Nexstone

NexstoneCapital

Quantitative trading infrastructure.

Open source. Statistically validated.

Flagship

RIFT

Scout & Trade manually or let algorithms run.

From idea to live trading.

>

Quick Start

# One command. Python engine and CLI in a single shot.

$ brew install Nexstone/tap/rift

Works on macOS and Linux. Pulls Python 3.13, Node 20, the engine, and the CLI. View on GitHub

Scout

One command ranks coins by funding, momentum, volume, positioning. Shows entry, stop loss, and target.

Spot & Perps

Buy and sell spot tokens — or trade perpetual futures with leverage. One tool for both. Server-side stops.

Algo Trading

Walk-forward validated strategies with auto-discovery. Health monitoring, auto-pause, position management.

Portfolio Manager

Multi-strategy supervisor. Risk coordination, correlation guard, scheduling, Slack alerts, health-based rotation.

Research Lab

Backtest, walk-forward, Monte Carlo, multi-pair, feature importance, vol forecast, health check, tearsheet.

Institutional Analytics

Execution analytics, PnL attribution, Value at Risk, compliance exports, strategy versioning, and REST API.

Institutional-Grade Validation

One command validates your strategy the way quant funds do. No shortcuts.

BacktestReturn, Sharpe, drawdown

2.5 years of real data

Walk-ForwardOOS return

27 train/test windows

Monte CarloProfit probability

10,000 shuffled sequences

Multi-PairGeneralization check

Cross-coin validation

Feature ScanWhat drives your edge

XGBoost feature ranking

Health CheckScore 0-100

Edge decay detection

GradeA / B / C / D / F

Pass or fail

One command: rift research your_strategy --pair SUI

Your Journey

From idea to live trading

9 steps
Create
rift workbench-create
Test
rift quick-test
Validate
rift research
Scan
rift scan
Simulate
rift simulate --duration 4h
Trade
rift algo

Every step outputs structured JSON. AI agents can run the entire pipeline autonomously via MCP.

Hyperliquid Native

Born on Hyperliquid

Not adapted. Not integrated. Built from the first line of code for Hyperliquid's architecture. Predicted funding entries before settlement. Stop losses that live on the exchange — not your machine. 2.5 years of native data for 50 coins, bundled and ready.

300+ crypto pairs74 TradFi perps2.5 years of dataServer-side stopsZero setup

Security

Your keys. Your machine. Your control.

RIFT runs locally. Nothing is sent to our servers — because there are no servers.

Keys never leave your machine

Your private key is stored locally. It's read from an environment variable at runtime — never passed as a CLI argument, never logged, never transmitted.

Stop losses survive crashes

The stop lives on Hyperliquid's servers — not yours. If RIFT crashes, your computer dies, or your internet goes down, the stop loss still executes.

No cloud dependency

There's no cloud account, no login, no telemetry. The only network calls are to Hyperliquid's public API. You can audit every line — it's open source.

Full audit trail

Every trade is logged with order IDs, fill prices, timestamps, and wallet addresses. Export compliance audits at any time. Every order is verifiable on-chain.

Drawdown Kill Switch

50% max

Health Auto-Pause

Grade D/F

Position Limits

Per-strategy USD cap

Portfolio Risk Gate

Net/gross limits

Comparison

How RIFT compares

Honest feature comparison.

Hyperliquid Native

RIFT

QC

FT

Market Scanner (Scout)

RIFT

QC

FT

No-Code Strategy Builder

RIFT

QC

FT

AI Agent Integration (MCP)

RIFT

QC

FT

Portfolio Supervisor

RIFT

QC

FT

Health Monitoring + Auto-Pause

RIFT

QC

FT

TCA + PnL Attribution

RIFT

QC

FT

VaR + Compliance Audit

RIFT

QC

FT

REST API

RIFT

QC

FT

Kelly Criterion Sizing

RIFT

QC

FT

Walk-Forward Validation

RIFT

QC

FT

Free & Open Source

RIFT

QC

FT

QC: $60-$1,080/moFT: Free, no HLRIFT: Free

AI Integration

AI-Native Trading

64 tools via Model Context Protocol. Claude, Cursor, or any AI agent can discover edge, build strategies, validate with walk-forward, execute trades, manage positions, and learn from past experiments — autonomously.

“Scout the market, trade the best opportunity, monitor it, and send me the TCA report when it closes.” — One prompt.

// MCP config

{

"rift": {

"command": "rift",

"args": ["serve"]

}

}

Audience

Who We Build For

Quant Researchers

Walk-forward, Monte Carlo, XGBoost, GARCH, VaR

Active Traders

Scout scanner, one-click trades, Kelly sizing, trailing stops

AI Developers

64-tool MCP server for autonomous agents

Strategy Builders

No-code workbench, 56 indicators, guided UI

Our Story

Why we built this

$300/mo for tools that don't support Hyperliquid.

Backtests that don't match live results.

Basic features locked behind paywalls.

We built what we wanted — and made it free.

Here's what real returns look like.

Alpha (edge)
0%
Funding
0%
Market (beta)
0%
Execution
0%

Most tools show you one number.
RIFT shows you where it came from.

Support

Common Questions

Install RIFT, run 'rift guide' for the 9-step journey, or 'rift scout' to see ranked opportunities in under 30 seconds. No API keys, no data downloads, no configuration. 2.5 years of data for 300+ pairs is bundled — your first backtest runs immediately.

Yes. RIFT is open source — research, backtesting, simulation, and analysis are free with no limits. Live perps trading has a 0.03% builder fee ($0.30 on a $1,000 trade). Spot trading has a 1% fee on sells. That's how we sustain development. No subscriptions, no tiers, no paywalls.

A small builder fee on live trades — 0.03% on perps, 1% on spot sells. That's it. We don't sell your data, we don't run a cloud service, and we don't lock features behind upgrades. The entire codebase is open source and auditable on GitHub.

No. Scout scans the market and tells you what looks interesting. Pick an opportunity, execute with one command, and monitor it live. For hands-off trading, build strategies with the workbench or let an AI agent create them via MCP.

No. The Strategy Workbench builds strategies from 56 indicators — including cross-asset, multi-timeframe, and adaptive — using guided menus. Advanced users can write Python too.

Building for one exchange deeply is better than supporting ten poorly. Every feature — predicted funding entries, server-side stop losses, native data, hourly settlement timing — is designed around Hyperliquid's specific architecture. Generic tools can't do this.

Yes. Trading perpetual futures carries substantial risk. RIFT includes multiple safety layers — server-side stop losses, drawdown kill switches, health monitoring that auto-pauses decaying strategies, and portfolio risk limits — but no software eliminates market risk. Never trade more than you can afford to lose.

QuantConnect is powerful but expensive ($60-$1,080/mo), built for equities first, and doesn't support Hyperliquid. It has no market scanner, no health monitoring, no portfolio supervisor, and no AI agent integration. RIFT gives you institutional-grade features for free on the exchange you actually trade.

Freqtrade is free and open source — we respect that. But it's a generic bot framework that requires you to build everything yourself. No Hyperliquid support, no predicted funding, no server-side stops, no VaR, no TCA, no research pipeline. RIFT ships with walk-forward validated strategies, 2.5 years of data, TradFi perps, and a 64-tool AI server out of the box.

Stop Waiting. Start Trading.

RIFT is on Homebrew, PyPI, and npm.

Install in one command. No signup. Free forever.

Use RIFT

Building the hosted Terminal next. Follow on X for early access.