November 17, 2026Technology

API 18: Bringing Advanced Global Anti-Fraud Technologies to Our Clients

API 18: Bringing Advanced Global Anti-Fraud Technologies to Our Clients
API 18: New Capabilities for Digital Risk Assessment arrow

JuicyScore continues to develop technologies for detecting fraud schemes, analyzing related events and protecting the software environment. In API 18, the new version of the JuicyScore API, we have delivered both significant improvements to existing technologies and new methods such as RASP and cluster-graph analysis.

API 18: New Capabilities for Digital Risk Assessment

The 18th-generation JuicyScore API is one of the largest updates in the service's history. Key improvements:

  • The response vector has been expanded with 24 new variables. In its maximum configuration, the JuicyScore response vector includes more than 270 variables and signals (not counting the additional markers in Advanced Parameters).
  • Device authentication algorithms have been refined, reducing the share of false positives on repeat requests.
  • Detection of proxy connections and real IPs is now significantly more accurate (this functionality is available from API 16 onwards).
  • Detection of randomizers is now more accurate (this functionality is available from API 16 onwards).
  • Anti-fraud scoring and IDX attributes are more informative, thanks to calibration and additional variables.
  • The section on detecting viruses and malicious applications has been expanded.
  • The data center network has been expanded, and the reliability and availability of our services have been further improved, with current risks worldwide in mind.

Examples of New Variables

API 18 introduces new stop-markers and risk markers that expand the analysis of devices, software and related events. Here are a few examples. To learn more about everything new in API 18, request a demo with our experts.

Mobile device activity. Determines whether the mobile device is active. Helps detect atypical patterns of interaction with the device.

SDK integrity check. Determines whether the SDK checksum hash matches the expected value. A mismatch may indicate that the software component has been modified or its integrity compromised.

DOM injections in crypto wallets. Detects signs of third-party code injected into the DOM of crypto wallet web interfaces. This may indicate attempts to tamper with data or interfere with the application.

Number of applications in a cluster over 7 days. Shows the number of applications associated with a specific account_id within one JuicyCluster cluster over seven days. Helps detect repeated activity and possible organized fraud schemes.

Model calibration and additional variables have made anti-fraud scoring and IDX attributes more informative. JuicyScore digital signals are also used in credit scoring and LTV models.

RASP (Runtime Application Self-Protection)

RASP is protection built directly into the application: the app itself monitors its runtime environment, recognizes attacks and responds to them in real time – with no external protection tools on the device.

RASP detects rooted and jailbroken devices, emulators and simulators, MITM attacks and certificate spoofing, debugging and hooking, app modification, and overlays and screen recording (currently more than 25 signals and markers, a number that keeps growing – follow the updates to the Advanced Parameters section).

Depending on the threat detected, the technology can block individual functions or close the application in real time, alert the user, and send telemetry to the server or SOC.

RASP is designed to protect user accounts, digital wallets, payments and business logic on devices, and it makes reverse engineering harder for fraudsters.

Cluster-Graph Analysis

The cluster-graph approach combines cluster analysis and graph models to detect complex fraud schemes. It shifts the analysis from individual transactions and sessions to the network of related events, participants and devices.

The technology works in three main stages:

  1. Data collection. For each session, technical and behavioral features are analyzed: device and browser parameters, network characteristics, language, clicks and various types of noise.
  2. Processing and cluster building. Graphs and/or the main clustering directions are built, and graph nodes or sessions are then merged around the principal components in real time.
  3. Cluster analysis. Signs of multi-accounting and linked risky behavior are detected, and investigations are run across a wider range of events.

Clustering reduces computational load and cost as graphs grow larger.

We have previously written in detail about how the technology works and its role in anti-fraud scoring and device authentication.

JuicyScore Dashboards: Monitoring Risk Events Across the Entire Customer Flow

JuicyScore Dashboards is a tool for monitoring and visualizing digital risks that complements JuicyScore scoring solutions. While scoring assesses each application individually, Dashboards shows the overall picture across the customer flow: where risk is growing, which accounts are linked to each other and what is happening with a specific customer.

Key Capabilities of JuicyScore Dashboards

Monitoring risk spikes. Track trends in key risk groups: multi-accounting, bot activity, device and internet connection anomalies, and stop-markers. Comparing metrics for the last hour and the last day with previous periods helps catch sharp changes early, before they affect the portfolio.

Account link graphs. A link graph based on Device ID and Real IP groups accounts into clusters. Multi-accounting and bonus abuse become visible even when each individual application looks legitimate.

Investigation down to the session level. Go from the CRM or back office to the events and risk signals of a specific customer by Client ID or Session ID, without exports or manual data reconciliation.

Integration and Deployment

JuicyScore Dashboards works with data from JuicyScore, JuicyID and JuicyLightning, using the risk signals already flowing into the client's system. The solution requires no in-house BI development and can be deployed on a dedicated virtual machine or in Juicy Cloud, with information security requirements in mind. JuicyScore Dashboards is available for API versions 16 and above.

Share this post

See How We Spot Fraud Before It Happens — Book Your Expert Session

  • list marker

    See It in Action with a Real Expert

    Get a live session with our specialist who will show how your business can detect fraud attempts in real time.

  • list marker

    Explore Real Device Insights in Action

    Learn how unique device fingerprints help you link returning users and separate real customers from fraudsters.

  • list marker

    Understand Common Fraud Scenarios

    Get insights into the main fraud tactics targeting your market — and see how to block them.

Our Contacts:

Leading Brands Trust JuicyScore:

robocash
id finance
tabby

Get in touch with us

Our dedicated experts will reach out to you promptly