Role Of Rebalancing In DeFi Portfolio Management | by Max Yampolsky

Is it profitable to rebalance DeFi positions and how frequently?

In the contemporary digital asset landscape, Decentralized Finance (DeFi) has evolved as a transformative ecosystem, enabling novel means of financial engagement via autonomous smart contracts. Within the expanding universe of DeFi protocols, yield farming has emerged as a popular and potentially profitable practice for crypto-asset investors. Yield farming refers to the strategic deployment of crypto-assets in various liquidity pools to maximize return on investment, an endeavor with a set of complexities and challenges that requires effective and data-driven management strategies.

This research aims to provide insight into the performance and sustainability of various DeFi yield farming portfolios, with a particular emphasis on differing rebalancing periods. By analyzing portfolio performance over a 12-month period, we examine the impact of weekly, biweekly, and monthly rebalancing strategies compared to a fixed portfolio with no rebalancing.

Our investigation considers both the potential for high yield and the inherent risks associated with these strategies. Furthermore, we incorporate the analysis of Arbitrum gas fees, which are integral to transactions and interactions with DeFi protocols and can significantly influence overall portfolio performance. However, our focus will not extend to the phenomenon of impermanent loss, a unique risk associated with liquidity provision in DeFi, which while crucial to the broader conversation, falls beyond the scope of this specific research.

Through a systematic backtesting approach, we will establish a comprehensive understanding of how different rebalancing periods can impact the overall performance of a DeFi yield farming portfolio. The findings of this research are intended to guide crypto investors, financial advisors, and DeFi enthusiasts in their portfolio management strategies, ultimately promoting greater efficiency and profitability in the fast-paced and ever-evolving DeFi space.

This research is premised on the expectation that yield farming portfolios with a higher frequency of rebalancing could potentially outperform those that are fixed or infrequently rebalanced. The logic behind this hypothesis is that, by adjusting the portfolio’s allocations more often, an investor can better navigate the rapidly shifting DeFi landscape, capitalizing on the high-yielding opportunities that emerge.

However, it is also proposed that the benefit of frequent rebalancing may diminish or even reverse for portfolios with lower investment amounts. This assumption is based on the economic principle that pits fixed costs against variable gains. In the context of DeFi yield farming, Arbitrum gas fees associated with each rebalancing act as the fixed cost, and the yield from various liquidity pools represents the variable gain. For smaller portfolios, these fixed costs could, over time, outweigh the benefits derived from capturing fleeting high-yield opportunities, thereby eroding overall performance.

To empirically test these hypotheses, we will employ a backtesting methodology using historical data from the top 50 TVL pools on Arbitrum, excluding Uniswap v3 pools due to their significant impermanent loss component. This data will be utilized to calculate the optimal portfolio for each week over a 12-month period, under different rebalancing frequencies: weekly, biweekly, monthly, and fixed. The subsequent comparison of these results will provide empirical evidence to support or refute our initial expectations, contributing to a more robust understanding of the dynamics at play within DeFi yield farming portfolio management.

In conducting this research, several assumptions will be applied to provide structure to the analysis, as well as to narrow the scope to a manageable domain. The assumptions guiding our study are as follows:

Investment size: Initially we will assume an investment size of $10,000, we will later explore the effect of lower investment size and examine the effect of investment size on profitability.

Portfolio Optimization: We will use the Modern Portfolio Theory (MPT) as our guiding principle for portfolio optimization. Thus, we will use mean-variance optimization to construct our portfolios.

Transaction Costs: The cost of Arbitrum gas fees for each rebalancing operation will be considered from self testing and then pro-rating to an average using data obtained from Dune Analytics, in a precautionary overestimation. In this study, it is assumed that every asset in the portfolio will require four separate transactions — two approvals (one for the underlying token conversion and one for liquidity provision) and two actual transactions (one for asset conversion and one for liquidity provision). The cost of each approval is assumed to be $0.225, while the cost of each transaction is assumed to be $0.4. This results in a total cost of $1.25 per asset per rebalancing period.

Portfolio Constraint: We will limit the number of asset in the portfolio to four. This constraint is based on the notion that rebalancing strategies are typically short-term and having too many assets can add unnecessary complexity and transaction costs, potentially diminishing the effectiveness of frequent rebalancing. Furthermore each pool has a max weight of 30%.

‍Impermanent Loss: In this research, we will not consider impermanent loss. Although impermanent loss can be a significant factor affecting the profitability of liquidity provision in certain DeFi protocols, it will be excluded from this particular study to isolate the impact of rebalancing frequency and gas fees on portfolio performance.

By applying these assumptions, we aim to create a structured framework within which we can conduct our analysis and generate meaningful insights into the implications of different rebalancing strategies on DeFi yield farming portfolios.

The following approach was utilized in our research to rigorously examine and compare the performance of DeFi yield farming portfolios under various rebalancing strategies:

Dataset: We used a dataset comprising the top 50 TVL pools on Arbitrum, excluding Uniswap v3 due to their significant impermanent loss component. Our data is specifically looking at the time period 25th April 2022 to 25th April 2023. Each included pool is from a protocol with a risk score of 6/10 or higher, as assessed by our proprietary risk evaluation method. It is crucial to note that not all pools have return data extending back to the start of the 12-month period; thus, the set of available assets for portfolio optimization expands as the backtest progresses. Rather than a limitation, this situation realistically mirrors the dynamic nature of the DeFi space, where new investment opportunities constantly emerge. Full list of included pools can be found in the bibliography.

Portfolio Construction: To construct the portfolios, we segmented the 12-month period into individual weeks. For each week, we employed mean-variance optimization to construct the maximum return portfolio, setting the number of assets in the portfolio to four. This constraint was applied for reasons of consistency, simplicity, and comparability across the varying rebalancing periods.

Return Computation: Returns were calculated at the end of each week based on the optimal portfolio from the previous week, as future returns are, of course, unknowable. For instance, for the weekly rebalance portfolio, the week 10 portfolio tracks the return of those assets in week 11. Essentially, at time T, we tracked the returns of the portfolio from time T-1. When rebalancing occurs, gains are compounded; conversely, when no rebalancing takes place, gains are simply added without compounding.

Gas Fees and Compounding: The gas fees were deducted at the start of any rebalancing period and incorporated within the compounding calculations. This approach allows for a realistic evaluation of the net returns that can be expected after considering the transaction costs associated with portfolio rebalancing.

This methodology provides a robust and realistic framework to evaluate the impact of varying rebalancing frequencies and transaction costs on DeFi yield farming portfolios’ performance over a 12-month period.