Insider Selling API

Insider Selling exposes a composite 0-10 hedge-pressure score per coin built from cross-venue negative funding, funding persistence, open interest, basis, CVD, sell-side flow, and supporting wallet/supply context. Basis and CVD exist only for the futures-covered majors, so most altcoin scores carry them null and are lower bounds (read signal_coverage_pct). High scores indicate hedge-like short pressure that deserves closer due diligence.

The signal updates every 30 minutes and is gated to coins where the score is strictly greater than zero, so an empty list means no qualifying pressure right now.

GET/v1/insider-selling/data

Get insider selling data

Returns coins ranked by their composite insider selling pressure score, descending.

Query parameters

Name
limit
Type
integer
Description
Maximum number of coins to return. 1 to 500. Default 100.
Name
min_score
Type
number
Description
Filter to coins scoring at or above this threshold. 0 to 10. Default 0.
Name
coin
Type
string
Description
One coin by its CoinGecko id (e.g. pepe): its row exactly as the scorer wrote it, whatever its score. A coin that stopped qualifying answers score: null with its sub-scores and phase cleared; an id with no row answers an empty list. limit and min_score do not apply.

Response fields

Name
coins
Type
array
Description
Array of scored coins, sorted by score descending.
Name
coins[].coin_id
Type
string
Description
Canonical coin slug.
Name
coins[].symbol
Type
string
Description
Ticker symbol.
Name
coins[].name
Type
string
Description
Display name.
Name
coins[].score
Type
number | null
Description
Composite pressure score from 0 to 10. Higher means stronger alignment across shortability, funding, market confirmation, flow, and supporting wallet/supply signals. If funding, price or derivatives inputs are missing, the score is a conservative lower bound rather than a measured neutral reading; missing on-chain inputs are left out and the remaining weights re-normalize, so that gap can move the score either way. null only on a coin request for a coin that stopped qualifying.
Name
coins[].signal_coverage_pct
Type
integer
Description
Weighted input completeness from 0 to 100 across funding, price, derivatives, and on-chain groups.
Name
coins[].signal_coverage_status
Type
string
Description
full when every signal group is observed, partial at 50% coverage or more, otherwise limited. Compare scores only with their coverage in view.
Name
coins[].incomplete_signal_groups
Type
string[]
Description
Signal families with one or more unavailable inputs.
Name
coins[].score_interpretation
Type
string
Description
lower-bound-partial-inputs when a funding, price or derivatives input is missing (it scores as zero), otherwise fully-observed. A row missing only on-chain inputs reads fully-observed: those are left out and the weights re-normalize, and the gap shows in incomplete_signal_groups.
Name
coins[].avg_funding_72h
Type
number
Description
Average funding rate across exchanges over the trailing 72 hours, expressed as a decimal (e.g. -0.00012 is -0.012%).
Name
coins[].funding_magnitude
Type
number
Description
Sub-score reflecting the absolute size of negative funding.
Name
coins[].funding_persistence
Type
number
Description
Sub-score reflecting how consistently the funding stays negative.
Name
coins[].wallet_concentration
Type
number | null
Description
Supporting sub-score [0, 1] from the top-10 holder share of supply. null when no on-chain holder data is available.
Name
coins[].holder_quality_deterioration
Type
number | null
Description
Sub-score [0, 1] summarizing fragile holder base, top-heavy supply, liquidity weakness, and contract/security context. null when none of those inputs is observed.
Name
coins[].distribution_pressure
Type
number
Description
Sub-score [0, 1] summarizing direct wallet/flow pressure and derivatives-confirmed selling pressure.
Name
coins[].market_confirmation
Type
number
Description
Sub-score [0, 1] summarizing funding, OI, basis, CVD, price path, and persistence confirmation.
Name
coins[].manipulation_phase
Type
string
Description
Current phase label such as active-hedge-pressure, persistent-shorting, or supply-risk-supported.
Name
coins[].consecutive_negative_days
Type
integer
Description
Number of consecutive trading days with negative average funding.
Name
coins[].price_change_24h
Type
number
Description
Trailing 24h price change as a percentage.
Name
coins[].price_change_7d
Type
number
Description
Trailing 7d price change as a percentage.
Name
coins[].top10_holders_pct
Type
number | null
Description
Share of supply held by the top 10 holders, percent; null when unknown. Contract-held supply (team treasuries, vesting) and exchange hot wallets count as holders.
Name
coins[].volume_to_liquidity_24h
Type
number | null
Description
24h DEX volume divided by available liquidity. High values can indicate fragile churn.
Name
coins[].holders_change_pct
Type
number | null
Description
Holder-count change since the previous successful scoring run, expressed as a decimal.
Name
coins[].top10_holders_pct_change
Type
number | null
Description
Change in top-10 holder share since the previous successful scoring run, expressed in percentage points.
Name
coins[].signal_start_date
Type
string
Description
ISO 8601 timestamp of when the negative funding streak began.
Name
coins[].updated_at
Type
string
Description
ISO 8601 timestamp of the most recent score recomputation.
Name
count
Type
integer
Description
Number of rows in coins.
Name
data_as_of
Type
string | null
Description
Latest updated_at timestamp represented by the returned rows, in ISO 8601 format. null when no valid score timestamp is available.
Name
stale
Type
boolean
Description
true when data_as_of is missing, invalid, or more than 1 hour old.
Name
freshness_sla_seconds
Type
integer
Description
Maximum supported score age in seconds. Currently 3600 (1 hour).
Name
description
Type
string
Description
Plain-language summary of the signal for client display.
Name
signal
Type
object
Description
Methodology metadata: methodology, update_frequency, and the list of exchanges monitored.

Request

GET

/v1/insider-selling/data

curl -G https://www.sharpe.ai/api/v1/insider-selling/data \
  -H "Authorization: Bearer sk_live_your_key_here" \
  -d limit=20 \
  -d min_score=4

Response

{
  "data": {
    "coins": [
      {
        "coin_id": "example-token",
        "symbol": "EXMP",
        "name": "Example Token",
        "score": 8.4,
        "signal_coverage_pct": 91,
        "signal_coverage_status": "partial",
        "incomplete_signal_groups": ["derivatives"],
        "score_interpretation": "lower-bound-partial-inputs",
        "avg_funding_72h": -0.00018,
        "funding_magnitude": 0.79,
        "funding_persistence": 0.88,
        "wallet_concentration": 0.91,
        "holder_quality_deterioration": 0.62,
        "distribution_pressure": 0.73,
        "market_confirmation": 0.68,
        "manipulation_phase": "active-hedge-pressure",
        "consecutive_negative_days": 12,
        "price_change_24h": -3.21,
        "price_change_7d": -11.4,
        "signal_start_date": "2026-04-14T00:00:00Z",
        "updated_at": "2026-04-26T07:00:00Z"
      }
    ],
    "count": 1,
    "data_as_of": "2026-04-26T07:00:00Z",
    "stale": false,
    "freshness_sla_seconds": 3600,
    "description": "Insider hedge-pressure scores based on cross-venue negative funding, open interest, basis, CVD, sell-side flow, and wallet/supply context.",
    "signal": {
      "methodology": "Composite 0-10 pressure score built around observable short-hedge behavior. Negative cross-venue funding is required; OI, basis, CVD, sell-side flow, and wallet/supply context raise conviction. Exact weighting is proprietary to reduce gaming.",
      "update_frequency": "every 30 minutes",
      "exchanges": [
        "Binance", "Bybit", "OKX", "Gate.io", "Bitget", "Hyperliquid",
        "Deribit", "HTX", "BingX", "KuCoin", "MEXC"
      ]
    }
  },
  "meta": {
    "request_id": "req_insider_abc123",
    "timestamp": "2026-04-26T07:00:05Z",
    "elapsed_ms": 12
  }
}

The response header X-Sharpe-Data-State is ok or stale using the same 1-hour SLA. Stale rows remain available with HTTP 200, but clients should treat them as historical context rather than a current trading signal. A database or query failure answers 503 (service_unavailable, Retry-After: 30) in the normal API error envelope.

How the score is built

The score is a composite risk model capped at 10. It does not depend on stated unlock calendars, because unlock timing and size can be incomplete or adversarial. Instead, it requires observable short-hedge pressure and then uses multiple signal families:

Hedge pressure: negative cross-venue funding, funding persistence, and exchange breadth.
Market confirmation: open interest, basis, true taker-flow/CVD, short crowding, and price path.
Sell-side flow: DEX flow imbalance and weakening liquidity context.
Supply context: top-10 holder share, holder count, liquidity depth, and mint/freeze authority flags.

Wallet and supply signals raise conviction, but they do not independently create an insider-selling signal. We publish the signal families and validation process, but not the exact formula, weights, or thresholds. That keeps the leaderboard harder to game while still making the approach understandable to users. Coins with score == 0 are filtered out of the response, so the coins array is always meaningfully populated when returned.

Unavailable derivatives, price and funding inputs contribute no risk. This intentionally keeps a partial-input score conservative, but it also means the number is a lower bound rather than directly comparable with a fully observed row. Unavailable on-chain inputs are left out instead and the remaining weights re-normalize, so an on-chain gap can move the score either way. Always inspect signal_coverage_pct, signal_coverage_status, and incomplete_signal_groups alongside score.

Borrow and OTC deal details are treated as optional context, not required inputs. Public borrow data is inconsistent across venues and private OTC hedges are intentionally opaque, so the production score prioritizes continuously observable perp behavior.

See the pump & dump signal