Application

Earning with Categories based Portfolio

How to earn with Categories based Portfolio - AI-driven Markowitz optimization across crypto categories for sector-level diversification and interest-bearing strategies.

Last updated:

Categories based Portfolio lets you build a Markowitz-optimized portfolio by describing what you want in plain language. The AI returns an allocation across crypto categories - DeFi, Layer 2, Stablecoins, and others - with risk metrics and a correlation matrix. Earnings come from portfolio appreciation, sector-level diversification, and optionally from yield-generating categories.

How returns are generated

The optimizer works at the category level rather than individual tokens, treating each crypto sector like an asset class. This produces two sources of return:

  • Price appreciation - the weighted performance of assets across the selected categories
  • Diversification benefit - categories with low or negative correlation reduce portfolio volatility without proportionally reducing return, improving the Sharpe Ratio of the overall portfolio

For interest-bearing strategies, the optimizer can focus on yield-generating categories - DeFi lending, liquid staking, and similar sectors - where the underlying assets produce income in addition to price exposure. Use the phrase "interest bearing" in your prompt to steer the optimizer toward these segments.

Prompt strategies for earnings

The quality of the portfolio depends on how you frame your prompt. Some approaches that target specific earning objectives:

  • Broad growth portfolio - Create a long only portfolio with 5 crypto categories - balanced across major sectors, optimized for Sharpe Ratio
  • Yield-focused - Create a long only portfolio with 5 interest bearing categories - focuses allocation on income-generating segments; useful when the goal is recurring yield rather than pure capital growth
  • Diversification check - Show me the correlation matrix for top 5 categories - reveals which categories move together before you commit; genuine diversification requires low pairwise correlation

You can refine the output conversationally - ask for more categories, swap one sector for another, or request metrics for a specific allocation.

Reading the output

For each proposed portfolio, the AI returns:

Output What to look at
Weights How much of the portfolio goes to each category
Expected annual return The optimizer's return estimate based on historical data
Volatility Portfolio-level standard deviation - lower is more stable
Sharpe Ratio Return per unit of total risk - higher is more efficient
Sortino Ratio Return per unit of downside risk - better measure for asymmetric crypto returns
Correlation matrix Pairwise correlations - categories near 0 or negative provide genuine diversification

A portfolio with a Sharpe Ratio above the individual category average is outperforming its components on a risk-adjusted basis - that is the diversification working.

Rebalancing

Category weights drift as markets move. Rebalance monthly: run a fresh prompt with the same objective, compare the new allocation to your current holdings, and adjust. Monthly rebalancing is a practical cadence - it captures most of the efficiency benefit without excessive transaction costs.

Blog articles

Further info