Alpha Leak API reference: signals and live feed

Alpha Leak exposes two primary interfaces: a Server-Sent Events stream for real-time signal delivery, and a set of REST endpoints for querying historical signals, wallets, tokens, and pipeline stats. Both are available from the same base URL as your hosted instance.

Base URL

All endpoints are served from your hosted instance. Replace your-instance with the host you have been given.

https://your-instance/api

Endpoints

GET /api/live/feed

SSE stream pushing signals, graduations, and anti-signals in real time.

GET /api/live/recent

Fetch recent signals, graduations, and large trades for initial page load.

GET /api/signals

Query historical signals filtered by ML score threshold.

GET /api/wallets

Retrieve tracked wallets filtered by alpha score and activity status.

GET /api/tokens

Fetch token data including lifecycle state and risk scores.

GET /api/stats

Current pipeline stats and active market regime.

SSE live feed

`GET /api/live/feed`

The live feed is a Server-Sent Events stream. It is the primary interface for consuming Alpha Leak data — no polling required. The stream pushes three event types as they occur on-chain.

Event	Description
`signal`	A tracked wallet or genesis model has produced a signal. Subtypes: `signal`, `genesis_signal`, `anti_signal`.
`graduation`	A token has graduated from the Pump.fun bonding curve.
`heartbeat`	Sent every 15 seconds to keep connections alive through proxies and load balancers.

SSE clients reconnect automatically on disconnection. No manual reconnect logic is needed in the browser or Node.js EventSource implementation.

JavaScript connection example

const es = new EventSource('https://your-instance/api/live/feed');

es.addEventListener('signal', (event) => {
  const data = JSON.parse(event.data);

  if (data.type === 'genesis_signal') {
    // New token scored highly in its first 60 seconds
    console.log('Genesis:', data.data.symbol, `score: ${data.data.genesisScore}`);

  } else if (data.type === 'anti_signal') {
    // Active token flagged — adversarial conditions detected
    console.warn('Anti-signal:', data.data.tokenMint, data.data.reasons);

  } else {
    // Standard tracked-wallet signal
    console.log('Signal:', data.data.tokenMint, data.data.action);
  }
});

es.addEventListener('graduation', (event) => {
  const token = JSON.parse(event.data);
  console.log(`${token.symbol} graduated — ${token.volume} SOL total volume`);
});

es.onerror = () => {
  // SSE clients reconnect automatically — no manual handling required
};

Signal subtypes

The signal SSE event carries a type field that identifies the signal subtype:

signal — a tracked wallet with a quality profile has made a significant buy. The payload includes wallet alpha score, ML probability score, token lifecycle state, buy rank, velocity data, and creator risk.
genesis_signal — a newly created token scored highly across the genesis model’s 75-feature vector during its first 60 seconds. No tracked wallet is required — this is based purely on launch dynamics.
anti_signal — a token with recent buy signals has triggered 2 or more adversarial detection rules simultaneously (e.g. coordinated bundle combined with high bot buyer percentage). Any open position in the flagged token should be treated as an exit cue.

Recent data

`GET /api/live/recent`

Returns recent signals, graduations, and large trades. Call this once on page or app load to seed your initial state, then let the SSE stream handle updates from that point forward.

Query parameters

limit

number

default:"20"

Maximum number of records to return per category. Applies to signals, graduations, and largeTrades independently.

Response fields

signals

object[]

Recent buy/sell signals with ML scores, in descending chronological order.

graduations

object[]

Recently graduated tokens, in descending chronological order.

largeTrades

object[]

Recent individual trades of 1.0 SOL or more, in descending chronological order.

Example

const res = await fetch('https://your-instance/api/live/recent?limit=20');
const { signals, graduations, largeTrades } = await res.json();

// signals      — recent buy/sell signals with ML scores
// graduations  — recently graduated tokens
// largeTrades  — recent trades >= 1.0 SOL

Signals

`GET /api/signals`

Returns a paginated list of historical signals filtered by ML score. Use this endpoint to query the signal archive or to build backtesting datasets.

Query parameters

min_ml_score

number

default:"0"

Filter to signals where the ML model’s calibrated probability is at or above this value. Range: 0.0–1.0.

limit

number

default:"50"

Maximum number of signals to return per page.

Example

// Signals with ML scores above 80% — paginated
const signals = await fetch('/api/signals?min_ml_score=0.80&limit=50').then(r => r.json());

Wallets

`GET /api/wallets`

Returns tracked wallets with their computed alpha scores and quality metrics. The system builds and maintains a scored profile for every wallet it observes, updated on a rolling 30-minute cycle across up to 5,000 wallets per pass.

Query parameters

min_alpha

number

default:"0"

Filter to wallets with an alphaScore at or above this value. Range: 0–100.

active_only

boolean

default:"false"

When true, returns only wallets that have made a trade within the current scoring window.

Example

// Tracked wallets with alpha score >= 70, active only
const wallets = await fetch('/api/wallets?min_alpha=70&active_only=true').then(r => r.json());

Tokens

`GET /api/tokens`

Returns token data including the current lifecycle state, risk score, bundle confidence, and creator risk score. Tokens remain tracked throughout their lifecycle on the bonding curve and after graduation.

Query parameters

graduated

boolean

Filter tokens by graduation status. true returns only graduated tokens. false returns only tokens still on the bonding curve. Omit to return all tokens.

Example

// Tokens still on the bonding curve
const tokens = await fetch('/api/tokens?graduated=false').then(r => r.json());

Stats

`GET /api/stats`

Returns a snapshot of current pipeline throughput and the active market regime. Useful for dashboards and health monitoring.

Response fields

regime

object

Active market regime snapshot, reclassified every 10 minutes.

Show regime properties

state

string

Current regime classification. One of: bull_euphoria, bull_normal, transition, bear.

confidence

number

Classification confidence between 0 and 1. For bull_euphoria and bear, scales with distance from thresholds. For bull_normal and transition, fixed at 0.6.

signalsLast1h

number

Number of signals emitted in the last 60 minutes.

walletsTracked

number

Total number of wallets currently tracked with scoring profiles.

tokensActive

number

Number of tokens currently being monitored on the bonding curve.

Example

// Current pipeline stats and market regime
const stats = await fetch('/api/stats').then(r => r.json());

{
  "ws": {
    "connected": true,
    "eventsPerSec": 42,
    "totalEvents": 4320000,
    "currentSlot": 312450000
  },
  "signals": {
    "total": 84200,
    "last24h": 1240
  },
  "wallets": {
    "tracked": 2840,
    "scored": 1920,
    "bots": 340
  },
  "regime": {
    "state": "bull_normal",
    "confidence": 0.6
  },
  "models": {
    "loaded": 4,
    "scored": 81200,
    "fallbacks": 42
  },
  "uptime": 86400,
  "memMb": 312
}

Signal payload reference

Every signal event on the SSE feed carries a structured JSON payload. The top-level type field identifies the signal subtype, and all signal data is nested under data.

type

string

required

Signal subtype. One of: "signal", "genesis_signal", "anti_signal".

data

object

required

The full signal payload.

Show data properties

tokenMint

string

required

The token’s on-chain mint address. This is the unique identifier for the token throughout the system.

wallet

string

The wallet address that triggered the signal. Present on signal type only — absent for genesis_signal and anti_signal.

action

string

Trade direction. One of: "buy", "sell". Present on signal type only.

solAmount

number

Size of this specific buy in SOL.

buyRank

number

The position of the triggering wallet among all buyers on this token. A buyRank of 1 means this wallet was the first buyer. Lower is earlier.

curvePct

number

Current fill percentage of the bonding curve, expressed as a decimal. 0.12 means the curve is 12% filled.

curveSol

number

Total SOL currently held in the bonding curve.

buysLast60s

number

Number of distinct buy transactions on this token in the last 60 seconds. A primary velocity indicator.

buysLast300s

number

Number of distinct buy transactions on this token in the last 5 minutes.

walletStats

object

Quality metrics for the wallet that triggered the signal. Present on signal type only.

Show walletStats properties

alphaScore

number

Composite wallet quality score on a scale of 0–100. Higher scores indicate wallets with a consistent track record of finding winning tokens early.

graduationRate

number

Fraction of tokens this wallet bought that subsequently graduated from the bonding curve. Expressed as a decimal — 0.41 means 41%.

winRate

number

Fraction of closed positions where the wallet exited in profit. Expressed as a decimal — 0.68 means 68% of trades closed green.

captureEfficiency

number

Average fraction of the available peak return this wallet captured. A wallet that exits at 72% of the peak on average has a captureEfficiency of 0.72.

tokenState

object

Risk and lifecycle metadata for the token being traded.

Show tokenState properties

lifecycleState

string

Current classification of the token’s lifecycle. Common values: "early_accumulation", "momentum", "distribution", "cooling". Tokens in "momentum" or "early_accumulation" have historically produced the strongest returns.

riskScore

number

Token-level risk score on a scale of 0–100. Lower values indicate safer tokens with fewer adversarial signals. A score above 50 warrants caution.

bundleConfidence

number

Model confidence that early buys on this token were part of a coordinated bundle. Expressed as a decimal — 0.12 means low bundle confidence (looks organic). Values above 0.5 indicate likely coordinated activity.

creatorRiskScore

number

Risk score assigned to the token’s creator address, based on historical behaviour across all previously created tokens. Scale of 0–100. Lower values indicate creators with a cleaner track record.

Full payload example

{
  "type": "signal",
  "data": {
    "tokenMint": "So11...",
    "wallet": "ABC1...",
    "action": "buy",
    "solAmount": 0.5,
    "buyRank": 4,
    "curvePct": 0.12,
    "curveSol": 8.3,
    "buysLast60s": 12,
    "buysLast300s": 34,
    "walletStats": {
      "alphaScore": 84,
      "graduationRate": 0.41,
      "winRate": 0.68,
      "captureEfficiency": 0.72
    },
    "tokenState": {
      "lifecycleState": "momentum",
      "riskScore": 18,
      "bundleConfidence": 0.12,
      "creatorRiskScore": 24
    }
  }
}

Signal quality guide

Not all signals carry the same conviction. The following criteria distinguish a high-quality signal from a marginal one. The more of these conditions a signal satisfies, the stronger the case for acting on it.

The best signals combine a high-scoring wallet entering early, rising buy velocity, a token in a favourable lifecycle state, and low adversarial risk across all three risk fields.

Field	High-quality indicator	What it means
`walletStats.alphaScore`	≥ 75	The wallet has a strong, consistent track record of early entries on winning tokens.
`buyRank`	≤ 5	The wallet entered very early — limited competition for price appreciation.
`buysLast60s`	Rising relative to `buysLast300s`	Buy velocity is accelerating, suggesting growing market interest.
`tokenState.riskScore`	≤ 25	Low adversarial signal density on this token. Fewer red flags.
`tokenState.bundleConfidence`	≤ 0.20	Early buys appear organic rather than coordinated. Lower manipulation risk.
`tokenState.lifecycleState`	`"momentum"` or `"early_accumulation"`	Token is in a phase historically associated with upward price movement.
`tokenState.creatorRiskScore`	≤ 30	Creator has a history of legitimate launches without rug patterns.

An anti_signal for a token overrides all quality indicators. If an anti_signal arrives for a token you hold a position in, the recommended response is an immediate exit — regardless of how strong the original signal appeared.

API Reference

Alpha Leak API reference: signals and live feed

Base URL

Endpoints

GET /api/live/feed

GET /api/live/recent

GET /api/signals

GET /api/wallets

GET /api/tokens

GET /api/stats

SSE live feed

`GET /api/live/feed`

JavaScript connection example

Signal subtypes

Recent data

`GET /api/live/recent`

Query parameters

Response fields

Example

Signals

`GET /api/signals`

Query parameters

Example

Wallets

`GET /api/wallets`

Query parameters

Example

Tokens

`GET /api/tokens`

Query parameters

Example

Stats

`GET /api/stats`

Response fields

Example

Signal payload reference

Full payload example

Signal quality guide

Build docs developers (and LLMs) love

API Reference

Documentation Index

​Base URL

​Endpoints

GET /api/live/feed

GET /api/live/recent

GET /api/signals

GET /api/wallets

GET /api/tokens

GET /api/stats

​SSE live feed

​GET /api/live/feed

​JavaScript connection example

​Signal subtypes

​Recent data

​GET /api/live/recent

​Query parameters

​Response fields

​Example

​Signals

​GET /api/signals

​Query parameters

​Example

​Wallets

​GET /api/wallets

​Query parameters

​Example

​Tokens

​GET /api/tokens

​Query parameters

​Example

​Stats

​GET /api/stats

​Response fields

​Example

​Signal payload reference

​Full payload example

​Signal quality guide

Build docs developers (and LLMs) love

Base URL

Endpoints

SSE live feed

`GET /api/live/feed`

JavaScript connection example

Signal subtypes

Recent data

`GET /api/live/recent`

Query parameters

Response fields

Example

Signals

`GET /api/signals`

Query parameters

Example

Wallets

`GET /api/wallets`

Query parameters

Example

Tokens

`GET /api/tokens`

Query parameters

Example

Stats

`GET /api/stats`

Response fields

Example

Signal payload reference

Full payload example

Signal quality guide