CRYLO^{^®} Crypto Portfolio Benchmarks

Public, reproducible benchmarks comparing CRYLO’s guardrailed crypto portfolios.

Home
CRYLO® Research on AI Managed Crypto Portfolios
CRYLO® Crypto Portfolio Benchmarks

Digital Asset Portfolio Benchmarks and Performance Metrics

Our Digital Asset Portfolio Benchmarks are the entry point to CRYLO’s public evaluation suites. Each suite publishes a leaderboard, methods summary, downloadable results (CSV), and a minimal script to regenerate key figures. The goal: make it easy to scrutinize outcomes—without revealing proprietary parameters.

Risk, Volatility and Drawdown Benchmarks for Crypto

Leaderboards for specific tasks and time windows
Baselines vs. guardrailed strategies (risk-first)
Methods, metrics, and limitations
Links to datasets (with DOIs) and code
“How to reproduce” instructions for each suite

How we evaluate crypto portfolio strategies

Guardrails v1 — Risk-First Portfolio Control

Summary of core risk metrics (Max Drawdown, ES, Sharpe/Sortino) across rolling out-of-sample windows. Includes stress injections (gaps, liquidity crunch) to test envelope bounds.

Explore the suite →

Download leaderboard (CSV) →

Figures →

DOI →

Baselines & Comparators

Reference implementations for BTC buy-and-hold, equal-weight top-N, and a public risk-parity proxy. These provide context for drawdown, volatility, and turnover versus our guardrailed approach.

See methods and baselines →

How to Read the Leaderboards

Key metrics

AR / AV: Annualized return and volatility
Sharpe / Sortino: Risk-adjusted performance (rf≈0)
Max Drawdown / ES(95): Tail risk and worst-case paths
Turnover / Shortfall: Implementation realism under costs and slippage

Slices & windows

We report rolling out-of-sample windows (e.g., 2019–2021, 2021–2023, 2024-2025) to avoid look-ahead. Each slice includes costs, exchange outages, funding, and depeg filters where applicable.

Methods at a Glance

What we disclose

Mechanisms and evaluation protocol: universe eligibility, rebalance cadence, guardrail triggers (vol targeting, drawdown circuits), stress scenarios, and metric definitions.

What we withhold

Thresholds, hyperparameters, feature recipes, and model weights remain confidential. Benchmarks focus on outcomes and guardrail effectiveness.

Reproduce the Results

One-command run

Each suite provides a short repro.sh (or notebook) that regenerates figures from a tiny sample dataset. Full datasets are hosted via DOI on Zenodo.

Steps:

Create environment (conda or pip)
Run repro.sh to render figures into /figures
Compare your outputs with our published CSV and images

Datasets & Code

Datasets

Public Eval Dataset — versioned, DOI-archived; schema and checksums provided.

Dataset Link →

DOI →

Code

Evaluation & Baselines — minimal, deterministic scripts to reproduce leaderboard tables and figures.

GitHub repository →

GitHub “Cite this repository” →

Citation & Versions

Cite the suite

Isler, M., & CRYLO Research. (2025). CRYLO Research: Methods & Evaluation Kit (Version v1.0.0) [Computer software]. DOI: (https://doi.org/10.5281/zenodo.17454874).

Changelog

v1.0.0 — Initial benchmark release (methods, CSVs, figures)
v1.1.0 — Dataset DOI + expanded slices
v1.1.1 — Documentation fixes (no metric changes)

Limitations & Responsible Use

Known constraints

Exchange data quality varies; filters applied, residual noise possible
Synthetic stress scenarios simplify liquidity dynamics
Results are sensitive to asset eligibility and costs

Disclaimer

This benchmark is for research and evaluation. It is not investment advice.

More Releases about Sharp Ratio and Max Drawdowns for Crypto Portfolios

Discover more about our research efforts on AI & ML for CRYLO's digital asset portfolios.

AI methods for crypto portfolio management

A neutral, citable description of how our portfolios are constructed and guarded—mechanisms, evaluation protocol, and limitations without revealing proprietary parameters. Explore the Methods →

Datasets for digital asset portfolios

ersioned, DOI-archived datasets with schema, checksums, and a simple “reproduce” guide for our public evaluation tasks. Browse Datasets →

Model-Cards & ethical thoughts on AI

Plain-English summaries of model families: intended use, inputs/outputs, evaluation notes, and ethical considerations. Read Model-Cards →

Notes on risk, drawdown, and volatility research

Short, timely research notes—ablations, stress tests, and replication updates that don’t require a full paper. See Research Notes →

For Investors

Our research on benchmarks for crypto portfolios is not just theory; the same AI models, guardrails, and performance results described above power the crypto portfolios and digital asset wealth management solutions you use with CRYLO^®. When you invest and buy digital assets through CRYLO^®, you benefit from systematic allocation, volatility targeting, and drawdown limits designed to create more stable outcomes than ad hoc coin picking or manual trading. Use the services below to turn this research into a concrete portfolio aligned with your risk profile.

AI crypto portfolio performance ->

Asset Management with Crypto ->

Investor Relations ->

More Insights

As we build a digital asset management platform fully powered by AI and ML, we report monthly on the latest topics in the financial world and news about crypto.

Frequently Asked Questions

You probably have many questions related to our benchmarks. Therefore, we answer the most important ones below.

To let practitioners, journalists, and regulators verify outcomes independently, using transparent baselines and reproducible code.

Yes—community submissions are welcome if they follow the same data slices, metrics, and leakage checks. Submission guidelines are provided in each suite.

No. We publish mechanisms and outcomes; proprietary thresholds and weights remain private.

CRYLO® Crypto Portfolio Benchmarks