> For the complete documentation index, see [llms.txt](https://mentat-minds-doc.gitbook.io/mentat/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://mentat-minds-doc.gitbook.io/mentat/invest/sectors.md). # Sectors All our indexes have the same functioning: they allocate your capital into the top 3 Bittensor subnets of the sector, selected and weighted based on trend-confirmation metrics. **Allocation:** 100% of the allocated capital is invested across the top 3 subnets of the sector. **Selection:** The ranking of subnets is determined using an Exponential Moving Average on the market cap. This ensures that only subnets with confirmed performance trends enter the index. **Rebalancing:** When a new subnet enters the EMA-based top 3, it is automatically added to the index, and the portfolio is reallocated. Mentat is also responsible for selecting the best validators on each subnet to optimize APY. **Distribution:** Within the top 3, allocation is weighted using an EMA applied to subnet prices. Higher-priced subnets receive a larger share of the portfolio. The allocation share in each subnet is automatically adjusted according to changes in their market cap. ### **Index Verification Process** * We maintain a master file listing all Bittensor subnets with their official links. * We update subnets and links by cross-checking against Bittensor's official Discord. * Once a week, our team reviews subnets, verifies official links, and proposes sector changes or removals. * Our team reviews and validates all proposed changes for accuracy before approval. * Validated changes are applied to the sectors. *Mentat uses subnets' official links to classify them. If a subnet changes its orientation without updating its official links, Mentat cannot be held responsible for adding the subnet to an index.* ### **Sector Definitions** #### **Prediction** The Prediction sector includes Bittensor subnets whose primary function is to forecast future outcomes. **Subnets currently included in the index:** 6; 8; 18; 30; 41; 50; 53; 88; 103 #### **Compute** The Compute sector includes Bittensor subnets whose primary function is to provide computational resources required to run, train, or serve artificial intelligence systems, as well as support general-purpose algorithmic and computational workloads. These subnets supply the hardware capacity and execution environments necessary for AI workloads, enabling models and applications to perform computation at scale. Unlike sectors focused on building models or applications, the Compute sector focuses on delivering the infrastructure that performs the computation itself. **Exclusions:** The Compute sector does not include subnets whose primary goal is to improve models, or solve domain-specific problems, even if they rely heavily on compute resources. **Subnets currently included in the index:** 4; 12; 51; 64; 75; 128 #### **DeFi** The DeFi sector includes Bittensor subnets whose primary function is to build, operate, or support decentralized financial protocols and financial infrastructure within blockchain ecosystems. These subnets focus on creating financial primitives and services such as decentralized exchanges, liquidity infrastructure, bridges, lending systems, or yield mechanisms that enable users or other systems to interact with decentralized financial markets. Subnets that provide capital allocation strategies based on current yields, liquidity conditions, or protocol incentives, rather than directional market views, are also included in this sector. Unlike sectors that focus on prediction or directional trading strategies, the DeFi sector focuses on the infrastructure and mechanisms that enable decentralized financial activity. **Exclusions:** Subnets that use DeFi protocols to take directional trading positions on digital assets are not included in this sector and are instead classified under the Prediction Index. **Subnets currently included in the index:** 35; 77; 106; 112; 116 #### **Security** The Security sector includes Bittensor subnets whose primary function is to identify, analyze, or mitigate risks, vulnerabilities, and malicious activities within digital systems, networks, or blockchain infrastructures. These subnets focus on producing systems or intelligence that help detect security threats, protect digital assets, and improve the resilience of software, protocols, or network environments. **Subnets currently included in the index:** 54; 60; 61 #### **AI Training** The AI Training sector includes Bittensor subnets whose primary function is to train, fine-tune, or improve foundational artificial intelligence models. These subnets focus on developing general-purpose model capabilities through training processes, optimization methods, feedback systems, or evaluation frameworks that contribute directly to the improvement of base models. Unlike sectors focused on domain-specific intelligence or applications, the AI Training sector focuses on training methodologies and systems used to develop foundational AI models. **Exclusions:** Subnets primarily training models for specific applications are not included and are classified under the sector corresponding to their application. **Subnets currently included in the index:** 1; 5; 9; 24; 56 #### **Agents** An AI agent is a system that can autonomously pursue goals by deciding and executing actions within an environment. Unlike a standard model that only produces an output for a single input, an agent maintains a process over time: it interprets information, chooses actions, and iteratively updates its behavior based on results. **Index Scope** The Agents index includes Bittensor subnets whose primary function is to develop, operate, or enable the deployment of AI agents. This sector includes two main categories of subnets: * **Agent Development**: Subnets that build, train, or operate agents capable of autonomously pursuing objectives by making decisions and executing actions. * **Agent Distribution Infrastructure**: Subnets that provide infrastructure for the discovery, distribution, coordination, or monetization of agents, such as agent marketplaces, agent registries, or platforms that allow users or systems to access and deploy agents. **Exclusions:** The Agents sector does not include subnets that focus solely on the underlying technologies used by agents. **Subnets currently included in the index:** 15; 62; 71; 121 **Disclaimer:** Only subnets for which the direction is perfectly clear to our team are added to the indexes. If you have suggestions, feel free to reach out to us on [Discord](https://discord.gg/z6H5YAzmNe).