Risk Insights

Copy Markdown View Source

Tink.RiskInsights and Tink.RiskCategorisation provide transaction-level risk scoring and categorisation signals. Use them to detect anomalous spending, assess creditworthiness, or flag high-risk financial behaviour.

Risk Insights

Fetching Insights

{:ok, insights} = Tink.RiskInsights.get(client, user_id: "user_123")

IO.inspect(insights.risk_score)        # 0.0–1.0
IO.inspect(insights.overdraft_events)  # count in last 90 days
IO.inspect(insights.income_stability)  # :high | :medium | :low

Insight Fields

FieldDescription
risk_scoreComposite risk score 0.0 (low) – 1.0 (high)
overdraft_eventsNumber of overdraft events in the analysis window
income_stability:high, :medium, or :low
average_balanceRolling 90-day average balance as Decimal
balance_volatilityStandard deviation of daily balances
gambling_detectedBoolean — gambling transactions present
credit_usageRatio of credit card spend to available credit

Risk Categorisation

Tink.RiskCategorisation provides raw category signals at the transaction level, giving you the building blocks to construct custom risk models.

Fetching Categorisations

{:ok, report} = Tink.RiskCategorisation.get_report(client, session.id)

Enum.each(report.categories, fn cat ->
  IO.puts("#{cat.type}: #{cat.amount} — #{cat.signal}")
end)

Category Signals

SignalDescription
GAMBLINGGambling platform transactions
PAYDAY_LOANHigh-interest short-term borrowing
OVERDRAFTTransactions that triggered an overdraft
IRREGULAR_INCOMEIncome with high variance
LARGE_CASH_WITHDRAWALATM withdrawals above threshold
FREQUENT_TRANSFERSUnusually high transfer frequency

Combining Insights and Categorisation

with {:ok, insights}  <- Tink.RiskInsights.get(client, user_id: user_id),
     {:ok, report}    <- Tink.RiskCategorisation.get_report(client, session_id) do

  if insights.risk_score > 0.7 or report.gambling_transaction_count > 0 do
    {:decline, :high_risk}
  else
    {:approve, insights.risk_score}
  end
end