Rating System Effectiveness Research

Interactive visualization of how our multi-factor rating system improves grid trading profitability by 37-66% across 480 backtests

Published: February 5, 2026 Open Standalone Version

0.886

Correlation with Returns

+37.6%

Return Improvement

78%

High-Rated Win Rate

34%

Low-Rated Win Rate

Abstract

This research validates the effectiveness of our multi-factor cryptocurrency rating system for grid trading. Through 480 backtests across 20 cryptocurrencies and 4 exchanges, we demonstrate that high-rated assets (A/B grades) achieve 37.6% average returns compared to 0% for low-rated assets (C/D grades). The rating system achieved a 0.886 correlation with actual grid trading returns, with the strongest predictive components being Grid Performance (30%), Sortino Ratio (15%), and Calmar Ratio (15%).

Loading interactive charts...

1 Correlation Improvement

Key Insight: Iterative optimization improved correlation from 0.700 to 0.886 (+26.6%)

2 Portfolio Performance

Key Insight: High-rated cryptos outperform by 37.6% with 2.3x better Sharpe ratio

3 Exchange Success Rates

Key Insight: Binance.US (85%) vs Coinbase (45%) - 40% swing in success rate!

4 Rating Component Weights

Key Insight: Grid Performance (30%) and risk metrics (30%) drive most of the rating

5 Component Importance (Ablation)

Key Insight: Grid Performance is most critical - removing it drops correlation by 0.089

6 Exchange Fee Breakdown

Key Insight: Coinbase costs 4.4x more than Binance.US per round-trip trade

7 Crypto Returns on Binance.US (All 20)

Excellent (>10%)

Good (5-10%)

Marginal (0-5%)

Loss (<0%)

Key Insight: Top 6 performers (LTC to ETH) are all A or B rated. Bottom 3 (ALGO, VET, NEAR) are D rated.

8 1% Spacing Disaster

Warning: NEVER use 1% spacing. Coinbase: 0/20 profitable, -11.3% average!

9 Rolling Test Results (19 Windows)

Key Insight: LTC wins 89% of test windows vs ALGO at 21%. Consistency validates ratings.

10 Profitability Heatmap: Crypto x Exchange

>15%

5-15%

0-5%

-5% to 0%

<-5%

Key Insight: Green clusters in top-left (good cryptos + low-fee exchanges). Avoid bottom-right red zone.

Methodology

Blind Forward Testing

All backtests used a blind forward methodology: Year 1 data (days 1-360) was used to calculate grid parameters, while Year 2 data (days 361-720) was used exclusively for trading simulation. This prevents lookahead bias and provides realistic performance expectations.

Rolling Window Validation

Additionally, 19 rolling 6-month test windows were used to validate consistency. Assets with high win rates across multiple windows (like LTC at 89%) demonstrate robust performance rather than lucky timing.

Exchange Coverage

Tests covered 4 major US exchanges: Binance.US, Kraken, Gemini, and Coinbase. Each exchange's actual fee structure (maker, taker, slippage) was incorporated to ensure realistic results.

Data Quality

All price data was sourced from exchange APIs with 4-hour candles over 2+ years. Assets with insufficient data history were excluded from the analysis.

Key Conclusions

1. Rating system works: 0.886 correlation between ratings and actual returns validates the multi-factor approach.
2. Exchange choice is critical: Binance.US achieves 85% success rate vs Coinbase at 45% - a 40 percentage point difference.
3. Never use 1% spacing: Tight spacing loses money on all exchanges due to fee accumulation.
4. Grid Performance is key: The single most important rating component, removing it drops correlation by 0.089.
5. High-rated = high returns: A/B rated assets average 37.6% returns with 78% win rate vs 0%/34% for C/D rated.

Related Research

Grid Trading Profitability Analysis

Comprehensive analysis of profitability across 480 backtests, 20 cryptos, 4 exchanges

View All Research

Browse our complete library of research papers and validation studies