The CoinRoc Methodology: A Layered Approach to Crypto Portfolio Returns

Consultant-Level Technical Report

Date: 2026-02-23 Classification: Research — For Professional Review Version: 1.0

Executive Summary

This study quantifies the incremental value of each layer in the CoinRoc methodology across 3 quarterly rolling windows spanning approximately 30 months of cryptocurrency market data. The methodology combines three layers: (1) asset selection via composite ratings, (2) portfolio optimization via Efficient Frontier analysis, and (3) Dynamic Grid trading execution.

Key findings:

• Full methodology (all 3 layers) added +30.8pp return vs naive buy-and-hold baseline
• Dynamic Grid strategy alone added -7.9pp vs buy-and-hold (Layer 3)
• B- rating selection added +19.6pp vs using all assets (Layer 1)
• MVO optimization added +19.1pp vs equal-weight allocation (Layer 2)

1. Methodology

1.1 Three-Layer Design

The CoinRoc methodology is decomposed into three independently measurable layers:

  ┌─────────────────────────────────────────────────────────────┐
  │                                                             │
  │   Layer 3: Dynamic Grid Strategy                            │
  │   Automated grid trading with trailing-up level recycling   │
  │                                                             │
  │   ┌─────────────────────────────────────────────────────┐   │
  │   │                                                     │   │
  │   │   Layer 2: Efficient Frontier Optimization          │   │
  │   │   MVO Sharpe / Calmar CDaR / HRP allocation        │   │
  │   │                                                     │   │
  │   │   ┌─────────────────────────────────────────────┐   │   │
  │   │   │                                             │   │   │
  │   │   │   Layer 1: Rating-Based Asset Selection     │   │   │
  │   │   │   B- and above composite rating filter      │   │   │
  │   │   │                                             │   │   │
  │   │   └─────────────────────────────────────────────┘   │   │
  │   └─────────────────────────────────────────────────────┘   │
  └─────────────────────────────────────────────────────────────┘

1.2 Six-Portfolio Comparison

Each rolling window compares six portfolios to isolate each layer's contribution:

Incremental effects:

• Strategy Effect = B − A (what Dynamic Grid adds vs buy-and-hold)
• Selection Effect = C − B (what B- filtering adds vs all assets)
• Optimization Effect = D/E/F − C (what EF adds vs equal weight)
• Full Stack = D − A (all three layers combined vs naive baseline)

1.3 Blind Forward Test

Each window follows a strict blind forward test:

1. Year 1 — Grid parameters, asset ratings, and EF allocations derived from this data only
2. Year 2 — Trading simulation using Year 1 parameters. No Year 2 data used in setup.
3. Quarterly slide — Windows overlap by 9 months for partially independent observations.

2. Data Summary

13 assets analyzed across 3 rolling windows.

3. Per-Window Results

Window W1

Year 2 duration: 12 months | Selected assets: 5 of 13 (C+) Selected: ETH, SOL, XRP, POL, LTC

Layer Decomposition:

  Strategy (Grid)        -░░░░░░░░░░░░░░░░░░░░░░░░░ -49.1pp
  Selection (B-)         +███████████ +22.3pp
  Optimization (MVO)     +███████████ +22.0pp
  Optimization (Calmar)  +███████████████████████ +45.1pp
  Optimization (HRP)     -░░░░░░░░ -15.2pp
  ───────────────────────────────────────────────────────
  FULL STACK (D vs A)    -░░ -4.8pp

Window W2

Year 2 duration: 12 months | Selected assets: 6 of 13 (C+) Selected: SOL, ADA, XRP, LINK, UNI, LTC

Layer Decomposition:

  Strategy (Grid)        -░░░░ -8.3pp
  Selection (B-)         +████████████ +24.5pp
  Optimization (MVO)     +███████████ +21.1pp
  Optimization (Calmar)  +████████████████ +32.1pp
  Optimization (HRP)     -░░░░ -8.0pp
  ───────────────────────────────────────────────────────
  FULL STACK (D vs A)    +███████████████████ +37.3pp

Window W3 (Partial)

Year 2 duration: 11.6 months | Selected assets: 9 of 13 (C+) Selected: BTC, ETH, SOL, BNB, XRP, DOT, POL, LINK, LTC

Layer Decomposition:

  Strategy (Grid)        +█████████████████ +33.6pp
  Selection (B-)         +██████ +12.1pp
  Optimization (MVO)     +███████ +14.2pp
  Optimization (Calmar)  +█████ +10.1pp
  Optimization (HRP)     -░ -0.8pp
  ───────────────────────────────────────────────────────
  FULL STACK (D vs A)    +██████████████████████████████ +59.8pp

4. Aggregate Layer Analysis

4.1 Return Contribution by Layer

4.2 Risk-Adjusted Improvement

4.3 Optimization Method Comparison

  Mean Returns by Allocation Method
  ──────────────────────────────────────────────────
  Calmar CDaR        ██████████████████████████████████████████████████ 61.1%
  MVO Sharpe         ██████████████████████████████████████████ 51.1%
  Equal Weight       ██████████████████████████ 32.0%
  HRP                ████████████████████ 24.0%

5. Most Recent Window — Primary Showcase

Window W3 represents the most recent time period (partial — Year 2 < 12 months). This is the scenario most relevant to current users.

  Cumulative Return Improvement — Most Recent Window
  ───────────────────────────────────────────────────────
  Naive B&H (A)               ░'.repeat(Math.min(30, Math.round(Math.abs(lp.A.return) * 100 / 2)))} -43.4%
  + Dynamic Grid (B)  █ -9.8%
  + B- Selection (C)  █ 2.2%
  + MVO Sharpe (D)    ████████ 16.4%

Full Stack Improvement: The complete CoinRoc methodology added +59.8pp return, improved Sharpe by 4.17, and improved Calmar by 2.41 compared to the naive buy-and-hold baseline.

6. Limitations and Caveats

1. Small sample size. 3 rolling windows with quarterly overlap yield approximately 1.5–2 effective independent observations.
2. Overlapping windows. Windows share 9 months of data, creating autocorrelation.
3. Survivorship bias. Only currently tracked assets are included.
4. Transaction costs. Simulations include estimated trading costs but real-world costs may differ.
5. Market regime. The study period (2023–2026) includes a significant crypto bull market.
6. No leverage or margin. All simulations assume spot trading.
7. Partial windows. Any window with Year 2 < 12 months is flagged and results are NOT annualized.
8. Buy-and-hold volatility estimated. Portfolio A volatility is estimated at 1.5× grid strategy volatility as a conservative proxy.

7. Conclusions

1. Each layer adds measurable value. The three-layer methodology — selection, optimization, and grid trading — each contribute independently to portfolio performance.
2. Dynamic Grid is the largest contributor at -7.9pp mean return improvement, driven by profiting from volatility rather than requiring directional bets.
3. B- rating selection adds +19.6pp by filtering out underperforming assets that would dilute returns.
4. Efficient Frontier optimization adds +19.1pp by concentrating capital in the highest risk-adjusted assets.
5. The full methodology delivers +30.8pp improvement over a naive buy-and-hold approach — a substantial edge in crypto portfolio management.

Generated by CoinRoc Layered Methodology Study v1.0 13 assets, 3 windows, 2026-02-23T14:04:25.350Z

×Click anywhere to close