DW · METHODOLOGY

How Deltawall measures integrity

Gross DEX volume on new Solana tokens mixes real demand with bots, repeat churn and coordinated wallets. Deltawall separates them with fixed, auditable rules — no AI guesses, no filled-in numbers.

Engines

Chain

Solana

Venues

Pump.fun · PumpSwap · Raydium

Wallet access

Never

Four-engine architecture

per token · per timeframe

01engine

Clean Execution

How much of the volume is real?

Inputs

Trade sizes, wallet trade counts, bot timing signatures

Outputs

Organic / insider / unclassified / bot split, organic floor & ceiling

02engine

Capital Flow

Is money staying in?

Inputs

Buy & sell volume by wallet class, hold windows

Outputs

Organic net flow, cumulative volume delta, absorption

03engine

Shadow Intent

What did traders try and fail to do?

Inputs

Failed transactions, slippage errors, priority fees

Outputs

Slippage-rejected buys/sells, bot retry share, fee pressure

04engine

Distribution

Who owns the supply?

Inputs

Holder balances, funding roots, transfers

Outputs

Top-10 share, clusters, transfer-only and dust holders

Score caps

applied after weighting

Cap	Trigger	Effect
Thin activity	Low trade count or volume in window	Score capped — not enough evidence
Bot dominance	Bot share of volume or trades high	Score capped regardless of other passes
Thin liquidity	Liquidity small vs market cap	Score capped — exit risk
Activity decay	Recent window far below earlier	Score capped — interest fading

Data rules

Unmeasured value: “—”, never 0
Unmeasured check: never counts as pass
Thresholds: guide, not backtested
Scores from LLM: never
Pattern bot: private evaluation
Wallet connect: never

Data pipeline

Layer	Source	Used for
Market	GeckoTerminal · DexScreener	Price, pools, trades, buy/sell counts
Execution	Jupiter	Live sell quotes for exit cost
On-chain	Helius RPC & webhooks	Authorities, holders, transfers, failed txns
Compute	Deltawall pure rule modules	Classification, scores, checks — tested