1.20.0 • Published 3 weeks agoCLI

crypto-quant-signal-mcp

Licence

MIT

Version

1.20.0

Deps

Size

2.7 MB

Vulns

Weekly

559

Stars

Summary Dependency Versions

crypto-quant-signal-mcp

The Brain Layer for AI Trading Agents
A self-tuning quant ML model with a published track record.
Composite trade calls · cross-venue funding arbitrage · regime-aware market classification
Across 5 exchanges (Binance · Hyperliquid · Bybit · OKX · Bitget) via MCP.

Try Free in Telegram · algovault.com · Live Track Record · How it works · Sign Up · Docs

Live Track Record

91.7%% PFE Win Rate · 151,968 trade calls · 55 on-chain batches · 98.9%% HOLD rate

Every call is hashed at emission and anchored on Base L2 daily. We can't edit history.

→ Open the live dashboard

ERC-8004 Verified Agent on Base

AlgoVault MCP is registered as a canonical trading-brain agent on the ERC-8004 Identity Registry (Base L2). The agent identity is on-chain; the track record links from it.

Identity Registry: 0x8004A169...e539a432
AlgoVault agentId: 44544 (8004-spec ERC-721)
Agent metadata (tokenURI): pinned to IPFS, mirrored at api.algovault.com/api/erc-8004-reputation
Trust signals: Merkle-anchored track record. On-chain attestation pipeline rolling out separately.

Verify the agent on Basescan: open agentId

Drop-in for every MCP client

AlgoVault MCP serves Streamable HTTP at https://api.algovault.com/mcp — MCP-spec compliant, no SDK or wrapper library required. Drop-in for:

Client	Config
Claude Desktop	Settings → Integrations → Add custom connector → `https://api.algovault.com/mcp`
Claude Code (CLI)	`claude mcp add crypto-quant-signal https://api.algovault.com/mcp`
Cursor	`~/.cursor/config.json` → `mcpServers` block → `url: "https://api.algovault.com/mcp"`
Cline	VS Code Cline extension → MCP server settings → add Streamable HTTP server
Codex (OpenAI CLI)	`~/.codex/config.toml` → `[mcp_servers.algovault]` table + `url = "https://api.algovault.com/mcp"` (or `codex mcp` CLI)
Windsurf	`~/.codeium/windsurf/mcp_config.json` → `mcpServers.algovault.serverUrl = "https://api.algovault.com/mcp"`
Continue.dev	`config.yaml` → `mcpServers: [{ name: algovault, type: streamable-http, url: "https://api.algovault.com/mcp" }]`
Any other MCP-spec-compliant client	Configure the Streamable HTTP transport with URL `https://api.algovault.com/mcp`

MCP tools/list + resources/list is the API surface. No @algovault/sdk; the protocol is the contract.

Building inside an agent framework? See the Framework integrations section below for drop-in tutorials.

Framework integrations

Drop-in tutorials for the major Python agent frameworks. Each tutorial pairs AlgoVault MCP with the framework's canonical MCP-adapter library — copy-pasteable demo code, no AlgoVault SDK required.

Framework	Tutorial	Runnable demo	Mirror
LangChain	`docs/integrations/langchain.md`	`examples/langchain/demo.py`	algovault.com/integrations/langchain
LlamaIndex	`docs/integrations/llamaindex.md`	`examples/llamaindex/demo.py`	algovault.com/integrations/llamaindex
Microsoft Agent Framework	`docs/integrations/maf.md`	`examples/maf/demo.py`	algovault.com/integrations/maf
CrewAI	`docs/integrations/crewai.md`	`examples/crewai/demo.py`	algovault.com/integrations/crewai

Each demo is runnable as python examples/<framework>/demo.py BTC 4h — gets a real BUY/SELL/HOLD verdict from api.algovault.com/mcp, prints it. ≤5 minutes to first call.

What's new in v1.20.0

Live since 2026-06-04:

scan_trade_calls — market-wide scanner. One call scans the top-N perps (1–100, ranked by open interest) on your chosen venue and returns every actionable BUY/SELL with confidence and regime. HOLDs stay free — quota counts only actionable calls.
Webhooks: dynamic top:N watchlists. Subscribe with assets: ["top:25"] and your webhook follows the venue's top perps automatically. No manual list upkeep.
TradFi-aware analysis. Both core tools now report the underlying market session (underlying_session, with weekend/holiday caveats), interpret fixed pre-IPO funding correctly, and aggregate cross-venue funding sentiment for stocks, indices, commodities, and FX across all 5 venues.
Smarter errors on young listings. Insufficient history returns a structured INSUFFICIENT_CANDLES error with suggested_timeframes instead of a plain string.

MCP clients cache tools/list at session start — toggle the connector off/on (or restart the MCP connection) to see scan_trade_calls.

v1.19.0 highlights (recap)

Webhook Delivery Service. Subscribe a URL; receive trade_call and regime_shift events as push, not poll. Payloads are HMAC-SHA256 signed, retried, and idempotent. Manage over HTTP: POST / GET / DELETE /api/webhooks, plus POST /api/webhooks/:id/test for a live test fire. API key required.
One-call signal verification. Every webhook payload carries a verify_url. Resolve it at GET /api/verify-signal?hash=<signal_hash> for that call's public, on-chain-anchored record.
Two more platform integration examples. The algovault-integrations mono-repo now ships 7 reference implementations — Hummingbot and FreqTrade join the existing five. New tiles on algovault.com/integrations.

→ Set up a webhook: algovault.com/integrations#webhooks · Webhook docs: docs/WEBHOOKS.md

v1.18.2 highlights (recap)

Two new platform integration examples. The algovault-integrations mono-repo now ships 5 reference implementations total (was 3 at v1.18.0):
- QuantDinger (reference-architecture doc) — cross-MCP IDE-mediated orchestration. AlgoVault MCP and QuantDinger MCP connect through the user's IDE (Claude Desktop / Cursor / Codex CLI); no transformer code.
- Cryptohopper (TypeScript) — Signaler webhook with HMAC-sha512 signed payloads. First HMAC-authed transport in the integrations repo.
"Connect Your Trading Platform" section added to algovault.com/integrations. Discoverable tiles for the 5 platform examples; "Why integrate with AlgoVault?" framing replaces the prior exchange-kit framing.

v1.18.0 highlights (recap)

3-minute timeframe goes LIVE on the public Live Track Record. The 3m card and Methodology > Evaluation Windows row are now visible at algovault.com/track-record. The 9th evaluated timeframe is live. (The 3m key shipped to /api/performance-public.byTimeframe in v1.16.0; v1.18.0 completes the dashboard-display chain.)
Platform integration examples. New public repo at github.com/AlgoVaultLabs/algovault-integrations ships reference implementations of the Verifiable-Signal v1.0 spec (introduced in v1.17.0). Three platforms in v1.18.0:
- AI4Trade (TypeScript) — REST POST with token-in-body auth.
- Nautilus Trader (Python) — Subclass of Nautilus's Data abstract base.
- 3Commas (TypeScript) — Signal Bot webhook via per-bot curry-pattern factory.

v1.17.0 highlights (recap)

Free-tier quota visibility. MCP tool responses surface usage progress at 75% and 90% of the monthly quota. At 100% the tool returns a structured error (TIER_LIMIT_REACHED) with an upgrade link.
Stripe webhook idempotency. checkout.session.completed events processed exactly once; UTM tags round-trip through Stripe metadata.
TensorBlock listing. 5th external discovery channel alongside npm registry, MCP Registry, LobeHub, and Claude Desktop DXT.

v1.16.0 highlights (recap)

3-minute timeframe public. Added to the public track record. PFE win rate and sample count visible at algovault.com/track-record.
Integration walkthroughs. Five new walkthroughs at algovault.com/integrations for Claude Desktop, Claude.ai, Cursor, Cline, and Codex CLI.
Knowledge tools docs. New documentation section at algovault.com/docs covering the search_knowledge and chat_knowledge MCP tools.
Bundle expansion. Knowledge tools now index blog posts, YouTube videos, and GitHub Discussions. Bundle refreshes weekly.

v1.15.1 highlights (recap)

Chat usage analytics. Admin-gated /admin/chat-analytics dashboard with weekly Telegram digest. PII-safe per-call telemetry.
Track-record dashboard polish. LATEST TRADE CALLS panel now matches surrounding card surfaces.
npm catch-up. v1.15.0 was published to MCP Registry + LobeHub + DXT + GitHub Release but skipped npm; v1.15.1 brought every surface into sync.

v1.15.0 highlights (recap)

search_knowledge MCP tool. BM25 lexical search over AlgoVault's full knowledge bundle. Free, fast, no LLM call, no quota cost. Also available via POST /api/search.
chat_knowledge MCP tool. Natural-language Q&A with citations, grounded in the canonical knowledge bundle. Backed by Claude Haiku 4.5 with prompt caching. Also available via POST /api/chat.
Zero manual refresh. Both tools index the auto-generated dist/knowledge/latest.json bundle via an in-process fs.watchFile poll (30s).

v1.14.1 highlights (recap)

Auto-generated knowledge bundle JSON. New public endpoints https://api.algovault.com/knowledge/latest.json (+ versioned /knowledge/v1.14.1.json + index /knowledge/index.json) serve a single source-derived KnowledgeBundle (every MCP tool description, response-shape audit snapshot, integration tutorial, package metadata, README "What's new" section) indexed for LLM consumption. Discoverable via MCP resource scheme algovault://knowledge/latest. Also attached as a GitHub Release asset on every npm version. Cache-Control: 1 hour. Foundation for v1.15.0's search_knowledge / chat_knowledge agent tools above.
get_trade_signal alias description refreshed. The [ALIAS] tag prefix replaces the prior parenthetical suffix. Future tool aliases follow the same shape.

v1.14.0 highlights (recap)

Framework integrations live. 4 drop-in tutorials shipped in algovault-skills: LangChain, LlamaIndex, Microsoft Agent Framework, CrewAI. Each tutorial pairs AlgoVault MCP with the framework's canonical MCP-adapter library. Copy-pasteable demo code; no SDK required. See the Framework integrations section above.
Cross-linked from algovault-skills + signal-MCP READMEs. Both repos surface the 4 tutorials so framework users discovering AlgoVault from either side find the right entry point.

v1.13.2 highlights (recap)

Tool descriptions rewritten for retrieval ranking. get_trade_call, scan_funding_arb, get_market_regime descriptions rewritten for Anthropic Tool Search (BM25 + regex retrieval over tools/list). Param describe() strings tightened. Zero schema mutation — same enum members, same defaults, same Zod constraints. Pure prose payload change.
ERC-8004 Verified Agent on Base. AlgoVault MCP is registered on the canonical ERC-8004 Identity Registry at 0x8004A169...e539a432. Each call traces back to a portable, censorship-resistant agent identity on Base L2. agentId 44544 on Basescan. Same on-chain track record; new verified agent handle that AI orchestrators can resolve.
/api/erc-8004-reputation endpoint. Read-only JSON aggregator exposing agentId, identity registry address, registration timestamp, and Basescan link. Cached 5 minutes. curl -s api.algovault.com/api/erc-8004-reputation | jq.

Upgrading from v1.19.x or earlier? The six original tools — get_trade_call, get_trade_signal, scan_funding_arb, get_market_regime, search_knowledge, chat_knowledge — keep their parameter shapes unchanged. v1.20.0 adds ONE new tool, scan_trade_calls (top-N market scanner), and additive output fields on the core tools (underlying_session, session_note, metrics.funding_by_venue) plus a structured INSUFFICIENT_CANDLES error — all backward-compatible. Because MCP clients (Claude Desktop, Cursor, Cline) cache tools/list at session start, reconnect (toggle the connector off/on) once to see scan_trade_calls; existing integrations keep working without changes.

Why AlgoVault

Most MCP trading servers give you raw data — prices, order books, candles. Your agent still has to figure out what to do with it.

AlgoVault is different. We give your agent one answer: a directional verdict with a confidence score, produced by a self-tuning quant ML model whose weights are calibrated from published trade outcomes. Every call is tracked, every outcome is measured, and the full track record is on-chain from day one. The agent itself is registered on ERC-8004.

What makes this not just another indicator wrapper:

Composite scoring, not single-indicator noise. Multiple orthogonal signals — momentum oscillators, trend structure, derivatives positioning, volume dynamics, open interest flow — fused into a single weighted verdict. Weights are calibrated from live market outcome data, not textbook defaults.
Regime-aware call generation. Calls are filtered through a market regime classifier before emission. The engine knows when to stay silent — a trend-following setup in a ranging market gets suppressed, not broadcast. (HOLD rate ~99%; we issue calls only when the edge is clear.)
Cross-venue intelligence. Full signal generation on 5 exchanges with native candle, OI, funding, and volume data per venue. 2 DEX venues (Aster, edgeX) in experimental shadow phase. Cross-venue funding arbitrage scanning across the major CEX — nobody else does multi-exchange derivatives analysis via MCP.
Published track record with every release. Every call is recorded with outcome prices at multiple horizons. PFE Win Rate, Profit Factor, Expected Value computed continuously. No cherry-picking, no survivorship bias. Anchored on-chain on Base L2 — we cannot rewrite history.
ERC-8004 verified agent. AlgoVault MCP holds a registered agentId on the Base L2 Identity Registry. The agent identity is portable and on-chain.
Drop-in for every major agent framework. First-party tutorials + runnable demos for LangChain, LlamaIndex, Microsoft Agent Framework, CrewAI. ≤5 minutes to first call. See the Framework integrations section.
Self-tuning model. Indicator weights are tuned from live outcome data by our Autonomous Optimization Engine. The model gets sharper with every signal.
Crypto + TradFi coverage. 760+ assets — standard crypto perps on all 5 CEX; TradFi perps (stocks, indices, commodities, FX); liquidity-filtered meme coins.

Try It in 30 Seconds

No code. No API key. No install.

Step 1. Open Claude → Settings → Integrations → Add custom connector

Step 2. Enter the name and URL:

Field	Value
Name	`Crypto Quant Signal`
URL	`https://api.algovault.com/mcp`

Add Connector

Step 3. Ask Claude anything:

"Get me a trade call for ETH on the 4h timeframe"

"Get me a trade call for BTC on Binance, 1h timeframe"

BTC trade call response — Binance, 1h

That's it. Your Claude now has a quant analyst built in.

Tools

`get_trade_call` _{(alias: get_trade_signal)}

Returns a composite BUY / SELL / HOLD verdict with confidence score for any supported asset on any of 5 supported exchanges — crypto perps, TradFi perps (stocks, indices, commodities, FX) on Binance / Bybit / Bitget / OKX / Hyperliquid, and liquidity-filtered meme coins.

Under the hood: a self-tuning quant ML model evaluates momentum, trend structure, derivatives sentiment, open interest dynamics, and volume conviction. Scores pass through regime-aware filters and adaptive post-processing gates before a final verdict is emitted. Only high-conviction calls are generated; the model stays silent when the edge is unclear.

Parameters:

coin (string, required): Asset symbol — e.g. "ETH", "BTC", "SOL", "GOLD", "TSLA", or any of 730+ supported assets
timeframe (string, default "15m"): all 11 timeframes — "1m", "3m", "5m", "15m", "30m", "1h", "2h", "4h", "8h", "12h", "1d"
exchange (string, default "BINANCE"): "BINANCE", "HL" (Hyperliquid), "BYBIT", "OKX", "BITGET". Asset availability varies per venue — pass exchange explicitly to target a specific venue. (DEX venues Aster + edgeX are in experimental shadow phase and not exposed via this parameter yet.)
includeReasoning (boolean, default true): Human-readable explanation of the call logic

Output: v1.10.0 sanitized shape — call direction, confidence (0–100), bucketed indicators (funding_rate / funding_24h_avg / funding_state / oi_change_pct / volume_24h / trend_persistence / breakout_pending), detected regime, sanitized reasoning prose, and _algovault metadata for downstream tool composability.

Responses also include optional closest_tradeable (on HOLD verdicts) and also_see (top-3 cross-asset leads), trimmed to {coin, timeframe, confidence} only — direction requires another get_trade_call invocation.

Example response:

{
  "call": "BUY",
  "confidence": 78,
  "price": 84250.50,
  "indicators": {
    "funding_rate": 0.0001,
    "funding_24h_avg": 0.00008,
    "funding_state": "NORMAL",
    "oi_change_pct": 2.4,
    "volume_24h": 2381602633,
    "trend_persistence": "HIGH",
    "breakout_pending": "INACTIVE"
  },
  "regime": "TRENDING_UP",
  "reasoning": "Trending regime, upward bias. Funding pressure mild. Volatility neither expanding nor compressed. Trend persistence elevated; momentum structure. Strong conviction from aligned signals.",
  "timestamp": 1712764800,
  "coin": "BTC",
  "timeframe": "1h",
  "also_see": [
    { "coin": "ETH", "timeframe": "1h", "confidence": 82 },
    { "coin": "SOL", "timeframe": "15m", "confidence": 73 }
  ],
  "_algovault": {
    "version": "1.20.0",
    "tool": "get_trade_call",
    "compatible_with": ["crypto-quant-risk-mcp", "crypto-quant-backtest-mcp"]
  }
}

`scan_funding_arb`

Scans cross-venue funding rate differentials across Hyperliquid, Binance, and Bybit. Normalizes hourly vs 8-hour rate conventions, computes basis-point spreads, ranks opportunities by composite score (spread magnitude, time urgency, funding conviction from 24h history). OKX and Bitget funding data is available via their respective adapters — arb scanning expansion is planned.

This is the only MCP server providing cross-venue funding arbitrage intelligence — long one exchange, short another, capture the spread.

Parameters:

minSpreadBps (number, default 5): Minimum spread in basis points to include
limit (number, default 10): Maximum results returned (free tier capped at 5)

Output includes: per-opportunity venue rates, optimal long/short direction, annualized spread percentage, and next funding timestamps.

`get_market_regime`

Classifies the current market environment: TRENDING_UP, TRENDING_DOWN, RANGING, or VOLATILE.

Combines directional strength measurement with ADX slope analysis (detecting trend strengthening vs exhaustion), volume-weighted pivot detection, ATR-adaptive funding thresholds, and cross-venue funding sentiment divergence. The regime classification directly informs how get_trade_call filters its output — agents can also use it independently for strategy selection and position sizing.

Parameters:

coin (string, required): Asset symbol
timeframe (string, default "4h"): Candle timeframe for analysis
exchange (string, default "BINANCE"): Exchange to analyze — same options as get_trade_call

Output includes: regime label, confidence score, underlying metrics (trend strength, volatility interpretation, price structure), cross-venue funding sentiment, and a plain-English strategy suggestion.

`scan_trade_calls`

Scans the top-N perpetual futures by open interest on a venue and returns composite BUY/SELL/HOLD verdicts for the whole set in one call — market-wide coverage for agents that would otherwise loop get_trade_call per coin.

Non-HOLD calls are ranked first by confidence; HOLDs are free and never consume quota (quota counts only the actionable BUY/SELL results returned).

Parameters:

topN (integer 1–100, default 20): How many top perps (by open interest) to scan
timeframe (string, default "15m"): Candle timeframe for the scan — 1m–1d
exchange (string, default "BINANCE"): Venue — BINANCE, HL, BYBIT, OKX, BITGET
minConfidence (number 0–100, optional): Drop non-HOLD calls below this confidence
includeHolds (boolean, default false): Append HOLD calls after the non-HOLD results
limit (integer 1–100, default 10): Max ranked calls returned (non-HOLD first)

Output includes: an array of { coin, timeframe, exchange, call, confidence, regime } entries, ranked non-HOLD first, plus the scan's venue and timeframe.

When You Hit the Free Limit

Free tier is 100 calls per calendar month. HOLD calls don't count against it — you only consume quota on BUY/SELL verdicts.

When the cap is reached, the next call's response includes:

Free tier limit reached (100/100 calls this month).
Upgrade to Starter ($9.99/mo) for 3,000 calls/mo,
or pay per call via x402.
→ https://api.algovault.com/signup?plan=starter

Two zero-friction upgrade paths:

Path	When to use	Friction
Starter $9.99/mo	Your agent runs on a known schedule (cron, hourly digests, daily scans).	Stripe checkout · API key delivered instantly · signup →
x402 micropayment	Your agent is autonomous and pays per call in USDC on Base. No signup.	Wallet · ~$0.01–0.05 per BUY/SELL · zero account state · x402.org

Pricing

Feature	Free	Starter ($9.99/mo)	Pro ($49/mo)	Enterprise ($299/mo)	x402 (per call)
Exchanges	All 5	All 5	All 5	All 5	All 5
Assets	All 760+	All 760+	All 760+	All 760+	All 760+
Asset classes	Crypto + TradFi	Crypto + TradFi	Crypto + TradFi	Crypto + TradFi	Crypto + TradFi
Timeframes	All 11	All 11	All 11	All 11	All 11
Funding arb results	Top 5	Unlimited	Unlimited	Unlimited	Unlimited
Track record	Full access	Full access	Full access	Full access	Full access
Monthly calls	100/mo	3,000/mo	15,000/mo	100,000/mo	Unlimited
Support	Community	Email	Priority	Dedicated	—
Price	$0	$9.99/mo	$49/mo	$299/mo	$0.01–0.05/call
HOLD calls	Free	Free	Free	Free	Free

* HOLD verdicts (engine says "don't trade") are always free across all tiers — no x402 charge, no quota deduction. We only get paid when we see a tradeable opportunity.

Subscriptions: Sign up at api.algovault.com/signup. Starter ($9.99/mo) unlocks 3,000 calls/mo. API key delivered instantly after checkout.

x402 micropayments: AI agents pay per HTTP call with USDC on Base — no signup, no API key, no billing. The payment receipt is the credential. See x402.org.

Integrations

End-to-end tutorials pairing AlgoVault with each major exchange's Agent Trade Kit. AlgoVault returns the analytics; the agent's risk policy decides what to execute. All demos run testnet/demo only — zero real-money risk.

#	Exchange	Tutorial	Demo	Mirror
01	Binance	`docs/integrations/binance.md`	`examples/binance/demo.mjs`	algovault.com/integrations/binance
02	OKX	`docs/integrations/okx.md`	`examples/okx/demo.mjs`	algovault.com/integrations/okx
03	Bybit	`docs/integrations/bybit.md`	`examples/bybit/demo.mjs`	algovault.com/integrations/bybit
04	Bitget	`docs/integrations/bitget.md`	`examples/bitget/demo.mjs`	algovault.com/integrations/bitget

Skills (20 ready-to-use Anthropic Agent Skills)

20 single-prompt wrappers over 1–3 AlgoVault tool calls — composite verdicts, regime gating, multi-timeframe consensus, funding-arb monitoring, and more.

claude plugin install AlgoVaultLabs/algovault-skills

Browse the full catalog at algovault.com/skills or github.com/AlgoVaultLabs/algovault-skills.

Show all 20 Skills

#	Slug	Name	Difficulty	Tools
01	`quick-btc-check`	Quick BTC Check	Beginner	`get_trade_call`
02	`portfolio-scanner`	Portfolio Scanner	Intermediate	`get_trade_call`
03	`regime-aware-trading`	Regime-Aware Trading	Intermediate	`get_market_regime`, `get_trade_call`
04	`funding-arb-monitor`	Funding Arb Monitor	Intermediate	`scan_funding_arb`
05	`full-3-tool-pipeline`	Full 3-Tool Pipeline	Advanced	`get_market_regime`, `get_trade_call`, `scan_funding_arb`
06	`multi-timeframe-confirmation`	Multi-Timeframe Confirmation	Advanced	`get_trade_call`
07	`tradfi-rotation`	TradFi Rotation	Advanced	`get_market_regime`, `get_trade_call`
08	`risk-gated-entry`	Risk-Gated Entry	Advanced	`get_market_regime`, `get_trade_call`
09	`funding-sentiment-dashboard`	Funding Sentiment Dashboard	Advanced	`get_market_regime`
10	`contrarian-meme-scanner`	Contrarian Meme Scanner	Advanced	`get_market_regime`, `get_trade_call`
11	`divergence-detector`	Divergence Detector	Advanced	`get_market_regime`, `get_trade_call`
12	`hourly-digest-bot`	Hourly Digest Bot	Advanced	`get_trade_call`, `get_market_regime`
13	`hedging-advisor`	Hedging Advisor	Advanced	`get_market_regime`, `get_trade_call`, `scan_funding_arb`
14	`volatility-breakout-watch`	Volatility Breakout Watch	Advanced	`get_market_regime`, `get_trade_call`
15	`cross-asset-correlation`	Cross-Asset Correlation	Advanced	`get_trade_call`
16	`funding-cash-and-carry`	Funding Cash-and-Carry	Advanced	`scan_funding_arb`, `get_trade_call`
17	`weekend-vs-weekday-patterns`	Weekend vs Weekday Patterns	Research	`get_trade_call`, `get_market_regime`
18	`agent-portfolio-rebalance`	Agent Portfolio Rebalance	Advanced	`get_market_regime`
19	`smart-dca-bot`	Smart DCA Bot	Advanced	`get_trade_call`
20	`multi-agent-war-room`	Multi-Agent War Room	Expert	`get_market_regime`, `get_trade_call`, `scan_funding_arb`

Performance Tracking & On-Chain Verification

Every call is tracked from emission to outcome. No exceptions.

What we measure:

Outcome prices at timeframe-appropriate evaluation windows
PFE Win Rate — did price move in the call direction at any point during the evaluation window
Expected Value — probability-weighted average return per call
Profit Factor — gross wins divided by gross losses
Peak Favorable Excursion (PFE) and Maximum Adverse Excursion (MAE)
Running statistics per asset, timeframe, and quality tier

HOLD calls are free. When the engine says "don't trade," you don't pay. Only BUY/SELL verdicts charge x402 or count against subscription quotas. Aligns incentives: we only get paid when we see a tradeable opportunity. Current HOLD rate ~99%.

On-Chain Verification

Every call is hashed (keccak256) at creation time and anchored on Base L2 via daily Merkle batches. The agent identity is registered on the ERC-8004 Identity Registry. The track record is tamper-proof — we cannot edit past calls.

Merkle Anchor Contract: 0x6485...0f81 (Base L2)
ERC-8004 Identity Registry: 0x8004...a432 — AlgoVault agentId 44544
Verify any call: api.algovault.com/api/verify-signal?signalId=<ID>
View all batches: api.algovault.com/api/merkle-batches
Agent metadata: api.algovault.com/api/erc-8004-reputation
Visual verification: algovault.com/verify
Live dashboard: algovault.com/track-record

For Developers

Remote endpoint (recommended)

https://api.algovault.com/mcp

Streamable HTTP transport. Compatible with any MCP-spec client — Claude Desktop, Claude Code, Cursor, Cline, custom agents. No SDK or wrapper library required; MCP is the API.

Local install via npx

npx -y crypto-quant-signal-mcp

Claude Desktop / Cursor config

{
  "mcpServers": {
    "crypto-quant-signal": {
      "command": "npx",
      "args": ["-y", "crypto-quant-signal-mcp"],
      "env": { "TRANSPORT": "stdio" }
    }
  }
}

Claude Code CLI

claude mcp add crypto-quant-signal https://api.algovault.com/mcp

npm install

npm install crypto-quant-signal-mcp

Self-hosting

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

Architecture

Agent / Claude Desktop / Claude Code / Cursor / Cline / any MCP client
  │
  ▼
api.algovault.com/mcp (Streamable HTTP)
  │
  ├─ x402 payment verification (USDC on Base)
  ├─ API key / subscription check
  ├─ Free tier fallback (100 calls/mo, all assets, all timeframes)
  │
  ▼
MCP Server (Express + @modelcontextprotocol/sdk)
  │
  ├─ Self-Tuning Quant ML Model
  │    ├─ Multi-factor indicator fusion
  │    ├─ Regime-aware signal filtering
  │    └─ Adaptive post-processing gates
  │
  ├─ Autonomous Optimization Engine (results published; mechanics confidential)
  │    └─ Closed-loop weight tuning from published outcomes
  │
  ├─ Asset Classification Engine
  │    ├─ 4-tier quality system (Blue Chip → Major Alt → TradFi → Meme)
  │    └─ Liquidity filter for meme/micro assets
  │
  ├─ Exchange Adapter Layer
  │    ├─ Binance USDT-M Futures (default)
  │    ├─ Hyperliquid (crypto + TradFi perps)
  │    ├─ Bybit Linear
  │    ├─ OKX Swap
  │    ├─ Bitget USDT-M
  │    ├─ Aster (DEX — experimental, shadow signal production)
  │    └─ edgeX (DEX — experimental, shadow signal production)
  │
  ├─ Performance Tracker
  │    └─ PostgreSQL (remote) / SQLite (local)
  │
  ├─ On-Chain Layer
  │    ├─ Merkle Anchor (daily) — Base L2 0x6485...0f81
  │    └─ ERC-8004 Identity Registry — Base L2 0x8004...a432
  │
  └─ Exchange Public APIs (free, no auth — all 5 CEX + 2 experimental DEX)

Suite Composability

Every tool output includes an _algovault metadata block declaring version and compatible downstream tools:

This tool	Feeds into (Phase 2+)
`get_trade_call`	`crypto-quant-risk-mcp` (position sizing) · `crypto-quant-backtest-mcp` (validation)
`scan_funding_arb`	`crypto-quant-execution-mcp` (optimal entry/exit) · `crypto-quant-risk-mcp` (exposure)
`get_market_regime`	`crypto-quant-risk-mcp` (regime-aware sizing) · `crypto-quant-backtest-mcp` (filtered backtests)

Schemas are designed for composability. All tools share consistent timestamp, coin, and _algovault fields — downstream tools accept these objects directly as input.

Privacy

Local mode: Zero telemetry. No data sent to AlgoVault servers. Call history stored on your machine only.

Remote mode: Request metadata logged for analytics (IP hashed, never stored raw). See privacy policy.

License

MIT

Disclaimer: AlgoVault provides directional entry interpretation for AI agents. Exit timing is determined by your agent or strategy. This is not financial advice. Past performance does not guarantee future results.

Built by AlgoVault Labs
algovault.com · how it works · track-record · MCP endpoint · Telegram bot

Keywords

mcp trading crypto trade-call signals hyperliquid binance bybit okx bitget funding-rate quant defi arbitrage technical-analysis model-context-protocol ai-agent x402