Strategy

1 Pair Trading Strategy Overview

Trading Setup

• Trading Time: Past three years
• Data Frequency: Daily resolution
• Hedge Ratio Window: 20 days
• Cointegration Test Window: 60 days
• Trading Session Timeout: 20 days

---

2 Strategy Logic

1. Calculate Hedge Ratio
   Compute the hedge ratio \(\beta_i\) based on the past 20 days of log prices: \[ \beta_i = \frac{\text{Cov}(\log P_{\text{RIVN}}, \log P_{\text{TSLA}})}{\text{Var}(\log P_{\text{TSLA}})} \] Then using hedge ratio to hold a notional value of $10000 position of both stocks

2. Compute Spread and Z-Score \[ \text{Spread}_t = \log(P_{\text{RIVN}, t}) - \beta_i \cdot \log(P_{\text{TSLA}, t})\\ \] \[ z_t = \frac{\text{Spread}_t-\text{mean}({\text{Spread}_{t-20:t}})}{\text{std}(\text{Spread}_{t-20:t})} \]

3. Cointegration Test
   Test 60-day rolling window of log prices for cointegration using the ADF test.

4. Z-Score Based Trading Rules
   If cointegration is statistically significant \(p<0.05\):

   • Entry: if \(|Z| < 1\)
   • Exit: if \(|Z| > 2.5\)

   Here we long TSLA and short RIVN.

5. Trading Session Timeout
   Exit trades when:

   • trade lasting over 20 days, and
   • \(|Z| < 1\) and \(p<0.05\) (still reverting)

3 Blotter

4 Ledger

5 Graphs

5.1 TSLA vs. RIVN Stock Log Prices

5.2 Rolling ADF Test on Spread (window=60)

5.3 PnL Over Time

5.4 Strategy Realized Returns Sensitivity

5.4.1 Strategy vs Long TSLA

5.4.2 Strategy vs Short RIVN

5.4.3 Summary

Most of the returns are attributable to alpha, not passive exposure to either stock. Our strategy achieves near market-neutrality with respect to RIVN, and gains a moderate positive sensitivity to TSLA’s upside — consistent with our thesis. This confirms our model successfully isolates pricing divergences and avoids directional bias.

6 Strategy Performance Summary

Here’s the performance overview of the TSLA-RIVN Divergence Strategy over the past 3 years:

Metric	Value
Number of Trades	14
Average Return per Trade	6.79%
Volatility of Returns	15.32%
Geometric Mean Return/Trade	5.93%
Sharpe Ratio (per trade)	0.44
Average Trades per Year	4.67
Average Holding Period	26.93 days
Annualized Return	71.38%
Annualized Volatility	46.85%
Annualized Sharpe Ratio	1.52
Total PnL	$12,902.38

---

7 Reflection and Future Work

1. TSLA ≠ Just an EV Company
   • One important thing for this strategy is understanding that Tesla isn’t just building electric vehicles as it’s also involved in AI, robotics, and energy.
   • The stock tends to move based on Elon Musk’s public actions, news headlines, and tech market sentiment, not just EV fundamentals.
   • In the earnings report 1Q of 2025 showed a signficant decline of 71% of net income and revenue falling 9% a year, but the stock actually went up!
   • As such, Elon’s fanbase (the Elon Musk Factor) makes TSLA behave differently from a company like RIVN, which is much more focused on vehicle production alone.
2. Reversing the Traditional Pair Trade
   • In most pair strategies, traders bet two related stocks will converge back together after moving apart.
   • Instead of betting on convergence, we anticipate divergence between TSLA and RIVN when they appear temporarily correlated.
   • This contrarian approach takes advantage of market misperceptions.
3. Managing Risk and Timing
   • Although the strategy delivers strong returns, the high per-trade volatility demands disciplined risk management and patience to wait for valid signals.
   • To control this, we set strict entry and exit rules, limit trade duration to 20 days, and only enter trades when there is clear cointegration.
4. Expanding the Strategy Universe
   • Currently, the strategy is based on a single pair.
   • Future work includes adding other EV stocks (e.g., LCID, BYDDF) and legacy automakers transitioning to EV (e.g., Ford, GM) to scale opportunity and diversify risk.
   • We may also use clustering algorithms to group similar stocks based on historical price behavior and fundamentals, helping us find more profitable and uncorrelated pair combinations.

---