Short-horizon order-flow pulse engine for Solana. Scores aggressive flow, buyer breadth, and liquidity migration before a burst becomes obvious on the chart.
Most Solana scanners tell you that a large trade happened. That is not enough. A real continuation setup depends on whether aggressive flow is broadening across wallets, whether liquidity is being consumed faster than it refills, and whether there is still enough topbook depth to exit without getting trapped.
Pulse ingests recent Helius transactions, derives short-horizon flow metrics, and asks a Claude agent to validate whether the current burst looks like a durable momentum pulse or a shallow false break. The output is a pulse-aware alert stream with regime context, confidence, and a watchlist of names where flow is compounding.
The engine is intentionally stricter on thin books where a signal can look strong but still be hard to exit.
DETECT -> SCORE -> VALIDATE -> REGIME -> REPORT
Quant Dashboard • Validation Flow • At a Glance • Operating Surfaces • How It Works • Example Output • Technical Spec • Risk Controls • Quick Start
Use case: short-horizon Solana flow detection before the chart becomes obviousPrimary input: trade imbalance, buyer acceleration, liquidity migration, wallet entropy, slippage pressurePrimary failure mode: mistaking a shallow burst for real continuationBest for: operators who need to know whether aggressive flow is compounding or already decaying
Live order-flow board for Pulse: composite pulse score, regime state, flow components, token watchlist, alert severity, and Claude's validation notes for each short-horizon setup.
How Pulse validates a burst: Helius transactions become flow components, the pulse score is checked against topbook depth and slippage pressure, and only durable setups reach the bullish pulse report.
Quant Dashboard: shows composite pulse, regime state, and component breakdownPulse Ladder: tracks whether a name is building, active, cooling down, or fadingValidation Loop: asks the agent whether the pulse still looks tradeable after the first moveValidation Flow: shows how raw flow becomes a durable or rejected pulse signal
The hard part of short-horizon Solana trading is not noticing that activity increased. The hard part is noticing whether that activity is becoming more tradeable or less tradeable as it accelerates.
Pulse exists to make that distinction explicit. It is not trying to be a generic volume monitor. It is trying to answer whether a burst still has enough breadth and enough exit quality to matter.
Pulse follows a five-step loop:
- ingest recent Solana transaction flow for the tracked symbols
- derive the short-horizon components that make a burst meaningful
- combine those components into a composite pulse score with regime logic
- validate the score against slippage and topbook depth so the move is still executable
- rank the surviving names into a pulse report the operator can actually use
That sequence matters because size alone is not a reliable signal. The board is designed to promote compounding flow and demote shallow optical bursts.
- trade imbalance is expanding, not just flickering
- buyer participation is broadening instead of staying concentrated
- liquidity is still refilling as the move develops
- slippage pressure is low enough that exits still look realistic
If those conditions break, the score should cool down quickly.
PULSE // FLOW ALERT
[HIGH] BONK
pulse score 1.34
regime bullish
buyer accel strong
liq delta positive
slippage pressure 12 bps
operator note: flow is broadening and depth still looks exitable
Pulse builds a short-horizon composite score from five flow components:
Pulse_t = 0.30 * z(trade_imbalance_30s) + 0.25 * z(unique_buyers_accel_2m) + 0.20 * z(liquidity_delta_1m) + 0.15 * z(wallet_entropy_5m - 0.5) - 0.10 * z(slippage_pressure_30s)
Where:
trade_imbalance_30smeasures directional aggression in the most recent burst windowunique_buyers_accel_2mrewards broadening participation instead of one-wallet printsliquidity_delta_1mmeasures how quickly liquidity is disappearing or refillingwallet_entropy_5mpenalizes overly concentrated flowslippage_pressure_30spenalizes setups that look good on paper but are hard to exit cleanly
Regime classification uses hysteresis so Pulse does not flap between states:
- enter
bullishwhenpulseScore >= 1.25andtrailingPulseScore >= 0.40 - remain
bullishuntilpulseScore < 0.60 - mark
cooldownwhen either pulse or trailing confirmation drops below zero
Execution-quality guardrails:
- reject high-conviction pulses when
topbookDepthUsd < MIN_TOPBOOK_DEPTH_USD - cap conviction when
slippagePressure30s > MAX_SLIPPAGE_PRESSURE_BPS - increase re-entry delay as realized volatility expands relative to
VOLATILITY_BASELINE_BPS
Severity is not just size-based. A mid-sized burst can still rank high if the pulse score is extreme and breadth confirms continuation.
depth filter: rejects bursts that are too thin to exit cleanlyslippage cap: prevents high-scoring but untradeable setups from being promotedregime hysteresis: reduces flapping between bullish and cooldown statesre-entry delay: slows the engine down when volatility expands too quickly
Pulse is intentionally strict because a fast signal without exit quality is usually worse than no signal at all.
Helius transaction feed
-> flow metric enrichment
-> pulse score + regime model
-> Claude validation loop
-> rolling pulse report
git clone https://github.com/PulseSolana/Pulse
cd Pulse && bun install
cp .env.example .env
bun run devANTHROPIC_API_KEY=sk-ant-...
HELIUS_API_KEY=...
SOLANA_RPC_URL=https://mainnet.helius-rpc.com/?api-key=...
ALERT_THRESHOLD_USD=75000
BULLISH_PULSE_THRESHOLD=1.25
MIN_TOPBOOK_DEPTH_USD=25000
MAX_SLIPPAGE_PRESSURE_BPS=32
SCAN_INTERVAL_MS=30000- Planned commit sequence:
docs/commit-sequence.md - Draft engineering issues:
docs/issue-drafts.md
MIT
catch the burst before the breakout looks obvious.