crypto-quant-signal-mcp
crypto-quant-signal-mcp
AI 交易代理的调用智能层 —— 通过 MCP 提供跨加密货币和传统金融永续合约的复合量化调用、跨交易所套利检测以及具备市场状态感知能力的分类。
实时业绩记录 — 在 900 多次交易调用中保持 90% 以上的方向准确率。公开透明,无需登录。
为什么选择 AlgoVault
大多数 MCP 交易服务器只提供原始数据——价格、订单簿、K 线。你的代理仍然需要自己决定如何处理这些数据。
AlgoVault 则不同。我们为你的代理提供一个答案:一个带有置信度评分的方向性结论,该结论由基于生产级量化系统调优的多因子复合评分引擎生成。每一次调用都被记录,每一个结果都被衡量,且完整的业绩记录从第一天起就是公开的。
它不仅仅是另一个指标封装器,其核心优势在于:
复合评分,而非单一指标噪音。 将多个正交信号(动量震荡指标、趋势结构、衍生品持仓、成交量动态、持仓量流向)融合为单一的加权结论。权重根据实时市场结果数据进行校准,而非使用教科书式的默认值。
具备市场状态感知能力的调用生成。 调用在发出前会经过市场状态分类器的过滤。引擎知道何时发出调用,何时保持沉默——在震荡市场中的趋势跟踪设置会被抑制,而不会被广播。
跨交易所智能。 实时对 Hyperliquid、Binance 和 Bybit 的资金费率进行标准化和对比。目前没有其他 MCP 提供跨交易所衍生品分析功能。
随每次发布公开业绩记录。 每一笔调用都记录了多个时间窗口的成交价格。胜率、利润因子和预期价值均持续计算。无幸存者偏差,无选择性展示。
自适应评分。 指标权重每月根据结果数据重新调整。引擎会学习什么有效并进行调整——你今天得到的调用比上个月的更精准。
覆盖加密货币 + 传统金融。 支持 290 多种资产——标准加密货币永续合约、传统金融永续合约(股票、指数、大宗商品、通过 Hyperliquid xyz dex 进行的外汇交易)以及经过流动性过滤的 Meme 币。资产按质量分级,并采用机构级调用过滤。
Related MCP server: Web3 Signals — Crypto Signal Intelligence
30 秒快速上手
无需代码。无需 API 密钥。无需安装。
第一步。 打开 Claude → 设置 → 集成 → 添加自定义连接器
第二步。 输入名称和 URL:
字段 | 值 |
名称 |
|
URL |
|

第三步。 询问 Claude 任何问题:
"Get me a trade signal for ETH on the 4h timeframe"

就是这样。你的 Claude 现在内置了一位量化分析师。
工具
get_trade_signal
为任何支持的资产(加密货币永续合约、传统金融永续合约、流动性过滤后的 Meme 币)返回复合的 买入 / 卖出 / 持有 结论及置信度评分。
底层逻辑:多因子评分引擎评估动量、趋势结构、衍生品情绪、持仓量动态和成交量信念。评分在发出最终结论前会经过市场状态感知过滤器和自适应后处理门控——包括资金流分析、波动率状态检测和趋势持续性衰减。
仅生成高信念调用。引擎旨在当优势不明确时保持沉默。
参数:
coin(string, 必填): 资产代码 — 例如"ETH","BTC","SOL","GOLD","TSLA"或 290 多种支持资产中的任意一种timeframe(string, 默认"15m"):"1m","3m","5m","15m","30m","1h","2h","4h","8h","12h","1d"includeReasoning(boolean, 默认true): 调用逻辑的人类可读解释
输出包括: 调用方向、置信度评分(0–100)、所有计算出的指标值、检测到的市场状态、推理叙述,以及用于下游工具组合的 _algovault 元数据。
scan_funding_arb
扫描 Hyperliquid、Binance 和 Bybit 之间的跨交易所资金费率差异。标准化小时费率与 8 小时费率惯例,计算基点价差,并按复合评分(价差幅度、时间紧迫性以及来自 24 小时历史的资金信念)对机会进行排名。
这是唯一提供跨交易所资金费率套利智能的 MCP 服务器——在一个交易所做多,在另一个做空,捕获价差。
参数:
minSpreadBps(number, 默认5): 包含的最小价差(以基点为单位)limit(number, 默认10): 返回的最大结果数
输出包括: 每个机会的交易所费率、最优多/空方向、年化价差百分比以及下一次资金费率结算时间戳。
get_market_regime
将当前市场环境分类为四种状态之一:TRENDING_UP(上涨趋势)、TRENDING_DOWN(下跌趋势)、RANGING(震荡)或 VOLATILE(波动)。
采用多维分类方法,结合方向强度测量与 ADX 斜率分析(检测趋势增强与衰竭)、成交量加权枢轴检测、ATR 自适应资金费率阈值以及跨交易所资金费率情绪背离。市场状态分类直接影响 get_trade_signal 如何过滤其输出——代理也可以独立使用它进行策略选择和仓位管理。
参数:
coin(string, 必填): 资产代码timeframe(string, 默认"4h"): 用于分析的 K 线时间周期
输出包括: 状态标签、置信度评分、底层指标(趋势强度、波动率解读、价格结构)、跨交易所资金费率情绪,以及通俗易懂的策略建议。
业绩追踪
每一笔调用都从发出到结果进行全程追踪。无一例外。
我们衡量的指标:
时间周期适当评估窗口内的结果价格
PFE 胜率 — 价格在评估窗口内的任何时刻是否向调用方向移动
预期价值 — 每次调用的概率加权平均回报
利润因子 — 总盈利除以总亏损
峰值有利偏移 (PFE) 和最大不利偏移 (MAE)
每个资产、时间周期和质量等级的运行统计数据
持有 (HOLD) 调用免费 — 当引擎说“不要交易”时,你无需付费。只有买入和卖出结论会通过 x402 收费或计入订阅配额。这使我们的激励机制保持一致:只有当我们看到可交易的机会时,你才付费。
持有率:引擎拒绝发出交易调用的扫描百分比。高持有率(目前约为 84%)意味着引擎是选择性的——它仅在多个指标条件一致时才发出买入/卖出调用。
基础设施:
远程模式:PostgreSQL,具有自动结果回填功能
本地模式:SQLite,位于
~/.crypto-quant-signal/performance.db仅追踪高置信度的买入/卖出调用 — 持有 (HOLD) 被排除在外
链上验证
每一笔调用在创建时都会进行哈希处理 (keccak256),并通过每日 Merkle 批次锚定在 Base L2 上。这使得业绩记录不可篡改——我们无法编辑过去的调用。
合约:
0x6485...0f81(Base L2)验证任何调用:
https://api.algovault.com/api/verify-signal?signalId=<ID>查看所有批次:
https://api.algovault.com/api/merkle-batches可视化验证: algovault.com/verify
定价
功能 | 免费 | 入门版 ($9.99/月) | 专业版 ($49/月) | 企业版 ($299/月) | x402 (按调用付费) |
资产 | BTC, ETH | 全部 290+ | 全部 290+ | 全部 290+ | 全部 290+ |
资产类别 | 仅加密货币 | 加密 + 传统金融 | 加密 + 传统金融 | 加密 + 传统金融 | 加密 + 传统金融 |
时间周期 | 15m, 1h | 全部 11 个 | 全部 11 个 | 全部 11 个 | 全部 11 个 |
资金费率套利结果 | 前 5 名 | 无限制 | 无限制 | 无限制 | 无限制 |
业绩记录 | 完全访问 | 完全访问 | 完全访问 | 完全访问 | 完全访问 |
每月调用 | ~100/天 | 3,000/月 | 15,000/月 | 100,000/月 | 无限制 |
支持 | 社区 | 邮件 | 优先支持 | 专属支持 | — |
价格 | $0 | $9.99/月 | $49/月 | $299/月 | $0.01–0.05/调用 |
持有 (HOLD) 调用 | 免费 | 免费 | 免费 | 免费 | 免费 |
* 持有 (HOLD) 结论(引擎说“不要交易”)在所有层级中始终免费——无 x402 费用,不扣除配额。
x402 微支付: AI 代理使用 Base 链上的 USDC 按 HTTP 调用付费——无需注册,无需 API 密钥,无需账单。付款收据即凭证。请参阅 x402.org。
订阅: 在 api.algovault.com/signup 注册。入门版($9.99/月)解锁所有资产和时间周期。结账后立即交付 API 密钥。
开发者指南
远程端点(推荐)
https://api.algovault.com/mcp流式 HTTP 传输。兼容任何 MCP 客户端 — Claude, Cursor, Cline, 自定义代理。
通过 npx 本地安装
npx -y crypto-quant-signal-mcpClaude Desktop / Cursor 配置
{
"mcpServers": {
"crypto-quant-signal": {
"command": "npx",
"args": ["-y", "crypto-quant-signal-mcp"],
"env": { "TRANSPORT": "stdio" }
}
}
}npm 安装
npm install crypto-quant-signal-mcp自托管
git clone https://github.com/AlgoVaultLabs/crypto-quant-signal-mcp
cd crypto-quant-signal-mcp
cp .env.example .env # Edit with your values
npm ci && npm run build
docker compose up -d架构
Agent / Claude / Cursor
│
▼
api.algovault.com/mcp (Streamable HTTP)
│
├─ x402 payment verification (USDC on Base)
├─ API key / subscription check
├─ Free tier fallback
│
▼
MCP Server (Express + @modelcontextprotocol/sdk)
│
├─ Composite Scoring Engine
│ ├─ Multi-factor indicator fusion
│ ├─ Regime-aware signal filtering
│ └─ Adaptive post-processing gates
│
├─ Asset Classification Engine
│ ├─ 4-tier quality system (Blue Chip → Major Alt → TradFi → Meme)
│ └─ Liquidity filter for meme/micro assets
│
├─ Exchange Adapter Layer
│ └─ Hyperliquid (standard + xyz TradFi perps) · Binance, Bybit (funding data)
│
├─ Performance Tracker
│ └─ PostgreSQL (remote) / SQLite (local)
│
└─ Hyperliquid Public API (free, no auth)交易所适配器模式: 所有交易所交互都通过 ExchangeAdapter 接口进行 — 支持 Hyperliquid 上的标准加密货币永续合约和 xyz 传统金融永续合约,并支持 Binance 和 Bybit 进行跨交易所资金费率对比。
套件组合性
每个工具的输出都包含一个 _algovault 元数据块,声明版本和兼容的下游工具:
此工具 | 馈送至 (阶段 2+) |
|
|
|
|
|
|
架构设计旨在实现组合性。所有工具共享一致的 timestamp、coin 和 _algovault 字段 — 下游工具直接接受这些对象作为输入。
隐私
本地模式: 零遥测。没有数据发送到 AlgoVault 服务器。调用历史仅存储在你的机器上。
远程模式: 请求元数据记录用于分析(IP 已哈希,从不存储原始数据)。请参阅 隐私政策。
许可证
MIT
由 AlgoVault Labs 构建 — 为自主交易时代提供调用智能。
Available Tools
11 toolschat_knowledgeARead-onlyInspect
Returns a synthesized natural-language answer with citations, grounded in the AlgoVault knowledge bundle (every MCP tool description, response shape, integration tutorial, and code example). Use when you need an explanation, code pattern, or how-to; for raw ranked snippets without LLM synthesis use search_knowledge (faster, no quota cost). Read-only: calls an LLM, no other side effects. Quota: Free 10/month, Starter 50, Pro 200, Enterprise 2000.
Inputs: question (required), model.
get_market_regimeARead-onlyInspect
Returns the market regime — TRENDING_UP TRENDING_DOWN RANGING VOLATILE — with confidence and a strategy hint, for one crypto perpetual futures. Composite verdict: trend ranging + cross-venue funding rate. Read-only, live exchange APIs. Verified track record: get_track_record or performance://signal-performance; on-chain verified merkle anchor.
Inputs: coin (required), exchange, timeframe.
get_support_resistance_zonesARead-onlyInspect
Returns support and resistance zones for one coin on Binance perpetuals, for one to three timeframes. Read-only. Per timeframe: up to 3 support and 3 resistance zones, ranked outward from price, each with its low and high, distance from price, bars touched in the last 90, tests in the last 30, last touch, and whether it flipped between support and resistance after two closes through it. Also returns volume against its 20-bar average and flags such as no zone above (price above every zone in its history). Use with get_trade_call or get_market_regime for the same coin. (Beta)
Inputs: coin (required), timeframes (required), venue.
get_track_recordARead-onlyInspect
Returns the AlgoVault track record — aggregated PFE win rates by call type, timeframe and asset tier, plus the evaluation methodology and window. The same verified aggregate the performance://signal-performance resource serves, callable from harnesses that bridge tools only. Defaults to the compact aggregate; use include for the per-asset, per-venue or recent-signal breakdowns. Read-only, no side effects.
Inputs: include.
get_trade_callARead-onlyInspect
Returns a composite verdict — BUY SELL HOLD trade call with confidence and market regime — for one crypto or tokenized-stock perpetual futures. One asset only; whole-market scan: scan_trade_calls. Read-only: live exchange APIs, no orders. Verified track record: get_track_record or performance://signal-performance; on-chain verified merkle anchor.
Inputs: coin (required), exchange, timeframe, assetClass, +1 more.
get_trade_signalARead-onlyInspect
Returns a composite verdict — BUY SELL HOLD trade call with confidence and market regime — for one crypto or tokenized-stock perpetual futures. One asset only; whole-market scan: scan_trade_calls. Read-only: live exchange APIs, no orders. Verified track record: get_track_record or performance://signal-performance; on-chain verified merkle anchor. [ALIAS] This tool is an alias of get_trade_call — same behavior, kept for backward compatibility. Prefer get_trade_call for new integrations.
Inputs: coin (required), exchange, timeframe, assetClass, +1 more.
pre_trade_checkARead-onlyInspect
Returns a pre-trade check of a crypto perpetual futures order against your own rules: each passes or fails with its numbers, then GO, CAUTION or WAIT. Adds isolated liquidation distance, whether it hits before your stop, fees, accuracy needed to break even over your hold, funding by venue (cross-venue) and the regime. Read-only: places no orders.
Inputs: coin (required), side (required), sizeUsd (required), rules, +8 more.
scan_funding_arbARead-onlyInspect
Ranked cross-venue funding arbitrage across major crypto perpetual futures venues — funding rate spreads, long one venue short another, as a BUY SELL HOLD composite verdict per pair. AI trading signal for crypto quant and Claude trading agents. Trade call via get_trade_call, market regime via get_market_regime. On-chain verified merkle anchor.
Inputs: limit, minSpreadBps.
scan_market_regimesARead-onlyInspect
Screens a whole crypto perpetual futures venue by market regime — TRENDING_UP TRENDING_DOWN RANGING VOLATILE — and returns each matching coin with confidence, trend strength, ADX, volatility ratio and how long the regime has held. Rows are the market regime classifier output for the last closed bar, sampled on Binance, Bybit, OKX and Bitget; another venue, or a timeframe a venue cannot serve, is refused with the alternatives. Billing: one call per returned row; non-matching coins are free. For a live single-coin read with cross-venue funding, call get_market_regime. Read-only.
Inputs: topN, coins, limit, offset, +11 more.
scan_trade_callsARead-onlyInspect
Returns ranked BUY SELL HOLD trade calls across the top crypto perpetual futures by open interest — one scan for whole-market coverage, each with confidence and market regime. Use this for breadth; use get_trade_call for per-coin depth and reasoning. Read-only: reads live exchange APIs, places no orders.
Inputs: topN, limit, rankBy, oiBasis, +7 more.
search_knowledgeARead-onlyInspect
Returns ranked snippets from the AlgoVault knowledge bundle answering a question about its MCP tools, response shapes, integration patterns (LangChain, LlamaIndex, MAF, CrewAI), or code examples. Call this BEFORE other tool calls to confirm parameter usage and avoid hallucinating tool shapes. Fast: BM25 lexical search, no LLM call, no quota cost. For a synthesized natural-language answer use chat_knowledge. Read-only, no side effects.
Inputs: query (required), limit.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 11 tools
Most tools occupy distinct niches: single-coin depth (get_trade_call, get_market_regime, get_support_resistance_zones) vs market-wide breadth (scan_trade_calls, scan_market_regimes, scan_funding_arb), plus a knowledge pair and a pre-trade validator, and the descriptions explicitly cross-reference when to use which. The one real overlap is get_trade_signal, which is documented as an exact alias of get_trade_call — a deliberate duplicate that an agent could still pick wrongly. chat_knowledge vs search_knowledge are also close, but the LLM-synthesis vs BM25-snippet distinction is spelled out clearly.
Strong snake_case consistency with verb-led names that encode scope: get_* for single-asset reads, scan_* for market-wide screening, plus search_knowledge and get_track_record. Minor deviations are pre_trade_check (order-of-words differs from the get_/scan_ pattern) and chat_knowledge, which reads as a noun-first name rather than a clear verb. Overall predictable and readable.
Eleven tools sit comfortably in the well-scoped range for a signal/analytics server, and each covers a genuinely different angle (validation, per-coin signal, regime, zones, three scans, track record, two knowledge lookups). The count is inflated by one tool: get_trade_signal is a backward-compatibility alias with no unique capability. Slightly redundant but otherwise lean.
For a read-only quant signal server the lifecycle is well covered: discovery (search/chat knowledge), per-coin analysis (trade call, regime, S/R zones), breadth (scans for calls, regimes, funding arb), pre-trade validation, and verification via track record. Deliberately no order placement, which fits the read-only framing. Gaps are minor — no standalone per-coin funding-rate or historical/OHLC lookup, so funding detail on a single asset is only available indirectly through get_market_regime or the arb scan.
Maintenance
Related MCP Connectors
AI research desk for Hyperliquid: ranked opportunities and trade theses. You confirm every trade.
Hyperliquid analytics for AI agents: whale flow, liquidations, funding, risk, signed signals.
Live crypto market signals for AI agents: perp liquidations, funding, OI positioning, yield, FX.
Hyperliquid market data, wallet analytics and liquidation alerts for AI agents.
Related MCP Servers
AlicenseAqualityBmaintenanceProvides comprehensive crypto intelligence for the Hyperliquid exchange, allowing users to query trader profiles, behavioral cohorts, and live market data. It enables AI agents to analyze over 1.8 billion trades, track whale positions, and access real-time liquidation heatmaps.10659 npm10MIT- AlicenseNot gradedqualityDmaintenanceAI-powered crypto signal intelligence for 20 assets (BTC, ETH, SOL, etc). 6 scoring dimensions: whale activity, technical analysis, derivatives flow, narrative strength, sentiment, market structure. Market regime detection (TRENDING/RANGING), portfolio optimization, and accuracy tracking. 9 read-only MCP tools. Free via MCP, $0.001 USDC via x402 on Base for REST API.3MIT
- AlicenseAqualityDmaintenanceFinancial intelligence for AI agents. 31 tools across 8 data sources — regime, derivatives, stablecoin flows, momentum, volatility, macro, DeFi, weather patterns, political cycles, seasonality. The context layer between your agent and a bad trade.319 npm9MIT
- AlicenseAqualityDmaintenanceReal-time crypto intelligence for AI agents. Technical analysis, liquidation heatmaps, sentiment, and funding rates for 50+ Hyperliquid perpetuals via x402 micropayments.1539 PyPI1MIT