A Quantitative Look at the Gold Overnight Strategy
Today, we do a quantitative look at the gold overnight strategy.
Many markets exhibit what traders call an overnight edge: most gains happen between the market close and the next day’s open.
This phenomenon is well documented in stocks and indices. But does the same pattern exist in gold?
In this article, we explore whether gold has an overnight bias and whether it can be turned into a trading strategy.
The idea behind an overnight strategy is simple:
Buy near the close and sell the next day at the open. The goal is to capture price movements that occur while the market is closed or less liquid.
Why might gold move overnight?
Gold trades almost 24 hours globally, but the ETFs that most traders use have defined trading hours.
For example, the ETF SPDR Gold Shares (GLD) closes at 16:00 New York time and opens again the next day at 09:30.
During the overnight period, global developments can influence prices:
macroeconomic news
central bank policy expectations
geopolitical events
currency movements
Because these events occur outside US trading hours, they can create gaps or price adjustments at the next open.
Testing the overnight edge in gold
To test the idea, we look at the performance of buying the close and selling the next open in gold-related instruments.
Gold miners ETF
First, consider the gold miners ETF VanEck Gold Miners ETF (GDX).
A simple strategy of buying at the close and selling at the next open shows:
average gain per trade: about 0.11%
However, because GDX has a relatively low price and trading costs matter, the edge is likely too small to trade profitably.
If we flip the test and hold only during the day (buy the open, sell the close), the results are negative. The average daily loss is around 0.06%.
This suggests that most of the positive performance occurs overnight rather than during the trading day.
Gold ETF
Next, we test the strategy using SPDR Gold Shares (GLD), one of the most liquid gold ETFs.
The results are interesting:
annual return: about 11.4%
buy-and-hold return: about 11.8%
average overnight gain: roughly 0.04% per trade
In other words, almost all the gains come from the close-to-open session.
If we instead hold only intraday (buy the open and sell the close), the performance is essentially flat, even though GLD has risen more than 976% over the test period.
This mirrors what researchers and traders have observed in equities: the overnight session often contains most of the market’s return.
Can the edge be improved?
The raw overnight edge in gold is small and noisy. On its own, it is probably not tradable after costs.
However, it can potentially be improved with filters.
For example, one experimental strategy we are testing:
Buy at the close
Sell at the next open
Use a few simple filters (including a seasonal component)
Results from testing:
836 trades
average gain: about 0.15% per trade
The trading rules are only available for our paying subscribers (on the website).
That level may be sufficient to trade if transaction costs are low.
A note about trading commodities
Our experience after decades of trading and backtesting is that commodities are generally harder to trade than stocks.
Edges tend to be:
less stable
more sensitive to regime changes
more prone to breaking down over time
This applies to gold and gold miners as well. Any strategy should therefore be carefully tested and continuously monitored.
Conclusion
Gold appears to exhibit a modest overnight bias, similar to what has been documented in the stock market.
Simple tests show that:
most gains occur from close to next open
intraday returns are close to zero
raw edges are small but may be improved with filters
While the basic overnight strategy is probably too small to trade by itself, it provides a useful starting point for developing more robust gold trading systems.