ScreenerBot
DocsAssistantExamples

AI Analysis Examples

Real-world examples of how ScreenerBot's AI analysis works in practice. Learn from actual scenarios to understand AI filtering, entry analysis, exit suggestions, and scam detection.

AI Filtering Examples

Example 1: Obvious Pump & Dump Scam

Token: "MOON ROCKET 🚀🌙💎"

Address: 8kJd...mN9x

SCAM DETECTED

Liquidity

$45,000

24h Volume

$850,000

Holders

1,250

Top 10 Holdings

92%

Numerical Filter Result:

PASS - Liquidity > $10k ✓, Volume > $100k ✓, Holders > 100 ✓

AI Analysis (GPT-4o-mini):

Recommendation: FAIL (Confidence: 0.96)

Reasoning: This token exhibits multiple high-risk indicators:

  • Suspicious name with emoji spam ("MOON ROCKET 🚀🌙💎") - classic pump signal
  • Extreme holder concentration (92% in top 10) - massive rug pull risk
  • Volume-to-liquidity ratio of 18.9x is abnormally high - likely wash trading
  • No metadata: missing description, website, or social links
  • 1,250 holders but 92% concentrated suggests fake distribution

Verdict: High probability pump & dump. Avoid.

Action Taken:

  • • Token marked as FAIL and rejected from trading consideration
  • • Auto-blacklisted (confidence 0.96 > threshold 0.8)
  • • Event logged to dashboard for review

Outcome: AI prevented a likely rug pull. Despite passing numerical filters, the token showed clear scam patterns that AI correctly identified.

Example 2: Legitimate New Project

Token: "SolanaTools"

Address: 3mP7...qR2k

PASSED

Liquidity

$125,000

24h Volume

$180,000

Holders

850

Top 10 Holdings

38%

AI Analysis (Claude-3.5-Sonnet):

Recommendation: PASS (Confidence: 0.82)

Positive Indicators:

  • Professional metadata with detailed description and roadmap
  • Healthy holder distribution (38% in top 10) shows good decentralization
  • Reasonable volume-to-liquidity ratio (1.4x) suggests organic trading
  • Valid social links: Twitter active, Discord community, GitHub repo
  • Verified website with working product demo
  • RugCheck score: 8.5/10 (Low risk)

Risk Factors:

  • Relatively new token (3 days old) - limited history
  • Liquidity could be higher for market cap

Verdict: Appears legitimate. Proceed with caution and proper risk management.

Action Taken:

  • • Token marked as PASS and forwarded to trading engine
  • • Entry analysis scheduled for next price check
  • • Added to watchlist for monitoring

Example 3: Borderline Case (Uncertain)

Token: "DOGE2.0"

Address: 7nQ4...pL8m

UNCERTAIN

Liquidity

$65,000

24h Volume

$320,000

Holders

2,400

Top 10 Holdings

62%

AI Analysis (Groq llama-3.1-70b):

Recommendation: UNCERTAIN (Confidence: 0.55)

Mixed Signals:

  • ✅ Name derivative of established meme (DOGE) - could be legitimate tribute
  • ⚠️ High volume-to-liquidity ratio (4.9x) - concerning but not extreme
  • ⚠️ Moderate holder concentration (62%) - not great, not terrible
  • ✅ Complete metadata with meme-appropriate branding
  • ❌ Social links present but low engagement (<500 Twitter followers)
  • ⚠️ Large holder count (2,400) but distribution skewed

Verdict: Insufficient data for confident recommendation. Could be community meme or coordinated pump. Requires human judgment.

Action Taken (Advisory Mode):

  • • Token forwarded to trading engine with UNCERTAIN flag
  • • Dashboard shows AI uncertainty warning
  • • Trader can manually review and decide
  • • Not auto-blacklisted (confidence 0.55 < threshold 0.8)

Note: UNCERTAIN results benefit from manual review. In enforcement mode (filtering_enforce = true), these would be rejected. In advisory mode, they proceed with a warning.

Entry Analysis Examples

Example 4: Good Entry Opportunity

Token: "SolProtocol" (SOL/USDC)

Current Price

$0.245

24h Change

+8.2%

Volume Trend

Rising

AI Entry Analysis (DeepSeek):

Recommendation: ENTER (Confidence: 0.78)

Favorable Conditions:

  • Price consolidating after recent dip - potential bounce setup
  • Volume increasing 45% over last 4 hours - building momentum
  • RSI at 42 - not overbought, room to move up
  • Buyer/seller ratio improving (60/40)
  • Liquidity stable at $180k - no major removes recently

Suggested Entry: Current price with tight stop-loss at -8% (below recent low).

Result: Bot entered position at $0.245. Token moved to $0.289 (+18%) before hitting first take-profit level. AI correctly identified favorable entry timing.

Example 5: Bad Entry (FOMO)

Token: "HYPE" (SOL/USDC)

Current Price

$1.85

24h Change

+285%

1h Change

+42%

AI Entry Analysis (GPT-4o-mini):

Recommendation: AVOID (Confidence: 0.88)

Warning Signs:

  • Extreme parabolic move (+285% in 24h) - unsustainable
  • RSI at 92 - severely overbought, correction imminent
  • Volume spiking but buyer exhaustion pattern emerging
  • Large sells starting to appear on chain (whales exiting)
  • Entry at current level = high risk of being exit liquidity

Verdict: Classic FOMO scenario. Wait for retracement before considering entry.

Result: AI prevented entry. Token peaked at $1.92 (+3.8%) then crashed to $0.58 (-68%) within 6 hours. Avoiding this trade saved significant loss.

Exit Suggestion Examples

Example 6: Smart Exit Timing

Position: "GameFi Token"

Entry Price

$0.42

Current Price

$1.18

Unrealized P/L

+181%

AI Exit Analysis (Anthropic Claude):

Recommendation: TAKE PARTIAL PROFIT (Confidence: 0.84)

Momentum Analysis:

  • Strong uptrend continues but showing early divergence signs
  • Volume declining on recent rallies - potential exhaustion
  • Major resistance at $1.25 from historical chart patterns
  • Holder distribution unchanged - no whale dumping yet
  • Risk/reward ratio deteriorating at current levels

Suggestion: Take 50% profit at $1.20, let 50% ride with trailing stop at -15%.

Result: Took partial profit at $1.19. Token peaked at $1.32 (+15%), then retraced to $0.88 (-25%). Remaining 50% exited at $1.05 via trailing stop. Overall realized: +156% (better than holding through dump).

Key Takeaways

Lessons from Real Examples

  • AI Catches What Numerical Filters Miss:Example 1 passed all numerical filters but AI detected scam patterns. Always use AI as a complementary layer, not a replacement for basic filters.
  • UNCERTAIN Requires Human Judgment:Example 3 shows AI can't always be certain. Use advisory mode and manually review borderline cases. Your risk tolerance determines how to handle uncertainty.
  • Entry Timing Matters More Than Discovery:Example 5 prevented a FOMO entry that would have resulted in -68% loss. Even good tokens can be bad entries at wrong prices. Use AI entry analysis to time positions better.
  • Partial Profits Beat Greed:Example 6 demonstrates the value of partial profit-taking. AI suggested de-risking at +181%, avoiding the full retracement. Lock in gains while leaving upside exposure.

Example Configurations

Here are the configurations used in the examples above:

Conservative Setup

Used for Examples 1-3 (Filtering)

[ai.analysis]
filtering_enabled = true
filtering_enforce = true
filtering_providers = ["claude", "gpt4o"]
auto_blacklist_enabled = true
auto_blacklist_confidence = 0.85

Dual premium models for maximum accuracy. High auto-blacklist threshold for safety.

Balanced Setup

Used for Examples 4-6 (Entry/Exit)

[ai.analysis]
entry_enabled = true
entry_providers = ["deepseek", "groq"]
exit_enabled = true
exit_providers = ["gpt4o-mini", "deepseek"]
auto_blacklist_confidence = 0.8

Free/cheap providers for speed. Premium for exits when precision matters.

Next Steps