08-19 | 5 份跨境档案 — Tengsheng/Inshiman/Yuansheng/GSR/Eisler(94 家公司) |
| 0.47.0 | 2026-08-19 | 8 份 CN Lite 档案 — Kaifeng/Honghu/Egret/Zhuoshi/Hande/Niankong/Mengxi/Xinhong(89 家公司) |
| 0.46.0 | 2026-08-19 | 7 份 CN Lite 档案 — Shenyi/Jasper/Liyi/Bodao/Zunjia/Qianyi/Pingfanghe(81 家公司) |
| 0.45.0 | 2026-08-19 | 3 份 CN Lite 档案 — Tianyan/Aifang/Maoyuan(74 家公司) |
| 0.44.0 | 2026-08-19 | 10 份 CN Lite 档案 — Zhicheng/Qianxiang/Blackwing/Inno/LongQi/JoinQuant/Evolution/Sixie/Bridgewater-CN/Beyang(71 家公司) |

| 0.43.0 | 2026-08-19 | Golden Bull special — 12 years of quant winners (2014-2025) + archive cross-analysis |
| 0.42.0 | 2026-08-19 | 5 Lite archives — Hongxi/Mingshi/Wenbo/Luoshu/Pansong (61 firms) + founding-date backfill |
| 0.41.0 | 2026-08-19 | Two-mode DD (Deep/Lite) + 3 Lite archives — ChaoQuanZi/YanSheng/Banyang (56 firms) |
| 0.40.0 | 2026-08-19 | DD standard v1 + China batch 1 re-due-diligenced (nine-section format, to-verify lists) |
| 0.39.0 | 2026-08-19 | China batch 2 — Zhixing Tongda/Chengqi/Ruitian/KuanDe/Lingjun/Xiaoyong (53 firms, WorldQuant lineage) |
| 0.38.0 | 2026-08-17 | Bank/brokerage lineage report — 13 firms, two waves, three generations |
| 0.37.0 | 2026-08-17 | China batch 1 — High-Flyer/Ubiquant/Minghong/Yanfu/Century Frontier (47 firms) |
| 0.36.3 | 2026-08-17 | AGENTS.md engagement loop — full vision + ask-your-human CTA |
| 0.36.2 | 2026-08-17 | AGENTS.md + CLAUDE.md agent onboarding |
| 0.36.1 | 2026-08-17 | Five-slot closed-loop case study + 10 supplyable candidates |
| 0.36.0 | 2026-08-17 | plugin/ five-slot external plugin library (22 repos & MCPs) |
| 0.35.2 | 2026-08-17 | Brand line 🐳 Dsh-Quant — The Everything-Plugin Quant OS |
| 0.35.1 | 2026-08-17 | Full English README |
| 0.35.0 | 2026-08-17 | Core UX: PDAT→PET onboarding (BTC example) + mcp/AGENT_GUIDE |
| 0.34.0 | 2026-08-17 | Quant lineage report (five motherships) |
| 0.33.0 | 2026-08-17 | Macro legends batch (42 firms) + first data analysis report |
| 0.32.0 | 2026-08-17 | Systematic Europe batch (37 firms) |
| 0.31.0 | 2026-08-17 | Market-making & crypto batch incl. Alameda failure case (32 firms) |
| 0.30.0 | 2026-08-17 | QRT/Capula/Winton/DRW/Tower batch (27 firms) |
| 0.29.0 | 2026-08-17 | SIG + quant chronicle timeline (22 firms) |
| 0.28.0 | 2026-08-17 | Balyasny/IMC/XTX/Five Rings + DE Shaw boost (21 firms) |
| 0.27.0 | 2026-08-17 | Man Group/AQR/GSA/Bridgewater batch (17 firms) |
| 0.26.0 | 2026-08-17 | Two Sigma/Virtu/DE Shaw/Renaissance batch (13 firms) |
| 0.25.0 | 2026-08-17 | HRT/Point72/Squarepoint batch (9 firms) |
| 0.24.0 | 2026-08-17 | Millennium/WorldQuant/Jump batch (6 firms) |
| 0.23.0 | 2026-08-17 | quant-history + quant-repo columns (Citadel/Optiver/Jane Street) |
| 0.22.0 | 2026-08-17 | Options & volatility board (Optiver-inspired) |
| 0.21.0 | 2026-08-17 | FICC link: quant_bond + bond data channels |
| 0.20.0 | 2026-08-16 | yahoo US/global klines + 13-channel guide + researchMultiAsset |
| 0.19.0 | 2026-08-16 | quant_linear_model + docs/ML_GUIDE + ml-workflow demo |
| 0.18.0 | 2026-08-16 | Chain completion: A-share klines, RankIC/IC decay, neutralization, walk-forward, drawdown, execution sim, pipeline |
| 0.17.0 | 2026-08-16 | dsh-community domain: quant_repo_stats / quant_npm_stats / quant_oss_pulse |
| 0.16.0 | 2026-08-16 | Domain-driven layout ↔ PDAT/PAAT/PCPT/PRT/PET + exchange fallback chain |
| 0.15.0 | 2026-08-16 | Kupiec VaR backtest + resample + report; 100 unit tests milestone |
| 0.14.0 | 2026-08-16 | quant_risk (VaR/CVaR/Beta/Alpha/IR/TE) |
| 0.13.0 | 2026-08-16 | quant_fund (1e8 capital, NAV 1.00, HWM 20% fee) + UI fund cards |
| 0.12.0 | 2026-08-16 | quant_metrics (9+ metrics) + Jane Street-style UI demo |
| 0.11.0 | 2026-08-16 | quant_chart (dsh-chart protocol) |
| 0.10.0 | 2026-08-16 | quant_factor_evaluate / combine (alphalens methodology) |
| 0.9.0 | 2026-08-16 | series stats + data quality + annotation (tribute to Scale AI) |
| 0.8.0 | 2026-08-16 | channel compare + decision-tree advice |
| 0.7.0 | 2026-08-16 | mcp/tools.json + pure-function re-exports + docs |
| 0.6.0 | 2026-08-16 | data channel guide (8 A-share channels) + rename to dsh-quant |
| 0.5.0 | 2026-08-16 | multi-exchange sources (OKX / Bybit) |
| 0.4.0 | 2026-08-16 | multi-asset portfolio backtest (periodic rebalancing) |
| 0.3.0 | 2026-08-16 | strategy family (Bollinger breakout / RSI reversion) + stop-loss/take-profit |
| 0.2.0 | 2026-08-16 | +6 indicators (KDJ / W%R / CCI / OBV / ADX / ROC) |
| 0.1.0 | 2026-08-16 | Launch: market data + 6 indicators + MA backtest/grid + CI/auto-release |

Full records: NEWS.md and CHANGELOG.md.

Known limitations & roadmap

- Market coverage is crypto-first: Binance / OKX / Bybit public APIs (automatic
fallback), no credentials; A-shares go through the channel knowledge base (akshare
et al. as future providers).
- Backtests are a built-in strategy family: dual-MA / Bollinger breakout / RSI
reversion / portfolio rebalancing / grid search; custom strategy callbacks are the
future route.
- presentCall/presentResult not customized: indicator results have no file /
terminal / diff semantics; UI falls back to generic cards.
- Market tools need network: live cases live in verify.ts; offline indicator /
backtest cases are unaffected.

Domain layout (PDAT→PET pipeline mapping)

src/dsh-data/ data (PDAT): 3 exchanges, 15 channels, quality/annotation, resample
src/dsh-alpha/ alpha (PAAT): 12 indicators + factor eval/combine (alphalens methodology)
src/dsh-ml/ portfolio (PCPT): strategy backtests + portfolio + metric catalog
src/dsh-risk/ risk (PRT): VaR/CVaR/Beta/Alpha/IR + Kupiec test + options + bonds
src/dsh-execution/ delivery (PET): chart data plane, fund sim, research report (no live trading)
src/dsh-community/ ecosystem (unique to the open side): GitHub/npm data + influence pulse

The boundary: data and conclusions stay internal; tools and methods ship to
dsh-quant — no alpha, no production strategies, no live-trading engineering, but frameworks, indicators, factor evaluation, UI and demos. See pinned Issue #9.

Quick start (after fork/pull)

npm ci && npm run build && npm test # offline full tests (215 unit + 4 Loader)
npm run test:verify # live market integration (needs network)
npm run gen:tools # regenerate mcp/tools.json

Build & use

build lib/ (tsc, NodeNext ESM; ships .js + .d.ts)
cd quant-indicators && tsc -p tsconfig.json

use in dsh: add one line to cordis.yml
- name: 'dsh-quant'
(the Loader resolves the package exports → lib/index.js from node_modules)

Verification

pure-function numeric correctness + market parsing + backtests (215 cases, node:test, zero deps)
cd deepseek-harness && pnpm exec tsx --test ../quant-indicators/tests/.spec.ts

REAL-composition: cordis.yml booted through the real Loader (registration visible / pipeline / isError / HMR-safety)
cd deepseek-harness && pnpm exec tsx --test ../quant-indicators/tests/loader-composition.spec.ts

harness integration (schemas → execution pipeline → isError → live fetch→indicators→backtest end-to-end)
cd deepseek-harness && pnpm exec tsx ../quant-indicators/verify.ts

consumer simulation: built lib loaded through real node_modules resolution (simulates post-install)
cd deepseek-harness && pnpm exec tsx ../quant-indicators/consumer-test/boot.ts

⭐ Support

If dsh-quant helps your research, a ⭐ makes the project visible to more dsh users.

This whale stands for DeepSeek Harness (dsh) — trading on its holographic screen 🐋

Issues / PRs / discussions welcome; share your domain perspective in
Discussion #10. 🐋

Ecosystem infrastructure: quant ecosystem directory · ecosystem playbook · ecosystem map Discussion #11

Research columns: quant-history (firm archives) · quant-repo (open-source special)

Plugin library (five slots × external repos & MCPs): plugin/

🤝 Contribute & maintain

- Want to add a plugin / data source / learning material? → Issue #109(插件征集) — data / risk / execution plugins, ML-DL & alpha learning resources all welcome
- Filing a bug / feature / help-wanted issue? → ISSUE_GUIDE.md (agent-friendly rules)
- Sending a PR? → CONTRIBUTING.md (dev loop + design contract)
- Maintainer view (how we review & merge): MAINTAINING.md

Every issue is a future PR; every contributor is a future maintainer. We review fast and merge small PRs quickly — credit goes into release notes + the author section. 🐳

👤 About the author

Pengyi Peng — AI-native builder with a mathematics & quant-research background, building this project as an open research sandbox.

- WorldQuant MAPC 2024 · Global 11 / 850 · UK 1 — official Credly badge
- WorldQuant IQC 2024 · Global 232 / 34,142 · UK 5 (finalist)
- dsh-quant: 59 tools · 6 domains · 215 unit tests · 110+ automated releases
- Quant-industry research: 95 firm archives + 53 research reports (in quant-history/)
- GitHub · LinkedIn

Methods open, secrets internal — the author's research is public, the strategies are not.* 🐳

Compatibility contributors

Thanks to @hufeide for the DSH 0.2.x adaptation in #136, and @ziladan for requesting 0.2.1-rc.2 verification in #139 / #140. The compatibility audit records tested versions and release boundaries.

Test contribution under review

@Yuren0312 contributed offline market-provider tests in
#144 for #106. The
independent audit confirms the coverage
gain and records one requested Node-type correction. This contribution is
under review, not yet merged or shipped.
预览分享卡片长什么样 ↓pengpengyi92/dsh-quant
分享 · DeepSeek Harness Hub
pengpengyi92/dsh-quant

🐳 Dsh-Quant:全能插件,AI 原生量化操作系统

回到顶部