<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
<channel>
  <title>Marketmaker.cc Blog</title>
  <link>https://marketmaker.cc/zh/blog/1/</link>
  <atom:link href="https://marketmaker.cc/zh/blog/rss.xml" rel="self" type="application/rss+xml" />
  <description>AI trading, market microstructure, and DeFi insights from Marketmaker.cc.</description>
  <language>zh</language>
  <lastBuildDate>Mon, 25 May 2026 00:00:00 GMT</lastBuildDate>
  <item>
    <title>我们的内部算法揭秘：HRP + 多空 + 基于Hull-White的CVaR</title>
    <link>https://marketmaker.cc/zh/blog/post/portfolio-pipeline-hrp-cvar/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/portfolio-pipeline-hrp-cvar/</guid>
    <pubDate>Mon, 25 May 2026 00:00:00 GMT</pubDate>
    <description>深入探讨 Pipeline——我们基于HRP构建的复合配置算法。以分层风险平价作为基础，通过代理信号和置信度驱动的多空叠加层，以及通过基于Hull-White波动率调整的CVaR进行最终风险校正。包含我们规范中的完整数学原理，以及实际的Rust实现。</description>
    <category>投资组合优化</category>
    <category>HRP</category>
    <category>分层风险平价</category>
    <category>CVaR</category>
    <category>Hull-White</category>
    <category>EWMA</category>
    <category>多空</category>
    <category>风险管理</category>
    <category>Rust</category>
    <category>量化金融</category>
    <enclosure url="https://marketmaker.cc/images/blog/portfolio-pipeline-hrp-cvar.webp" type="image/webp" />
  </item>
  <item>
    <title>12种投资组合优化算法比较：HRP、Black-Litterman、NCO及其他</title>
    <link>https://marketmaker.cc/zh/blog/post/portfolio-optimization-algorithms-compared/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/portfolio-optimization-algorithms-compared/</guid>
    <pubDate>Fri, 22 May 2026 00:00:00 GMT</pubDate>
    <description>一篮子加密货币，十二种配置算法，一次诚实的比较。我们开源了一个Rust投资组合优化器，它通过一个单一接口运行HRP、HERC、MVO、Black-Litterman、NCO、熵池等算法——这里是每种算法的思路以及为什么没有单一赢家。</description>
    <category>投资组合优化</category>
    <category>分层风险平价</category>
    <category>HRP</category>
    <category>均值-方差</category>
    <category>Black-Litterman</category>
    <category>NCO</category>
    <category>风险平价</category>
    <category>Rust</category>
    <category>量化金融</category>
    <category>资产配置</category>
    <enclosure url="https://marketmaker.cc/images/blog/portfolio-optimization-algorithms-compared.webp" type="image/webp" />
  </item>
  <item>
    <title>OneTick：交易所捕捉欺骗者、对冲基金寻找Alpha的平台</title>
    <link>https://marketmaker.cc/zh/blog/post/onetick-tick-analytics-platform/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/onetick-tick-analytics-platform/</guid>
    <pubDate>Fri, 15 May 2026 00:00:00 GMT</pubDate>
    <description>OneTick架构——面向Tick数据的企业级时间序列引擎。通过Event Processor的DAG查询、实时与历史数据统一、市场监控（MiFID II、MAR、SEC）、TCA、量化研究，以及与kdb+的比较。</description>
    <category>onetick</category>
    <category>tick数据</category>
    <category>时间序列</category>
    <category>监控</category>
    <category>TCA</category>
    <category>量化研究</category>
    <category>MiFID II</category>
    <category>kdb+</category>
    <category>算法交易</category>
    <enclosure url="https://marketmaker.cc/images/blog/onetick-overview.png" type="image/png" />
  </item>
  <item>
    <title>TradingAgents：模拟对冲基金的多智能体AI交易框架</title>
    <link>https://marketmaker.cc/zh/blog/post/tradingagents-multi-agent-framework/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/tradingagents-multi-agent-framework/</guid>
    <pubDate>Tue, 12 May 2026 00:00:00 GMT</pubDate>
    <description>深入解析TradingAgents架构 — 基于LangGraph的开源框架，LLM智能体（分析师、研究员、交易员、风控、投资组合经理）通过结构化辩论做出交易决策。</description>
    <category>AI</category>
    <category>多智能体</category>
    <category>LangGraph</category>
    <category>LLM</category>
    <category>交易</category>
    <category>风险管理</category>
    <enclosure url="https://marketmaker.cc/images/blog/tradingagents-multi-agent.png" type="image/png" />
  </item>
  <item>
    <title>预测市场套利：隐藏成本、手续费与真实数学</title>
    <link>https://marketmaker.cc/zh/blog/post/prediction-markets-arbitrage-fees/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/prediction-markets-arbitrage-fees/</guid>
    <pubDate>Tue, 12 May 2026 00:00:00 GMT</pubDate>
    <description>解析Polymarket、Limitless、Predict.fun、Opinion和Kalshi之间的套利。动态手续费、跨链桥、滑点、结算风险——以及为什么5%的价差可能仍然亏损。</description>
    <category>预测市场</category>
    <category>套利</category>
    <category>Polymarket</category>
    <category>手续费</category>
    <category>DeFi</category>
    <category>风险</category>
    <category>跨链</category>
    <enclosure url="https://marketmaker.cc/images/blog/prediction-markets-arbitrage.png" type="image/png" />
  </item>
  <item>
    <title>T-Bricks (Broadridge)：自营交易公司使用的平台架构解析</title>
    <link>https://marketmaker.cc/zh/blog/post/tbricks-broadridge-hft-platform/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/tbricks-broadridge-hft-platform/</guid>
    <pubDate>Sat, 09 May 2026 00:00:00 GMT</pubDate>
    <description>T-Bricks架构 — 用于做市、ETF套利和集中风险管理的C++模块化HFT平台。100+客户，150+交易所，纳秒级延迟。</description>
    <category>tbricks</category>
    <category>broadridge</category>
    <category>hft</category>
    <category>c++</category>
    <category>做市</category>
    <category>低延迟</category>
    <category>自营交易</category>
    <category>算法交易</category>
    <enclosure url="https://marketmaker.cc/images/blog/tbricks-overview.png" type="image/png" />
  </item>
  <item>
    <title>Flowsurface：开源加密货币订单流分析平台</title>
    <link>https://marketmaker.cc/zh/blog/post/flowsurface-open-source-orderflow/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/flowsurface-open-source-orderflow/</guid>
    <pubDate>Sat, 09 May 2026 00:00:00 GMT</pubDate>
    <description>Flowsurface评测——基于Rust的免费桌面应用，实时可视化DOM热力图、足迹图、成交明细和深度梯子。支持Binance、Bybit、Hyperliquid、OKX和MEXC。</description>
    <category>订单流</category>
    <category>热力图</category>
    <category>DOM</category>
    <category>足迹图</category>
    <category>rust</category>
    <category>开源</category>
    <category>加密货币</category>
    <category>做市</category>
    <category>订单簿</category>
    <category>流动性</category>
    <enclosure url="https://marketmaker.cc/images/blog/flowsurface-overview.png" type="image/png" />
  </item>
  <item>
    <title>Fincept Terminal：基于C++和AI的开源Bloomberg Terminal替代方案</title>
    <link>https://marketmaker.cc/zh/blog/post/fincept-terminal-review/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/fincept-terminal-review/</guid>
    <pubDate>Sat, 09 May 2026 00:00:00 GMT</pubDate>
    <description>Fincept Terminal v4深度评测 — 基于C++20和Qt6构建的原生桌面应用，内置37个AI代理、QuantLib和100多个数据连接器。</description>
    <category>fincept</category>
    <category>terminal</category>
    <category>c++</category>
    <category>qt6</category>
    <category>ai</category>
    <category>quant</category>
    <category>open-source</category>
    <category>算法交易</category>
    <enclosure url="https://marketmaker.cc/images/blog/fincept-terminal-overview.png" type="image/png" />
  </item>
  <item>
    <title>Kronos：让K线图说Transformer语言的基础模型</title>
    <link>https://marketmaker.cc/zh/blog/post/kronos-foundation-model-review/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/kronos-foundation-model-review/</guid>
    <pubDate>Sun, 19 Apr 2026 00:00:00 GMT</pubDate>
    <description>Kronos评测——OHLCV蜡烛图预测基础模型。BSQ分词器、层次化解码器、两阶段采样、Qlib训练管线。模型如何学习交易所的&apos;语言&apos;。</description>
    <category>kronos</category>
    <category>机器学习</category>
    <category>transformer</category>
    <category>基础模型</category>
    <category>qlib</category>
    <category>评测</category>
    <category>开源</category>
    <category>分词</category>
    <enclosure url="https://marketmaker.cc/images/blog/kronos-foundation-model-review.png" type="image/png" />
  </item>
  <item>
    <title>Jesse：基于Python和Rust的加密货币量化交易框架</title>
    <link>https://marketmaker.cc/zh/blog/post/jesse-habr-review/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/jesse-habr-review/</guid>
    <pubDate>Sun, 19 Apr 2026 00:00:00 GMT</pubDate>
    <description>Jesse深度评测——加密市场量化交易框架。分钟级模拟引擎、策略状态机、Rust加速指标、防过拟合优化。</description>
    <category>jesse</category>
    <category>量化交易</category>
    <category>加密货币</category>
    <category>回测</category>
    <category>python</category>
    <category>rust</category>
    <category>评测</category>
    <category>开源</category>
    <enclosure url="https://marketmaker.cc/images/blog/jesse-habr-review.png" type="image/png" />
  </item>
  <item>
    <title>AI Hedge Fund：AI分析师投票决定交易的多智能体基金</title>
    <link>https://marketmaker.cc/zh/blog/post/ai-hedge-fund-review/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/ai-hedge-fund-review/</guid>
    <pubDate>Sun, 19 Apr 2026 00:00:00 GMT</pubDate>
    <description>深度解析virattt的AI Hedge Fund——一个开源系统，多个LLM智能体以不同分析风格通过风险过滤器构建投资组合。架构、智能体、局限性及对实际系统的启示。</description>
    <category>LLM</category>
    <category>多智能体</category>
    <category>风险管理</category>
    <category>投资组合</category>
    <category>评测</category>
    <category>github</category>
    <category>开源</category>
    <enclosure url="https://marketmaker.cc/images/blog/ai-hedge-fund-review.png" type="image/png" />
  </item>
  <item>
    <title>AI4Finance Foundation：FinGPT、FinRL和FinRobot量化交易生态系统</title>
    <link>https://marketmaker.cc/zh/blog/post/ai4finance-foundation-guide/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/ai4finance-foundation-guide/</guid>
    <pubDate>Mon, 13 Apr 2026 00:00:00 GMT</pubDate>
    <description>AI4Finance Foundation生态系统完整指南：FinGPT（金融LLM + LoRA）、FinRL（交易强化学习）、FinRobot（多智能体编排）。</description>
    <category>AI4Finance</category>
    <category>FinGPT</category>
    <category>FinRL</category>
    <category>FinRobot</category>
    <category>LoRA</category>
    <category>强化学习</category>
    <category>LLM</category>
    <category>开源</category>
    <enclosure url="https://marketmaker.cc/images/blog/ai4finance-foundation-guide.png" type="image/png" />
  </item>
  <item>
    <title>VectorBT：Python 中最快的回测框架</title>
    <link>https://marketmaker.cc/zh/blog/post/vectorbt-overview/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/vectorbt-overview/</guid>
    <pubDate>Sun, 12 Apr 2026 00:00:00 GMT</pubDate>
    <description>VectorBT 概览 —— 一款创新的量化分析库，凭借 NumPy 和 Numba 的强大功能改变了回测方法。</description>
    <category>回测</category>
    <category>python</category>
    <category>vectorbt</category>
    <category>算法交易</category>
    <category>numba</category>
    <category>量化分析</category>
    <enclosure url="https://marketmaker.cc/images/blog/vectorbt-overview.png" type="image/png" />
  </item>
  <item>
    <title>Copula模型：加密投资组合联合风险建模</title>
    <link>https://marketmaker.cc/zh/blog/post/copula-models-joint-risk-crypto/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/copula-models-joint-risk-crypto/</guid>
    <pubDate>Mon, 30 Mar 2026 00:00:00 GMT</pubDate>
    <description>超越线性相关——使用Copula模型捕获加密投资组合中的尾部依赖和联合风险，精确估计VaR和CVaR。Vine-copula、GARCH-EVT管线和Python实现。</description>
    <category>风险</category>
    <category>copula</category>
    <category>投资组合</category>
    <category>VaR</category>
    <category>尾部依赖</category>
    <enclosure url="https://marketmaker.cc/images/blog/copula-models-joint-risk-crypto.png" type="image/png" />
  </item>
  <item>
    <title>ZigBolt：为什么我们用 Zig 从零打造了自己的 Aeron，实现了每条消息 20 纳秒延迟</title>
    <link>https://marketmaker.cc/zh/blog/post/zigbolt-zig-messaging-hft/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/zigbolt-zig-messaging-hft/</guid>
    <pubDate>Fri, 27 Mar 2026 00:00:00 GMT</pubDate>
    <description>详解我们如何用 Zig 语言从零构建了一套面向高频交易的超低延迟消息系统。无 JVM、无 GC、无意外。SPSC 环形缓冲区 20 纳秒，IPC 30 纳秒，编解码器 0 纳秒。附基准测试数据。</description>
    <category>zigbolt</category>
    <category>zig</category>
    <category>高频交易</category>
    <category>低延迟</category>
    <category>消息系统</category>
    <category>aeron</category>
    <category>ipc</category>
    <category>开源</category>
    <enclosure url="https://marketmaker.cc/images/blog/zigbolt-zig-messaging-hft.webp" type="image/webp" />
  </item>
  <item>
    <title>ETF 投资组合自动再平衡：我们如何为 Tinkoff 投资编写机器人</title>
    <link>https://marketmaker.cc/zh/blog/post/etf-balancer-bot-tinkoff/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/etf-balancer-bot-tinkoff/</guid>
    <pubDate>Thu, 26 Mar 2026 00:00:00 GMT</pubDate>
    <description>基于 TypeScript/Bun 的开源机器人，用于在 Tinkoff 投资平台上自动进行 ETF 投资组合再平衡。四种平衡模式、保证金交易、多账户支持。含源代码。</description>
    <category>etf</category>
    <category>tinkoff</category>
    <category>再平衡</category>
    <category>投资组合</category>
    <category>机器人</category>
    <category>量化交易</category>
    <category>开源</category>
    <enclosure url="https://marketmaker.cc/images/blog/etf-balancer-bot-tinkoff.webp" type="image/webp" />
  </item>
  <item>
    <title>Aeron：驱动半个HFT行业的消息传递系统揭秘</title>
    <link>https://marketmaker.cc/zh/blog/post/aeron-messaging-overview/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/aeron-messaging-overview/</guid>
    <pubDate>Wed, 25 Mar 2026 00:00:00 GMT</pubDate>
    <description>深入解析Aeron——Real Logic为高频交易打造的消息传递系统架构。Transport、Archive、Cluster、Sequencer。内部原理、工作机制与性能瓶颈。</description>
    <category>aeron</category>
    <category>hft</category>
    <category>低延迟</category>
    <category>消息传递</category>
    <category>java</category>
    <category>ipc</category>
    <category>raft</category>
    <category>金融</category>
    <enclosure url="https://marketmaker.cc/images/blog/aeron-messaging-overview.webp" type="image/webp" />
  </item>
  <item>
    <title>如何捕捉山寨币拉盘后的暴跌：系统化方法</title>
    <link>https://marketmaker.cc/zh/blog/post/shitcoin-pump-dump-strategies/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/shitcoin-pump-dump-strategies/</guid>
    <pubDate>Tue, 24 Mar 2026 00:00:00 GMT</pubDate>
    <description>系统化解析山寨币拉盘后的做空策略。资金费率、持仓量、成交量分析、K线形态、连环清算。附实用算法。</description>
    <category>山寨币</category>
    <category>拉盘</category>
    <category>砸盘</category>
    <category>做空</category>
    <category>衍生品</category>
    <category>资金费率</category>
    <category>算法交易</category>
    <enclosure url="https://marketmaker.cc/images/blog/shitcoin-pump-dump-strategies.webp" type="image/webp" />
  </item>
  <item>
    <title>算法交易中的订单类型：从追价限价单到虚拟订单</title>
    <link>https://marketmaker.cc/zh/blog/post/algotrading-order-types/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/algotrading-order-types/</guid>
    <pubDate>Mon, 23 Mar 2026 00:00:00 GMT</pubDate>
    <description>算法交易中订单类型的全面解析：标准交易所订单、追价限价单、基于时间的订单、虚拟/合成订单。附代码示例和实际案例。</description>
    <category>订单</category>
    <category>算法交易</category>
    <category>限价单</category>
    <category>追价</category>
    <category>虚拟订单</category>
    <category>网格机器人</category>
    <category>做市</category>
    <enclosure url="https://marketmaker.cc/images/blog/algotrading-order-types.webp" type="image/webp" />
  </item>
  <item>
    <title>算法交易的K线类型与聚合方法</title>
    <link>https://marketmaker.cc/zh/blog/post/beyond-time-bars-candle-construction/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/beyond-time-bars-candle-construction/</guid>
    <pubDate>Sun, 22 Mar 2026 00:00:00 GMT</pubDate>
    <description>K线构建的双轴分类：17种基础K线类型（时间、成交笔数、成交量、美元、砖形、范围、波动率、Heikin-Ashi、卡吉、折线突破、点数图、TIB、VIB、游程、CUSUM、熵、Delta）× 3种聚合方法（日历、滚动、自适应）= 51种组合。附实现代码和实践建议。</description>
    <category>algotrading</category>
    <category>candles</category>
    <category>market microstructure</category>
    <category>Lopez de Prado</category>
    <category>order flow</category>
    <category>backtesting</category>
    <category>research</category>
    <enclosure url="https://marketmaker.cc/images/blog/beyond-time-bars-candle-construction.webp" type="image/webp" />
  </item>
  <item>
    <title>隐马尔可夫模型在交易中的应用：如何根据市场状态调整策略</title>
    <link>https://marketmaker.cc/zh/blog/post/regime-detection-hmm-adaptive-trading/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/regime-detection-hmm-adaptive-trading/</guid>
    <pubDate>Sat, 21 Mar 2026 00:00:00 GMT</pubDate>
    <description>如何利用隐马尔可夫模型识别当前市场状态（牛市、熊市、横盘），并自动切换交易策略。附 Python 代码和回测结果。</description>
    <category>hmm</category>
    <category>市场状态</category>
    <category>机器学习</category>
    <category>算法交易</category>
    <category>自适应策略</category>
    <category>波动率</category>
    <enclosure url="https://marketmaker.cc/images/blog/regime-detection-hmm-adaptive-trading.webp" type="image/webp" />
  </item>
  <item>
    <title>墙内排队：订单簿密集区的挂单位置分析</title>
    <link>https://marketmaker.cc/zh/blog/post/queue-position-order-book-wall-analysis/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/queue-position-order-book-wall-analysis/</guid>
    <pubDate>Fri, 20 Mar 2026 00:00:00 GMT</pubDate>
    <description>理解自己在价格层级中的排队位置，如何将高频交易从猜测变为精确的工程问题</description>
    <category>订单簿</category>
    <category>queue position</category>
    <category>高频交易</category>
    <category>做市</category>
    <category>FIFO</category>
    <category>算法交易</category>
    <category>市场微观结构</category>
    <enclosure url="https://marketmaker.cc/images/blog/queue-position-order-book-wall-analysis.webp" type="image/webp" />
  </item>
  <item>
    <title>LLM Alpha挖掘：如何从财报电话会议和金融文档中提取交易信号</title>
    <link>https://marketmaker.cc/zh/blog/post/llm-alpha-mining-earnings-calls/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/llm-alpha-mining-earnings-calls/</guid>
    <pubDate>Fri, 20 Mar 2026 00:00:00 GMT</pubDate>
    <description>如何使用大语言模型从投资者电话会议、财报和新闻中提取交易信号。Chain-of-thought提示、结构化提取、信号回测。</description>
    <category>llm</category>
    <category>alpha</category>
    <category>earnings</category>
    <category>nlp</category>
    <category>gpt</category>
    <category>金融分析</category>
    <category>算法交易</category>
    <enclosure url="https://marketmaker.cc/images/blog/llm-alpha-mining-earnings-calls.webp" type="image/webp" />
  </item>
  <item>
    <title>加密货币市场的统计套利与配对交易：从协整到卡尔曼滤波</title>
    <link>https://marketmaker.cc/zh/blog/post/statistical-arbitrage-pairs-trading-crypto/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/statistical-arbitrage-pairs-trading-crypto/</guid>
    <pubDate>Thu, 19 Mar 2026 00:00:00 GMT</pubDate>
    <description>加密货币市场统计套利完全指南。协整分析、卡尔曼滤波、基差策略、跨交易所套利。包含回测与Python代码。</description>
    <category>统计套利</category>
    <category>配对交易</category>
    <category>协整</category>
    <category>卡尔曼</category>
    <category>套利</category>
    <category>算法交易</category>
    <category>加密货币</category>
    <enclosure url="https://marketmaker.cc/images/blog/statistical-arbitrage-pairs-trading-crypto.webp" type="image/webp" />
  </item>
  <item>
    <title>交易者的数字指纹：如何通过订单簿行为识别做市商</title>
    <link>https://marketmaker.cc/zh/blog/post/digital-fingerprint-trader-identification/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/digital-fingerprint-trader-identification/</guid>
    <pubDate>Thu, 19 Mar 2026 00:00:00 GMT</pubDate>
    <description>每个算法都会留下独特的指纹。学会解读它——你就能知道交易对手方是谁。</description>
    <category>fingerprint</category>
    <category>做市商</category>
    <category>订单簿</category>
    <category>Hawkes过程</category>
    <category>聚类</category>
    <category>幌骗</category>
    <category>Hyperliquid</category>
    <category>市场微观结构</category>
    <enclosure url="https://marketmaker.cc/images/blog/digital-fingerprint-trader-identification.webp" type="image/webp" />
  </item>
  <item>
    <title>按活跃时间计算PnL：改变策略排名的指标</title>
    <link>https://marketmaker.cc/zh/blog/post/pnl-active-time-metric/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/pnl-active-time-metric/</guid>
    <pubDate>Wed, 18 Mar 2026 00:00:00 GMT</pubDate>
    <description>为什么年度原始PnL不适合比较不同交易时间的策略。如何计算有效收益率，为什么需要fill_efficiency，以及为什么27% PnL的策略可能优于300%的策略。</description>
    <category>算法交易</category>
    <category>回测</category>
    <category>指标</category>
    <category>PnL</category>
    <category>编排</category>
    <category>投资组合</category>
    <category>风险管理</category>
    <enclosure url="https://marketmaker.cc/images/blog/pnl-active-time-metric.webp" type="image/webp" />
  </item>
  <item>
    <title>自适应下钻：从分钟到原始交易的可变粒度回测</title>
    <link>https://marketmaker.cc/zh/blog/post/adaptive-resolution-drill-down-backtest/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/adaptive-resolution-drill-down-backtest/</guid>
    <pubDate>Tue, 17 Mar 2026 00:00:00 GMT</pubDate>
    <description>自适应数据粒度如何加速回测并节省存储空间：仅在价格显著波动或成交量异常的位置从1分钟下钻到1秒、100毫秒和原始交易，而非对整个历史序列进行处理。</description>
    <category>算法交易</category>
    <category>回测</category>
    <category>parquet</category>
    <category>优化</category>
    <category>粒度</category>
    <category>下钻</category>
    <category>自适应分辨率</category>
    <enclosure url="https://marketmaker.cc/images/blog/adaptive-resolution-drill-down-backtest.webp" type="image/webp" />
  </item>
  <item>
    <title>聚合 Parquet 缓存：如何将多时间框架回测加速数百倍</title>
    <link>https://marketmaker.cc/zh/blog/post/parquet-cache-multitimeframe-backtest/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/parquet-cache-multitimeframe-backtest/</guid>
    <pubDate>Mon, 16 Mar 2026 00:00:00 GMT</pubDate>
    <description>如何从分钟级K线预计算时间框架和指标，保存为 parquet 文件，并在大规模策略测试中使用，避免重复计算。</description>
    <category>算法交易</category>
    <category>回测</category>
    <category>多时间框架</category>
    <category>parquet</category>
    <category>优化</category>
    <category>缓存</category>
    <enclosure url="https://marketmaker.cc/images/blog/parquet-cache-multitimeframe-backtest.webp" type="image/webp" />
  </item>
  <item>
    <title>Walk-Forward 优化：唯一诚实的策略测试方法</title>
    <link>https://marketmaker.cc/zh/blog/post/walk-forward-optimization/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/walk-forward-optimization/</guid>
    <pubDate>Sun, 15 Mar 2026 00:00:00 GMT</pubDate>
    <description>为什么单次训练/测试分割无法防止过拟合，walk-forward 优化如何系统性地验证参数稳健性，以及为什么一个具有 21 个参数、PnL@ML 达 +3342% 的策略在没有 WFO 的情况下是一颗定时炸弹。</description>
    <category>算法交易</category>
    <category>回测</category>
    <category>walk-forward</category>
    <category>过拟合</category>
    <category>验证</category>
    <category>优化</category>
    <enclosure url="https://marketmaker.cc/images/blog/walk-forward-optimization.webp" type="image/webp" />
  </item>
  <item>
    <title>信号相关性：需要监控多少个交易对</title>
    <link>https://marketmaker.cc/zh/blog/post/signal-correlation-pairs/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/signal-correlation-pairs/</guid>
    <pubDate>Sat, 14 Mar 2026 00:00:00 GMT</pubDate>
    <description>为什么10个加密货币交易对无法提供10倍分散化，如何通过correlation_factor计算effective_N，以及实际需要监控多少个交易对才能达到80-90%的编排器插槽利用率。</description>
    <category>算法交易</category>
    <category>相关性</category>
    <category>分散化</category>
    <category>投资组合</category>
    <category>编排</category>
    <category>加密货币</category>
    <enclosure url="https://marketmaker.cc/images/blog/signal-correlation-pairs.webp" type="image/webp" />
  </item>
  <item>
    <title>Polars vs Pandas 算法交易实战基准测试</title>
    <link>https://marketmaker.cc/zh/blog/post/polars-vs-pandas-algotrading/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/polars-vs-pandas-algotrading/</guid>
    <pubDate>Fri, 13 Mar 2026 00:00:00 GMT</pubDate>
    <description>Polars 与 Pandas 在算法交易任务上的详细对比：过滤、聚合、滚动信号计算、I/O 和内存消耗的基准测试。Polars + Numba 混合架构实现最大回测性能。</description>
    <category>算法交易</category>
    <category>Polars</category>
    <category>Pandas</category>
    <category>基准测试</category>
    <category>性能</category>
    <category>数据工程</category>
    <enclosure url="https://marketmaker.cc/images/blog/polars-vs-pandas-algotrading.webp" type="image/webp" />
  </item>
  <item>
    <title>平台分析：如何区分稳健最优与过拟合</title>
    <link>https://marketmaker.cc/zh/blog/post/plateau-analysis-overfitting/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/plateau-analysis-overfitting/</guid>
    <pubDate>Thu, 12 Mar 2026 00:00:00 GMT</pubDate>
    <description>为什么找到最佳策略参数只是工作的一半。如何从视觉和定量两方面区分稳定的平台与脆弱的尖峰，以及为什么Optuna等高线图是将优化后的策略投入生产前的必要步骤。</description>
    <category>算法交易</category>
    <category>回测</category>
    <category>优化</category>
    <category>过拟合</category>
    <category>平台分析</category>
    <category>参数稳定性</category>
    <enclosure url="https://marketmaker.cc/images/blog/plateau-analysis-overfitting.webp" type="image/webp" />
  </item>
  <item>
    <title>坐标下降法 vs 贝叶斯优化：哪种方法能找到更好的参数</title>
    <link>https://marketmaker.cc/zh/blog/post/optuna-vs-coordinate-descent/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/optuna-vs-coordinate-descent/</guid>
    <pubDate>Wed, 11 Mar 2026 00:00:00 GMT</pubDate>
    <description>为什么穷举搜索对12个以上参数不可行，坐标下降法如何遗漏参数交互作用，以及Optuna的TPE采样器如何在500次迭代中找到OAT在96次中无法找到的结果。实用代码示例、采样器对比与多目标优化。</description>
    <category>算法交易</category>
    <category>回测</category>
    <category>优化</category>
    <category>Optuna</category>
    <category>TPE</category>
    <category>贝叶斯优化</category>
    <category>坐标下降法</category>
    <category>超参数</category>
    <enclosure url="https://marketmaker.cc/images/blog/optuna-vs-coordinate-descent.webp" type="image/webp" />
  </item>
  <item>
    <title>多品种验证：在所有交易对上检验你的策略</title>
    <link>https://marketmaker.cc/zh/blog/post/multi-symbol-validation/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/multi-symbol-validation/</guid>
    <pubDate>Tue, 10 Mar 2026 00:00:00 GMT</pubDate>
    <description>为什么在ETHUSDT上优化的策略可能在山寨币上失败。如何正确按交易对分组（蓝筹、大市值、垃圾币）进行测试，以及什么水平的cross-symbol robustness score才算足够。</description>
    <category>算法交易</category>
    <category>回测</category>
    <category>验证</category>
    <category>multi-symbol</category>
    <category>分散化</category>
    <category>加密货币</category>
    <enclosure url="https://marketmaker.cc/images/blog/multi-symbol-validation.webp" type="image/webp" />
  </item>
  <item>
    <title>资金费率扼杀你的杠杆：为什么 PnL×50x 是虚构的</title>
    <link>https://marketmaker.cc/zh/blog/post/funding-rates-kill-leverage/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/funding-rates-kill-leverage/</guid>
    <pubDate>Mon, 09 Mar 2026 00:00:00 GMT</pubDate>
    <description>Binance/Bybit 上的资金费率如何将高杠杆下漂亮的回测结果变成确定的亏损。公式、真实策略的重新计算，以及资金费率不吞噬利润的最大杠杆倍数。</description>
    <category>算法交易</category>
    <category>回测</category>
    <category>资金费率</category>
    <category>杠杆</category>
    <category>风险管理</category>
    <category>加密货币</category>
    <category>Binance</category>
    <enclosure url="https://marketmaker.cc/images/blog/funding-rates-kill-leverage.webp" type="image/webp" />
  </item>
  <item>
    <title>Cascade策略：优先执行与fallback填充</title>
    <link>https://marketmaker.cc/zh/blog/post/cascade-strategies-orchestration/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/cascade-strategies-orchestration/</guid>
    <pubDate>Sun, 08 Mar 2026 00:00:00 GMT</pubDate>
    <description>「回测无幻觉」系列终篇。如何构建N策略×M交易对的编排器，实现级联模式的优先和fallback填充，选择dual_size，以及为什么不能通过简单汇总PnL来回测策略组合。</description>
    <category>算法交易</category>
    <category>编排</category>
    <category>投资组合</category>
    <category>cascade</category>
    <category>策略</category>
    <category>槽位管理</category>
    <enclosure url="https://marketmaker.cc/images/blog/cascade-strategies-orchestration.webp" type="image/webp" />
  </item>
  <item>
    <title>回测与实盘一致性：为什么你的机器人交易表现与回测不同</title>
    <link>https://marketmaker.cc/zh/blog/post/backtest-live-parity/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/backtest-live-parity/</guid>
    <pubDate>Sat, 07 Mar 2026 00:00:00 GMT</pubDate>
    <description>回测与实盘交易之间差异的完整分类：从滑点和部分成交到代码库不同步。实现一致性的架构模式、共享核心模块的 Python 示例以及生产环境监控清单。</description>
    <category>算法交易</category>
    <category>回测</category>
    <category>实盘交易</category>
    <category>回测-实盘一致性</category>
    <category>执行</category>
    <category>NautilusTrader</category>
    <enclosure url="https://marketmaker.cc/images/blog/backtest-live-parity.webp" type="image/webp" />
  </item>
  <item>
    <title>蒙特卡洛自助法：如何用10行代码获取回测的置信区间</title>
    <link>https://marketmaker.cc/zh/blog/post/monte-carlo-bootstrap-backtest/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/monte-carlo-bootstrap-backtest/</guid>
    <pubDate>Fri, 06 Mar 2026 00:00:00 GMT</pubDate>
    <description>为什么回测的单点估计是一种危险的幻觉。蒙特卡洛自助法如何在2秒内计算出PnL和MaxDD的95%置信区间，以及为什么这是策略上线前的必要步骤。</description>
    <category>算法交易</category>
    <category>回测</category>
    <category>蒙特卡洛</category>
    <category>自助法</category>
    <category>置信区间</category>
    <category>风险管理</category>
    <category>统计学</category>
    <enclosure url="https://marketmaker.cc/images/blog/monte-carlo-bootstrap-backtest.webp" type="image/webp" />
  </item>
  <item>
    <title>跨交易所资金费率套利：如何从费率差异中获利</title>
    <link>https://marketmaker.cc/zh/blog/post/funding-rate-arbitrage-cross-exchange/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/funding-rate-arbitrage-cross-exchange/</guid>
    <pubDate>Thu, 05 Mar 2026 00:00:00 GMT</pubDate>
    <description>加密交易所之间的资金费率套利如何运作，为什么 Binance、Bybit、OKX 和 dYdX 上的费率不同，以及如何构建监控和执行系统来从这些差异中提取利润。</description>
    <category>算法交易</category>
    <category>资金费率</category>
    <category>套利</category>
    <category>加密货币</category>
    <category>Binance</category>
    <category>Bybit</category>
    <category>做市</category>
    <enclosure url="https://marketmaker.cc/images/blog/funding-rate-arbitrage-cross-exchange.webp" type="image/webp" />
  </item>
  <item>
    <title>QuestDB 算法交易实战：改变游戏规则的 SQL 扩展</title>
    <link>https://marketmaker.cc/zh/blog/post/questdb-algotrading-sql/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/questdb-algotrading-sql/</guid>
    <pubDate>Wed, 04 Mar 2026 00:00:00 GMT</pubDate>
    <description>深入探讨 QuestDB 的时序 SQL 扩展：SAMPLE BY、ASOF JOIN、HORIZON JOIN、WINDOW JOIN、LATEST ON 及真实交易查询模式。</description>
    <category>QuestDB</category>
    <category>SQL</category>
    <category>ASOF JOIN</category>
    <category>SAMPLE BY</category>
    <category>time-series</category>
    <category>algorithmic trading</category>
    <enclosure url="https://marketmaker.cc/images/blog/questdb-algotrading-sql.webp" type="image/webp" />
  </item>
  <item>
    <title>QuestDB 算法交易实战：从订单簿到生产架构</title>
    <link>https://marketmaker.cc/zh/blog/post/questdb-algotrading-production/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/questdb-algotrading-production/</guid>
    <pubDate>Wed, 04 Mar 2026 00:00:00 GMT</pubDate>
    <description>物化视图、2D 数组订单簿分析，以及基于 QuestDB 的算法交易平台参考架构。</description>
    <category>QuestDB</category>
    <category>materialized views</category>
    <category>order book</category>
    <category>OHLC</category>
    <category>production</category>
    <category>algorithmic trading</category>
    <enclosure url="https://marketmaker.cc/images/blog/questdb-algotrading-production.webp" type="image/webp" />
  </item>
  <item>
    <title>QuestDB 算法交易实战：读懂市场语言的架构设计</title>
    <link>https://marketmaker.cc/zh/blog/post/questdb-algotrading-architecture/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/questdb-algotrading-architecture/</guid>
    <pubDate>Wed, 04 Mar 2026 00:00:00 GMT</pubDate>
    <description>深入解析 QuestDB 三层存储架构——WAL、列式存储与对象存储中的 Parquet——以及算法交易系统的数据库模式设计原则。</description>
    <category>QuestDB</category>
    <category>time-series</category>
    <category>algorithmic trading</category>
    <category>architecture</category>
    <category>infrastructure</category>
    <enclosure url="https://marketmaker.cc/images/blog/questdb-algotrading-architecture.webp" type="image/webp" />
  </item>
  <item>
    <title>算法交易系统中的数据通信：技术综述</title>
    <link>https://marketmaker.cc/zh/blog/post/data-communication-algotrading/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/data-communication-algotrading/</guid>
    <pubDate>Tue, 03 Mar 2026 00:00:00 GMT</pubDate>
    <description>解析算法交易平台各个层级的数据通信技术：从交易所连接协议（REST, WebSocket, FIX）到内部 IPC、消息代理和数据存储。</description>
    <category>算法交易</category>
    <category>架构</category>
    <category>WebSocket</category>
    <category>FIX</category>
    <category>gRPC</category>
    <category>Kafka</category>
    <category>Aeron</category>
    <category>Redis</category>
    <category>QuestDB</category>
    <category>HFT</category>
    <enclosure url="https://marketmaker.cc/images/blog/data-communication-algotrading.webp" type="image/webp" />
  </item>
  <item>
    <title>亏损与盈利的不对称性：正在摧毁你账户的数学原理</title>
    <link>https://marketmaker.cc/zh/blog/post/loss-profit-asymmetry/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/loss-profit-asymmetry/</guid>
    <pubDate>Sun, 01 Mar 2026 00:00:00 GMT</pubDate>
    <description>为什么亏损50%需要盈利100%才能恢复，波动率拖累如何在横盘市场中摧毁资本，以及每位算法交易者必须掌握的风险管理公式。</description>
    <category>风险管理</category>
    <category>数学</category>
    <category>波动率拖累</category>
    <category>算法交易</category>
    <category>凯利准则</category>
    <enclosure url="https://marketmaker.cc/images/blog/loss-profit-asymmetry.webp" type="image/webp" />
  </item>
  <item>
    <title>在 Rust 中执行复杂套利：从纳秒到原子化多步操作</title>
    <link>https://marketmaker.cc/zh/blog/post/complex-arbitrage-execution-rust/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/complex-arbitrage-execution-rust/</guid>
    <pubDate>Sat, 28 Feb 2026 00:00:00 GMT</pubDate>
    <description>如何从 Rust 中榨取执行多步套利的最高性能：io_uring、无锁订单簿、LMAX Disruptor、SIMD、类型状态机和 Arena 分配器。</description>
    <category>Rust</category>
    <category>套利</category>
    <category>HFT</category>
    <category>低延迟</category>
    <category>无锁</category>
    <category>SIMD</category>
    <category>量化交易</category>
    <category>多步交易</category>
    <category>订单执行</category>
    <enclosure url="https://marketmaker.cc/images/blog/complex-arbitrage-execution-rust.webp" type="image/webp" />
  </item>
  <item>
    <title>用于套利的 GNN、Transformer 与 RL：当神经网络学会交易</title>
    <link>https://marketmaker.cc/zh/blog/post/complex-arbitrage-ml-approaches/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/complex-arbitrage-ml-approaches/</guid>
    <pubDate>Fri, 27 Feb 2026 00:00:00 GMT</pubDate>
    <description>图神经网络如何在 78 毫秒内找到套利链，为什么 RL 智能体显示 142% 的年化收益率（而基于规则的机器人仅为 12%），以及如何在 Rust 中构建集成系统。</description>
    <category>套利</category>
    <category>机器学习</category>
    <category>GNN</category>
    <category>Transformer</category>
    <category>强化学习</category>
    <category>Rust</category>
    <category>贝叶斯方法</category>
    <category>在线学习</category>
    <category>期货</category>
    <category>现货</category>
    <enclosure url="https://marketmaker.cc/images/blog/complex-arbitrage-ml-approaches.webp" type="image/webp" />
  </item>
  <item>
    <title>矩阵、张量与热带代数：用于套利检测的线性代数</title>
    <link>https://marketmaker.cc/zh/blog/post/complex-arbitrage-vector-matrix/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/complex-arbitrage-vector-matrix/</guid>
    <pubDate>Thu, 26 Feb 2026 00:00:00 GMT</pubDate>
    <description>汇率矩阵、特征值、热带代数和张量分解如何将加密货币市场的混乱转化为清晰的套利信号。</description>
    <category>套利</category>
    <category>线性代数</category>
    <category>热带代数</category>
    <category>矩阵</category>
    <category>张量</category>
    <category>rust</category>
    <category>加密货币</category>
    <category>优化</category>
    <category>PCA</category>
    <category>特征值</category>
    <enclosure url="https://marketmaker.cc/images/blog/complex-arbitrage-vector-matrix.webp" type="image/webp" />
  </item>
  <item>
    <title>套利中的 Vine Copulas：高维依赖关系建模</title>
    <link>https://marketmaker.cc/zh/blog/post/complex-arbitrage-vine-copulas/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/complex-arbitrage-vine-copulas/</guid>
    <pubDate>Wed, 25 Feb 2026 00:00:00 GMT</pubDate>
    <description>如何使用 Vine Copulas 识别数十个加密资产之间的隐藏依赖关系，并构建强大的高维统计套利策略。</description>
    <category>套利</category>
    <category>Vine Copulas</category>
    <category>统计学</category>
    <category>高维</category>
    <category>rust</category>
    <category>加密货币</category>
    <category>风险管理</category>
    <category>建模</category>
    <enclosure url="https://marketmaker.cc/images/blog/complex-arbitrage-vine-copulas.webp" type="image/webp" />
  </item>
  <item>
    <title>期货-现货套利：从期现套利到 DeFi-CeFi</title>
    <link>https://marketmaker.cc/zh/blog/post/complex-arbitrage-futures-spot/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/complex-arbitrage-futures-spot/</guid>
    <pubDate>Tue, 24 Feb 2026 00:00:00 GMT</pubDate>
    <description>资金费率、基差以及去中心化与中心化市场的融合如何为加密市场的资本创造无风险机会。</description>
    <category>套利</category>
    <category>期货</category>
    <category>现货</category>
    <category>期现套利</category>
    <category>资金费率</category>
    <category>基差</category>
    <category>DeFi</category>
    <category>CeFi</category>
    <category>rust</category>
    <category>德尔塔中性</category>
    <enclosure url="https://marketmaker.cc/images/blog/complex-arbitrage-futures-spot.webp" type="image/webp" />
  </item>
  <item>
    <title>套利检测的图算法：从 Bellman-Ford 到 RICH</title>
    <link>https://marketmaker.cc/zh/blog/post/complex-arbitrage-graph-algorithms/</link>
    <guid isPermaLink="true">https://marketmaker.cc/zh/blog/post/complex-arbitrage-graph-algorithms/</guid>
    <pubDate>Mon, 23 Feb 2026 00:00:00 GMT</pubDate>
    <description>负环、多资产图以及 RICH 算法如何以亚毫秒级的精度在深层加密货币市场中识别套利机会。</description>
    <category>套利</category>
    <category>图算法</category>
    <category>Bellman-Ford</category>
    <category>RICH</category>
    <category>rust</category>
    <category>加密货币</category>
    <category>优化</category>
    <category>负环</category>
    <enclosure url="https://marketmaker.cc/images/blog/complex-arbitrage-graph-algorithms.webp" type="image/webp" />
  </item>
</channel>
</rss>