## What does the subnet do?

As the subnet name and tagline imply, it’s investing with a decentralized approach. In essence the subnet is a decentralized hedge fund, with a global network of quants and analysts providing competing investment strategies. We combine top strategies into ensemble strategies, turning them into the final product of the subnet – actionable intelligence that we use to manage assets, and/or offer to 3rd parties.

With a unique design and solid technical foundation that follow the KISS and First Principles, the subnet can support any asset class, including Tao/Alpha staking, US stocks and stocks from other countries, forex, commodities, and more. It also supports all investment styles, from classic quant algorithms to machine learning & AI models to human insights and everything in between.

## How does the subnet work?

**Overview**: Miners provide competing strategies. Validators score and reward the top performing. We further process top strategies into the final product. UID space and emissions are partitioned to support different asset classes.

The subnet relies on our “miners” as quants and analysts in a high-fidelity, simulated trading environment. A miner starts with an asset class of their choice, a fixed amount of initial capital, and provides a continuously updated asset allocation table within the asset class. This is called a strategy or portfolio. The miner can update their strategy or “actively rebalance” the portfolio as often as they like, according to their preferred timeframe, based on their quant/AI models and/or human insights.

Shorting is supported with simple negative allocation numbers for any asset class that allows it.

Our “validators” gather strategies from miners, calculate each strategy’s return on a daily basis. Using our exclusive scoring algorithm that incorporates daily returns and other factors such risk and volatility, validators assign a score to each strategy. The higher the score, the more emissions are rewarded to the miner. This process naturally surfaces low risk, high return strategies from expert quants and analysts.

With top strategies as raw inputs, we apply our own quant methods/models and combine them into one meta or ensemble strategy in each asset class (can also be more than one strategy in a class, depending on different constraints such as timeframe, in-class asset selection). This is the final product – actionable intelligence that we can directly use to manage assets and portfolios, and/or offer to 3rd parties.

The subnet’s UID space and emissions are partitioned to accommodate different asset classes, using a ratio that can be adjusted based on demand and profitability of each asset class. For example, a ratio of 0.6 assigned to US stocks means a fixed 60% the subnet’s emissions go to all miners who focus on US stocks. Roughly speaking, we only need the top ~10 strategies from each asset class. A UID space of ~250 has plenty of room to support many asset classes.

## Why is the subnet better than the conventional approach?

**Overview**: The subnet’s key to success is twofold: 1. Wisdom of the crowd – a network of expert quants and analysts that spans the globe; 2. The fundamental math and science underpinning the Nobel Prize winning concept of MPT and CAPM.

A typical hedge fund consists of a tight group of experts that produce the fund’s actionable investment strategy. The organizational nature can have limits in scope, human bias, and group think. 

Our decentralized approach can bring all these experts together, each independently offering their unique expertise, perspective, and investment insights. With their competing strategies, the best come out on top.

As both Bittensor and the subnet continue to mature and gain mainstream attention, we expect to attract an increasing number of quants and analysts from every corner of the world.

In investing, beginners chase returns, while pros manage risks. A return without the context of risk is meaningless, or even dangerously misleading. This is the core concept of MPT/CAPM. It can also be summarized in one word: diversification. It’s a bit oversimplified but gets the point across intuitively.

Imagine two stocks, or for that matter two investment strategies. They both go up in the long-term. But in the short-term, when one stock/strategy goes up, the other goes down, and vice versa. This is called negative covariance in statistics. Investing in each stock/strategy individually can be profitable but with associated risks. A better approach is to invest in both stocks/strategies as a diversified portfolio. The return remains the same but the risk becomes much lower.

Negative covariance can also be artificially introduced with carefully planned shorting. The holy grail of MPT/CAPM is to construct a portfolio that reduces covariance and stays on the “efficient frontier”.

Following the same principle, our final ensemble strategy constructed from the top miner strategies will have improved return with reduced risk. This is our advantage over the conventional investment approach. The devil is in the details of course. The process involves heavy modeling, back testing, and forward testing. But the general concept remains.

## How does the subnet generate revenue?

**Overview**: Main sources of revenue include asset management fees and subscription fees from 3rd parties. In the beginning phase, we capture alpha from the staking market. In later phases, we expand to US equities, forex, and commodities. We plan to onboard clients both in and outside the US, leveraging Interactive Broker’s institutional program with a worldwide reach. We can also generate direct profit by investing with our own private funds. A portion of revenues can flow back into the subnet to reward miners and sustain growth.

The subnet launched with Tao/Alpha staking. After 5 months (as of this writing) of improvements, staking is here and now. The staking market is young and inefficient, offering ample opportunities. We’ve seen a clear alpha edge from our miners’ strategies, and we’re in the process of bringing the opportunities to market, leveraging an exclusive partnership with TrustedStake. Major efforts include premium newsletter contents, an ETF subnet in the making, as well as trading with our own Tao wallets, all powered by the subnet’s actionable intelligence.

However, the staking market is relatively small. The total market cap of all subnets is currently a few hundred million USD. A staking portfolio of just a few million will start to influence the market and see degradation of the alpha edge.

On the other hand, our current strategy timeframe of active rebalancing every few days is a great fit for a young and fast-moving staking market. Following link is our recent X post that showed a comparison chart between a processed ensemble strategy – a simple average without any optimization, vs the single top miner strategy. With both strategies’ risk% normalized to 1%, the ensemble strategy doubles the return of the single strategy.

https://x.com/Investing88ai/status/1970171234467094689

The subnet launched phase II US stocks two months ago. The time window is too short to draw any meaningful conclusion, but preliminary results have been very promising. Assuming our miners and Bittensor continue to deliver, we will start acquiring clients when the clear alpha edge emerges. The initial focus will be onboarding retail investors and high-networth individuals in the US and the APAC region, leveraging Interactive Broker’s institutional program as our frontend AUM app. The app is currently in the alpha development stage and moving towards beta. Institutional clients will follow next, which is a slower process and requires a proven track record in a longer time window. But once the ball starts rolling, institutional adoption rate can potentially be faster than individuals.

The US equity market stands at roughly 100 trillion USD. With our current strategy timeframe of active rebalancing every few days and main focus on large-cap stocks and ETFs, the portfolio size or AUM ceiling can approach hundreds of billions before degrading alpha.

The asset management fees in the US are tightly regulated on the retail client side, and much less so on high-networth and institutional clients, allowing performance-based fees. Roughly speaking, the annual fees generated can reach double digit percentage of the baseline AUM size.

The subscription fee model can have a lower barrier of entry and faster adoption. Once we see the clear alpha edge, we plan to offer free trial to 3rd parties. The key is painless onboarding of CIOs with a turnkey solution that seamlessly integrates with their existing infrastructure. The simplicity of our design here provides a clear advantage. The annual fees generated will be a significant portion of the 3rd party’s revenue/profit.

When a sustainable revenue stream is established, a portion of it will be injected back into the subnet on a regular basis, as a bonus reward to our existing miners and means to attract future miners. Within the community, the subnet’s alpha tokens will function as access tokens to gated contents, driving value back to the subnet.

The subnet will launch phase III forex as well as commodities following our roadmap that has been aggressively delivered:

https://github.com/mobiusfund/investing?tab=readme-ov-file#Roadmap