Docs / ML Confidence Tiers

💎 ML Confidence Tiers

Understanding Diamond, Gold, and Verified labels and how to use them effectively

VETTR uses machine learning to predict which tokens are likely to hit 2x. Our AI model has been trained on 4,400+ historical tokens and uses 51 proprietary features to assign a confidence score to each signal.

Based on this score, some signals receive a tier label indicating our AI's confidence level. These labels help you focus on the highest-probability opportunities.

Key Insight: Labeled picks have a 77.4% success rate vs 30.6% for unlabeled picks. That's a +47 percentage point improvement from the ML model.

The Three Tiers

💎

Diamond Tier

Score Range: 0.85 - 1.00 (Top tier signals)

90.8%
Success Rate
59/65
Hit 2x
~7%
Of Signals

What It Means

Diamond picks represent the highest confidence our AI can assign. These tokens have all the right signals: strong creator history, smart wallet involvement, solid fundamentals, and low rug risk.

When You See Diamond

  • Very high probability of hitting 2x (90.8%)
  • ✅ Excellent risk/reward ratio
  • ✅ Prioritize these over other signals
  • ⚠️ Still check rug risk and liquidity yourself

Real Examples (Recent Wins)

💎 $LION — Hit 109x
ML Score: 0.85
Smart Wallet: Elite tier detected
Creator: 12/15 successful (80%)
Result: 109x peak

Gold Tier

Score Range: 0.60 - 0.84 (High confidence signals)

80.2%
Success Rate
77/96
Hit 2x
~11%
Of Signals

What It Means

Gold picks have strong fundamentals but may be missing one or two ideal characteristics. Still well above baseline success rates.

When You See Gold

  • High probability of hitting 2x (80.2%)
  • ✅ Good risk/reward (3 in 4 succeed)
  • ✅ Secondary priority after Diamond
  • ⚠️ Review creator history and rug risk carefully

Real Examples

⭐ $SIGN — Hit 16.5x
ML Score: 0.67
Creator: New deployer (first launch)
Holder Quality: Top 10 concentrated 18%
Result: 16.5x peak

Verified Tier

Score Range: 0.50 - 0.59 (Above baseline signals)

68.6%
Success Rate
116/169
Hit 2x
~19%
Of Signals

What It Means

Verified picks meet the minimum threshold for ML labeling. Better than random but not as strong as higher tiers.

When You See Verified

  • Above-average probability (68.6% vs 37.7% baseline)
  • ⚠️ More risk than Diamond/Gold (2 in 5 may fail)
  • ⚠️ Lowest priority of labeled signals
  • ⚠️ Do extra due diligence

Real Examples

✅ $littleguy — Hit 46.5x
ML Score: 0.54
Creator: 3/8 successful (38%)
Volume: High initial activity
Result: 46.5x peak

Unlabeled Signals

Most signals (88%) don't receive a tier label because their ML score is below 0.50. These signals:

Important: Even unlabeled signals beat random picks. VETTR's filtering alone improves your odds. But for best results, focus on labeled picks.

Comparison Table

Tier Score Range Success Rate Frequency Risk Level
💎 Diamond 0.85 - 1.00 90.8% ~7% of signals Very Low
🥇 Gold 0.60 - 0.84 80.2% ~11% of signals Low
✅ Verified 0.50 - 0.59 68.6% ~19% of signals Medium
(No label) < 0.50 37.7% ~63% of signals High

How to Use the Tiers

1. Prioritize by Tier

Focus your attention and capital on higher-tier signals:

2. Combine with Other Signals

Tiers don't replace due diligence. Always check:

3. Set Realistic Expectations

Even Diamond picks can fail (4.7% failure rate). Practice good risk management:

Pro Tip: Focus on Diamond picks exclusively if you're risk-averse. Gold and Verified tiers are for more aggressive traders who want higher volume of opportunities.

Understanding the ML Model

What the Model Evaluates (High-Level)

Our proprietary model analyzes 51 features across these categories:

Specific features and weights are proprietary to maintain our competitive advantage.

Model Performance

What "Success" Means

A token is considered "successful" if it reaches 2.0x from our entry signal at any point. This is our baseline threshold because:

Note: Many successful tokens go far beyond 2x (10x, 50x, 100x+). The model predicts the 2x threshold, not peak performance.

Disclaimer: Past performance does not guarantee future results. The model's accuracy may vary based on market conditions, new scam techniques, and other factors. Always DYOR.

Summary