L. E. Blose, V. Gondhalekar and A. Kort, stated back in 2017 systematic deviations between opening and closing prices of gold-related assets. Based on their findings, this work in hand investigates the investment hypothesis if overnight returns of gold-related Closed Ended Mutual Funds (CEF), Exchange Traded Funds (ETF) and gold mining company stocks yield significant positive returns for suitable investment strategies. The obvious investment strategy, to repeatedly invest the whole portfolio in those funds at their closing price and immediately selling them at the next market open, could have been affirmed in the absence of slippage and even outperformed the benchmark model. Poorly, this last-mentioned simplifying assumption turned out to be crucial for the success of such trading behavior within this research. In particular, the natural illiquidity of those assets is suspicious to serve as a key factor for the underwhelming performance within the backtest environment. As a result, this investment strategy cannot be recommended as it is highly likely to fail under realistic circumstances.

For their research, Blose et al. analyzed the COMEX gold front futures contract, the gold spot market (London Fix), several gold mining company stocks and ETFs and CTFs. The specific time span considered July 1985 through June 2015. Despite just a few outliers, all observed assets yielded positive results even within bear markets and under consideration of transaction costs of up to 10 $ per transaction. This work at hand considered six ETF’s/CTF’s and two gold mining company stocks from the timespan of January 2009 through January 2019. In addition, the timespan had also been split into 2009 through June 2015 and July 2015 through January 2019. This extra effort seemed plausible as July 2015 marks the date out of consideration by Blose et al. and also the end of the open-outcry method for the COMEX gold futures open outcry trading. Anyway, by eyeballing, the results do not point towards structural breaks at this date. First off the statistical properties of the assets and of the overnight returns had been calculated. This includes a test for normality (Jarque Bera), which had to be rejected also for logged variables. Finally, the total returns of the strategy, if applied to each asset on its own, had been compared to the revenue of the Buy & Hold strategy, outperforming it. The results are supported by plots.
To reduce conflicts regarding the execution of trades, the assets in consideration for the backtest had been reduced to GLD, IAU and NUGT. Furthermore, the time span under consideration had been set to January 2014 through 2018. The Portfolio had been split evenly among these assets. The commission fee used had been set to $ 0.001 per share (default). In terms of simplification, the backtest doesn’t take slippage or volume limits into account. A backtest, serving results under Quantopian’s default settings (Slippage: 0.05%; Volume Share Slippage: 0.25%), suggests immense underperformance of the strategy if applied to “non-laboratory” conditions. This concludes no recommendation as a valid strategy under realistic circumstances.