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