# Stablecoin Strategy Analysis

## 1. Project Overview

This project studies whether **stablecoin-related fundamentals** can help explain the daily yield earned from a simple DeFi yield strategy. Instead of building a complex trading system, the goal is to:

* Pull on-chain fundamentals (interest rates, supply, TVL, prices).
* Align everything on a daily frequency.
* Define a simple stablecoin yield strategy.
* Run **single‑factor regressions** to see which variables move with the strategy’s returns.

The strategy we evaluate is:

> **Strategy:** Hold sUSDe and earn its daily yield.

Since sUSDe is a yield-bearing stablecoin, the daily yield is clean, stable, and easy to regress on fundamentals.

Our regressions answer the intuitive question:

> **What on-chain variables help explain daily variations in stablecoin yield?**

---

## 2. Data Sources and Variables

We used several datasets exported from Dune:

### Aave USDC / USDT Interest Rates

* **USDC supply APY** (Aave v2 + v3)
* **USDT supply APY** (Aave v2 + v3)

These represent the risk‑free-ish yield level in DeFi for dollar assets.

### sUSDe (Ethena)

* **Daily share price**
* **Daily yield**
* **Cumulative supply**

### Stablecoin Prices

* USDC, USDT, USDe

### Stablecoin Supply

* Circulating supply for USDC / USDT / DAI

### Aave TVL

* Aggregate USD value locked in Aave v2 + v3

### BTC / ETH Prices

* Daily close price
* Daily returns

---

## 3. Strategy Definition

Our dependent variable (**strategy_ret**) is:

* sUSDe **daily yield** (percentage change in share price).

This represents the passive return from holding sUSDe.

We do *not* build a leveraged borrow‑lend loop or complex delta‑hedged structure; instead we focus on:

> "Which macro / stablecoin / DeFi variables statistically explain yield behavior?"

---

## 4. Regression Structure

For each independent variable X, we run:

strategy_ret ~ X

This gives us a clean one‑factor sensitivity view.
We are not building a multivariate model because many variables are collinear and we want to interpret relationships one at a time.

---

## 5. Regression Results (Intuitive Explanation)

Below is what each regression tells us in plain language.

### 5.1 USDC Supply APY → sUSDe Yield

* **R² = 0.207**
* **Coefficient ≈ +0.0049** (highly significant)

Interpretation:

* When **Aave USDC yield goes up**, sUSDe yield tends to increase.
* This makes sense: both reflect the broader cost of capital and stablecoin demand.

### 5.2 USDT Supply APY → sUSDe Yield

* **R² = 0.244**
* **Coefficient ≈ +0.0055**

Interpretation:

* USDT APY explains even *more* of sUSDe’s movement.
* USDT demand spikes often correlate with on-chain liquidity crunches—raising yields across ecosystems.

### 5.3 sUSDe Yield → sUSDe Yield

* **R² = 1.000** (as expected)
* This is just a self-check: regressing the variable on itself gives perfect fit.

### 5.4 USDC Circulating Supply → sUSDe Yield

* **R² = 0.124**
* **Coefficient: Negative**

Interpretation:

* When **USDC supply expands**, sUSDe yields tend to slightly decrease.
* More stablecoin supply → more liquidity → lower returns across systems.

### 5.5 DAI Circulating Supply

* **R² = 0.042**
* Weak explanatory power.

Interpretation:

* DAI supply changes are less tied to broader stablecoin funding markets.

### 5.6 Aave TVL → sUSDe Yield

* **R² = 0.145**
* **Coefficient: Negative**

Interpretation:

* Rising TVL means capital is plentiful → yields compress.

### 5.7 BTC Returns

* **R² = 0.001** (zero relationship)

Interpretation:

* BTC price volatility has no effect on stablecoin yields.

### 5.8 ETH Returns

* **R² = 0.006** (essentially zero)

Interpretation:

* ETH’s market trends don’t influence stablecoin yield mechanics.

---

## 6. High-Level Takeaways

1. **Stablecoin yields are driven by liquidity conditions, not crypto prices.**

* BTC/ETH have near‑zero explanatory power.

2. **Aave supply yields are strong predictors of sUSDe yield.**

* Both reflect demand for leverage and stablecoin borrowing.

3. **Liquidity expansion reduces yields.**

* Higher USDC supply or higher Aave TVL → lower returns.

4. **Stablecoin ecosystems are interconnected.**

* USDT APY (often more volatile) is particularly predictive.

Overall, the project shows that **stablecoin yield behaves like a money‑market rate inside DeFi**—rising when liquidity tightens and falling when liquidity expands.

---

## 7. Summary for Presentation / Report

* We built a dataset merging Aave yields, stablecoin supply, TVL, and crypto prices.
* We defined a simple stablecoin yield strategy using sUSDe.
* We ran single‑factor regressions to understand sensitivity.
* The strongest signals come from **USDC and USDT APY**, and from overall liquidity measures.
* Crypto market volatility has little or no influence.

This gives an intuitive picture of how **on-chain money markets behave similarly to TradFi money markets**.

---

## 8. Next Steps (Optional Extensions)

If we wanted to extend the work:

* Try multi‑factor regressions
* Include lagged terms (1–7 days)
* Analyze cross‑chain flows
* Include CEX stablecoin funding rates
* Try to predict sUSDe yield ahead of time

But for the purposes of this project, the single‑factor insights already form a clean narrative.