6.6.2 Nearbys and Distortions
Nearbys exhibit certain distortions related to contract expirations. Depending upon the underlier, price standard deviations may rise or fall as the last trade date approaches. Correlations with other prices may also be affected. If futures on the same commodity trade on two geographically separate exchanges, the correlation between the futures prices will tend to decline as the contracts approach expiration. Localized supply and demand imbalances will become evident and it will become increasingly infeasible to transport the commodity from one location to the other in time for delivery. During the final days prior to a contract’s last trade date, distortions can be pronounced as positions are closed out, open interest declines, and liquidity migrates to the subsequent contract.
If futures prices are formed into nearbys, such expiration-related behavior manifests itself as a cyclical pattern in each nearby. This may be especially pronounced in the first nearby. One way to mitigate this effect in the first nearby is to construct nearbys by rolling over contracts prior to each last-trade date.
The rollover date for a set of nearbys is the date when each nearby is switched to reflect prices of the subsequent futures contract. The last possible rollover date is the last trade date of the expiring contract, but earlier dates, such as the first day of the expiring contract’s delivery month, are possible.
The earlier we make a rollover date, the more we can mitigate expiration-related distortions in the first nearby. If we set our rollover date 2 weeks prior to the last-trade date, prices during each contract’s last 2 weeks of trading will be discarded. They will appear in none of the nearbys.
Early rollover of nearbys raises a problem if a portfolio holds futures contracts to expiration. We don’t want to stop recording data for a soon-to-expire contract while it is still held by the portfolio! A simple solution is to track data right until contracts expire. Only after they expire do we go back and modify nearbys to reflect the earlier rollover date.
The judicious selection of a rollover date is just one technique that addresses a specific form of distortion in the first nearby. Other distortions may remain across all nearbys. Exhibit 6.7 is a graph of the daily second nearby for IPE natural gas futures. The first trading day of each delivery month is used as a rollover date. The graph covers the second half of 1998.
The nearby prices exhibit a pattern. This is more pronounced in Exhibit 6.8, which shows daily returns for the same nearby. The six largest (positive or negative) returns in Exhibit 6.8 are evenly spaced in time. Indeed, each occurs on the first day of a month. These returns do not reflect market events. They are artifacts of how the nearby was constructed. IPE natural gas futures expire monthly. Every month, each nearby rolls over to the next contract. Because of seasonality effects, there is usually a price jump between consecutive contracts. These are incorporated into each nearby. The monthly price jumps are evident in Exhibit 6.7. They cause the large evenly spaced returns in Exhibit 6.8.
Such artifacts can be addressed in various ways. One approach is to remove artifact price jumps by directly adjusting nearby prices upward or downward as needed prior to each rollover date. Prices prior to the most recent rollover date are adjusted by the price difference between the new and old contracts. Prices prior to the previous rollover date are adjusted by that amount plus the price difference between the new and old contracts for that rollover date, etc. Nearbys constructed in this manner are called price-adjusted nearbys. This solution eliminates artifact price jumps. However, the accumulation of price adjustments may cause older nearby prices to stray significantly from their true levels. Indeed, if we go back far enough, nearby prices may even become negative.
Alternatively, we may remove artifact returns by directly adjusting those returns that straddle a rollover date. The artifact returns arise because, when a return straddles a rollover date, it is calculated using a price from the old contract and another price from the new contract. The solution is to calculate such returns exclusively from the new contract’s prices. Based upon these returns, and returns that can be calculated from the rest of the nearby’s prices, we obtain a series of returns for the nearby. We then convert these into what is called a return-adjusted nearby by starting with the most recent futures price and applying the most recent return to obtain the previous price. Working backwards in this manner, we obtain an entire series of prices. Prior to the most recent rollover date, these may not reflect actual price levels or actual price changes. However, prices will not become negative, and historical returns will be correctly represented.
Through the judicious selection of a rollover date and the use of price-adjusted or return-adjusted nearbys, we can mitigate the most obvious distortions associated with nearbys. More subtle distortions can remain. Primarily, these will be cyclical patterns in standard deviations or correlations. If contracts expire frequently, say every month, these will be modest. With bimonthly or quarterly expirations, they may be more severe. Such distortions are not easily mitigated.