Wallet Intelligence Model
The Wallet Intelligence Model (WIM) is Raphael Sentinel’s long-term behavioral profiling system that tracks how wallets behave across the Solana ecosystem. It acts as a memory layer for the platform, enabling predictive insights and improving the accuracy of risk scoring over time.
WIM is designed around a simple principle:
Wallets behavior is often more reliable than token behavior.
By continuously analyzing how wallets participate in launches, trade, fund each other and exit, Sentinel builds an evolving intelligence graph that becomes increasingly powerful with every new token processed.
1. Purpose of the Wallet Intelligence Model
While most scanners look only at token-level metadata, Sentinel focuses on people-level behavior:
Who funded the early buyers?
Who repeatedly launches farm-style tokens?
Who consistently dumps on retail?
Which clusters of wallets operate together?
Which wallets behave like bots vs real users?
This intelligence fundamentally amplifies the accuracy of:
dev cluster detection
swarm analysis
exit pattern prediction
risk scoring
wallet quality scoring
2. Long-Term Behavioral Profiling
Every wallet interacting with analyzed tokens is evaluated across multiple dimensions.
Tracked Attributes
number of past launches participated in
timing behavior (sniping, synchronized bursts)
holding duration patterns
exit intensity
consistency of buy/sell sequencing
funding origins (direct + multi-hop)
interaction with known dev clusters
involvement in past rugs or farms
cross-token behavior similarity
Example Insights
“Wallet A frequently participates in dev-led farms.”
“Wallet B funds 80% of wallets inside a dev cluster.”
“Wallet C repeatedly dumps within 3 minutes of launch.”
“Wallet D behaves like an automated sniper bot.”
This information feeds directly into risk assessments.
3. Funding Graph Analysis
WIM reconstructs funding graphs for every wallet:
direct SOL transfers
multi-hop transfers
loop structures
timing proximity
correlated outflows
Purpose:
link wallets into clusters
reveal hidden dev wallets
detect wallets controlled by the same operator
identify pre-funded swarm structures
Bonding curve tokens rely heavily on obfuscation of wallet identities — WIM deobfuscates them.
4. Cross-Token Reputation System
WIM assigns each wallet a reputation score derived from its behavior across all analyzed tokens.
Positive Reputation Signals
organic buying behavior
natural holding duration
non-synchronized patterns
low frequency dumping
no cluster affiliation
Negative Reputation Signals
recurrent early dumping
involvement in high-risk dev clusters
repeated rug-adjacent patterns
highly synchronized activity
multiwallet swarm behavior
WIM reputation directly enhances the Risk Engine.
5. Cluster Membership Scoring
Every wallet is evaluated for cluster association:
funding correlations
timing overlap
trade pattern similarity
identical sequence behavior
repeated co-participation
Outputs include:
likelihood of being dev-controlled
likelihood of belonging to a swarm
likelihood of acting as a bot partner
These insights form the backbone of Pump.fun analysis.
6. Integration With Risk Engine
WIM intelligence contributes to multiple key risk dimensions:
For SPL Tokens
wallet quality scoring
detection of malicious LP providers
surfacing wallets linked to past manipulation
For Bonding Curve Tokens
dev cluster detection
swarm detection
early buyer graph quality
exit pattern interpretation
A token with risky wallets automatically inherits elevated risk.
7. Continuous Learning & Intelligence Growth
WIM updates itself with every new token analyzed:
new wallet behaviors → new patterns
new dev clusters → updated fingerprints
bot signatures → better classification
more tokens → higher predictive accuracy
The more the system sees, the more powerful it becomes.
WIM is what transforms Raphael Sentinel from a scanner into a true on-chain intelligence platform.
Last updated