Quantifying Overnight and Weekend Gap Risk: A Probabilistic Approach

The Illusion of Continuous Markets

For traders accustomed to intraday price action, the market can appear as a continuous stream of data. However, this is a dangerous illusion. Financial markets are fundamentally discrete, punctuated by closures—overnight and over the weekend. During these periods, information continues to accumulate, but liquidity vanishes. The result is the gap: a discontinuity between the previous session's close and the current session's open. Gap risk is the exposure to an adverse price movement during these non-trading periods, a risk that can render traditional stop-loss orders useless.

Effective gap risk management begins with quantification. Simply acknowledging the risk is insufficient for professional traders. We must measure it, model its probabilities, and integrate it into our overall risk framework. This involves moving from a deterministic mindset (e.g., "I'll place a stop at X") to a probabilistic one (e.g., "There is a Y% probability of a gap larger than Z% on this instrument").

Statistical Analysis of Historical Gaps

The foundation of quantifying gap risk is the statistical analysis of historical data. The first step is to define and identify a gap. A common definition is any instance where the current session's opening price is outside the high-low range of the previous session. For an upward gap, Open(t) > High(t-1). For a downward gap, Open(t) < Low(t-1).

Once identified, we can build a distribution of gap sizes. This requires a sufficiently long dataset covering various market regimes. For a specific instrument, a trader should calculate:

• Gap Frequency: What percentage of trading days exhibit a gap? This should be analyzed separately for overnight (weekday) gaps and weekend (Monday open) gaps, as their characteristics differ significantly.
• Gap Magnitude Distribution: What is the statistical distribution of gap sizes? This is rarely a normal distribution. More often, it exhibits "fat tails," meaning extreme gap events are more common than a normal distribution would predict. Traders should calculate the mean, median, standard deviation, skewness, and kurtosis of gap sizes.
• Conditional Gap Analysis: How do gap characteristics change based on market conditions? For example, are gaps larger or more frequent when the VIX is improved? Do they correlate with specific economic announcements or geopolitical events?

Let's consider a practical example. A trader analyzing the SPY ETF might download 20 years of daily data. They would write a script to iterate through each day, comparing the open to the previous day's high and low. For each identified gap, they would calculate the percentage change: (Open(t) / Close(t-1)) - 1. After collecting thousands of these data points, they can plot a histogram to visualize the distribution and calculate the key statistical moments. This analysis might reveal that while the average overnight gap is small, there is a 2% chance of a gap greater than 3% occurring over a weekend, a piece of information important for sizing positions held into the close on Friday.

Modeling Gap Volatility with GARCH

While historical analysis is useful, it is static. Market volatility is dynamic. The probability of a large gap tomorrow is not the same as it was last year. To account for this, we can employ more advanced time-series models, such as the Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model.

GARCH is designed to model and forecast volatility clustering—the observation that periods of high volatility are followed by more high volatility, and periods of low volatility are followed by more low volatility. Since large gaps are a function of volatility, GARCH can be adapted to forecast the potential magnitude of future gaps.

A GARCH(1,1) model, a common variant, forecasts future variance (the square of volatility) based on a long-run average variance, the previous period's variance forecast, and the previous period's squared return. In the context of gap risk, we are not interested in the intraday returns but the overnight return (the gap itself).

The process would be:

1. Create a Time Series of Gap Returns: For each trading day, calculate the overnight return: r_t = (Open_t / Close_{t-1}) - 1.
2. Fit a GARCH Model: Use a statistical package (e.g., in Python or R) to fit a GARCH(1,1) model to this time series of gap returns.
3. Forecast Next Period's Variance: The fitted model provides parameters that can be used to forecast the conditional variance for the next non-trading period (σ^2_{t+1}).
4. Convert to Volatility: The square root of the forecasted variance gives the expected volatility (σ_{t+1}), which represents the one-standard-deviation expectation for the next gap's size._

This forecasted volatility is a effective tool. For example, if the GARCH model forecasts an overnight volatility of 1.5% for an upcoming earnings announcement, a trader can use this to set more intelligent profit targets or to price the cost of an options hedge. It allows for a dynamic adjustment of position size. On a quiet day with forecasted gap volatility of 0.3%, a trader might take on a larger position. Ahead of a major event with forecasted gap volatility of 2.0%, they would reduce their size accordingly.

Pricing Gap Risk with Options

Options markets provide a market-implied measure of expected volatility, which inherently includes gap risk. The price of an option is a function of several variables, including implied volatility (IV). A higher IV means the market is pricing in a greater probability of a large price swing, including an overnight or weekend gap.

Traders can use options to not only hedge gap risk but also to quantify its market-perceived cost. For instance, consider a stock trading at $100 ahead of a weekend. A trader wants to quantify the market's expectation of a weekend gap. They could look at the price of an at-the-money straddle (buying both a call and a put) that expires the following Monday. The cost of this straddle represents the "breakeven" move the market is pricing in. If the straddle costs $3.00, the market is implying that there is a significant probability of the stock opening above $103 or below $97 on Monday.

This provides a real-time, forward-looking measure of gap risk. A trader can compare the GARCH-forecasted volatility to the options-implied volatility. If the GARCH model suggests a 1.5% gap is likely, but the options market is pricing in a 3% move, it indicates that the market has a higher perception of risk than the historical data alone would suggest. This could be due to a known upcoming event not captured in the historical price series. This discrepancy itself can be a source of trading opportunities, such as selling volatility if one believes the market is over-pricing the risk.

By combining historical statistical analysis, dynamic volatility forecasting with GARCH, and the market-implied view from options, a trader can build a robust, multi-faceted framework for quantifying gap risk. This transforms risk management from a guessing game into a quantitative discipline, providing a significant edge in navigating the market's inherent discontinuities.