> For the complete documentation index, see [llms.txt](https://docs.deep3.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.deep3.ai/the-technology/ai-models/deepshield-hft.md).
# DeepShield-HFT
## Overview
DeepShield-HFT is our specialized security model designed to identify high-frequency trading (HFT) addresses on Ethereum. With an accuracy of over 95%, it helps dapp developers and users alike distinguish between organic trading activity and machine-controlled bots that can disrupt market integrity. Although DeepShield-HFT currently focuses on historical detection, its architecture can be extended to support real-time use cases in the future.
Built on the same foundational architecture as our HODL-C1 model, DeepShield-HFT benefits from the rigorous development and testing methods that have proven successful in our other solutions. This ensures a robust and scalable approach to detecting rapid trading patterns that might otherwise go unnoticed.
## Key features
DeepShield-HFT excels at flagging addresses that consistently perform multiple trades per minute, indicating the potential presence of bots or automated systems. Its core feature set enables dapp developers to proactively enforce security measures—such as banning or restricting suspicious addresses—while also providing a clearer view of on-chain trading dynamics.
Beyond security, DeepShield-HFT can serve as a valuable on-chain intelligence tool. By pinpointing which tokens are heavily traded by bots, the model helps users and developers gauge how much of a token’s volume is truly organic. This insight supports more transparent market analyses and informed decision-making.
## Data sources & inputs
Like HODL-C1, DeepShield-HFT is trained exclusively on Ethereum on-chain data. Although it can be extended to other networks, the current model focuses on analyzing wallet addresses and their transaction histories within the Ethereum ecosystem. All insights are derived from publicly available blockchain records, ensuring transparency and verifiability.
## Methods & technical details
DeepShield-HFT adopts the same classification framework used in HODL-C1, benefiting from advanced feature engineering and robust model training. For labeling, addresses are considered high-frequency traders if their last 3 out of 50 trades each had a holding period of one minute or less. This strict criterion ensures the model accurately captures addresses engaging in rapid, bot-like trading behaviors.
By leveraging proven machine learning techniques and extensive training data, DeepShield-HFT offers a high degree of reliability in flagging suspicious activity. Its flexible architecture allows for future enhancements, including adaptation to other blockchains and potential real-time detection capabilities.
## Performance & accuracy
DeepShield-HFT achieves over 95% accuracy in identifying high-frequency trading addresses, a performance level consistent with our broader suite of security and analytics models. This accuracy metric stems from rigorous training and validation, where out-of-sample testing confirms the model’s ability to generalize effectively to new data.
Because the model borrows directly from the HODL-C1 architecture, it also inherits the proven methodologies that minimize overfitting and ensure stable performance. This gives developers confidence in deploying DeepShield-HFT for mission-critical security applications.
## Use cases
DeepShield-HFT is primarily intended to help dapp developers detect and mitigate bot-driven trading activity. By identifying high-frequency trading addresses, developers can implement tailored security measures such as outright bans, restricted permissions, or clear flagging within user interfaces. This proactive approach preserves the integrity of on-chain markets and protects human users from unfair or manipulative trading practices.
In addition, DeepShield-HFT provides valuable market intelligence for anyone interested in understanding the nature of token trading volume. By highlighting the presence of automated activity, it enables more transparent analyses of market liquidity and trends. As the Web3 ecosystem continues to evolve, DeepShield-HFT stands ready to adapt, offering a reliable foundation for both current security needs and emerging real-time detection scenarios.
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter:
```
GET https://docs.deep3.ai/the-technology/ai-models/deepshield-hft.md?ask=
```
The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.