Cointegrated Pairs Trading with Gold Tracking ETFs
Author
Ngeh, Rodrick
Term
4. semester
Publication year
2024
Submitted on
2024-02-05
Pages
35
Abstract
Dette speciale undersøger, om en simpel pairs trading-strategi kan give overskud ved at handle to amerikanske børsfonde (ETF’er), der følger guldprisen: SPDR Gold Shares (GLD) og iShares Gold Trust (IAU). Vi analyserer data fra 1. januar 2015 til 25. oktober 2023. Metoden bygger på kointegration, dvs. at vi tester, om de to prisserier bevæger sig sammen over tid. Konkret anvender vi ADF- og Johansen-testene til at vurdere stationaritet og kointegration. Strategien udvikles på en træningsperiode og evalueres derefter i en separat backtestperiode for at se, om resultaterne holder på nye data. Resultaterne viser, at GLD og IAU er kointegrerede og stationære i de testede perioder. Afkastmæssigt præsterer træningsperioden bedre end backtesten, selv om backtesten stadig giver acceptable, annualiserede afkast. Dette rejser spørgsmål for den effektive markedshypotese, som ville forvente, at sådanne muligheder hurtigt udlignes. En følsomhedsanalyse peger på robusthed, da scenariernes resultater ligger tæt på backtestens. På trods af studiets begrænsninger konkluderes det, at pairs trading fortsat kan være profitabel, men at effektiviteten falder, når strategien anvendes på nye data.
This thesis examines whether a simple pairs trading strategy can be profitable using two U.S. gold-tracking exchange-traded funds (ETFs): SPDR Gold Shares (GLD) and iShares Gold Trust (IAU). We analyze data from January 1, 2015 to October 25, 2023. The approach relies on cointegration, meaning we test whether the two price series move together over time. Specifically, we apply the ADF and Johansen tests to assess stationarity and cointegration. The strategy is built on a training period and then evaluated on a separate backtesting period to see if results hold on new data. The findings show that GLD and IAU are cointegrated and stationary in the tested periods. In terms of returns, the training period outperforms the backtesting period, although the backtest still delivers acceptable annualized returns. This pattern raises questions for the efficient market hypothesis, which would expect such opportunities to be competed away. A sensitivity analysis indicates robustness, as scenario outcomes were close to the backtest results. Despite the study’s limitations, the conclusion is that pairs trading remains profitable, though its efficiency declines when applied to new data.
[This summary has been rewritten with the help of AI based on the project's original abstract]
Documents