Timing and Price Ratios in Backward-Adjusted Futures Series
Summary
The document explains how to construct a backward ratio-adjusted continuous futures series when open interest indicates a roll from one contract to another. In its corn example, the front contract's open interest falls behind the deferred contract's, but the adjustment is made using the price ratio on the first day the new contract is used. That ratio is then applied to earlier observations, including the prior day's bar.
It also clarifies how to report prior-day open interest: the cited series carries forward the open interest of the contract traded on the preceding day, even after the roll. The worked prices and open-interest observations illustrate the timing convention and reconcile differences from an alternative adjustment. The example describes one vendor's convention; continuous-series construction can depend on roll rules and data definitions, so the procedure should be made explicit when comparing series or evaluating strategies.
Key ideas
- Apply the new-to-old contract price ratio on the day the new contract enters the series.
- Use that ratio to adjust earlier bars in a backward-adjusted series.
- Prior-day open interest can refer to the contract traded on the previous day.
- Roll timing and open-interest conventions affect the resulting continuous series.
Tags
Full text
# Continuous Contract - which bar do I adjust given a roll date? # Continuous Contract - which bar do I adjust given a roll date? I have the following bars around an open interest roll date (all bars stated date, contract, settle, previous day open interest) for CME corn futures. ``` date,symbol,settle,previous_day_open_interest 2018-06-11,ZCN2018,367.25,589007 2018-06-11,ZCZ2018,388.25,587840 2018-06-12,ZCN2018,377.5,566301 2018-06-12,ZCZ2018,398.25,594274 2018-06-13,ZCN2018,376,527519 2018-06-13,ZCZ2018,397,600335 2018-06-14,ZCN2018,363,487860 2018-06-14,ZCZ2018,384.5,604879 ``` Note that the most liquid contract flips from the N to the Z on 6/12. Question - how exactly would I build a backwards ratio adjusted continuous contract on these bars? What have I tried I'd have thought the following: 1) The "signal" to switch from one contract to the other happens on 6/12, meaning the first date I can be on the new contract would be 6/13. 2) I would adjust the 6/12 and 6/11 bars by the ratio of settlement prices on 6/12 (398.26/377.5=1.054967). The 6/12 settlement would now be stated at the back contract value while 6/11 would be truly ratio adjusted. ``` date,settle,previous_day_open_interest 2018-06-11,387.436589403973,589007 2018-06-12,398.25,566301 2018-06-13,397,600335 2018-06-14,384.5,604879 ``` However, I purchased the Stevens continuous roll series, which differed in 2 spots: 1) The ratio used for the 6/11 and 6/12 bars seemed different, and 2) They reported the front contract "previous_day_open_interest" on the 6/13 bar, even though there had already been enough time to move to the new contract on 6/13. ``` date,settle,previous_day_open_interest 2018-06-11,387.7613032,589007 2018-06-12,398.5837766,566301 2018-06-13,397,527519 2018-06-14,384.5,604879 ``` ## Answer by Alex C (score 1, accepted) https://quant.stackexchange.com/a/40465 If the previous_day_open_interest is interpreted to mean "the OI yesterday in the contract that we were trading yesterday" then Stevens choice of 527519 for 2018-06-13 makes sense. In your interpretation you are reporting the OI yesterday for a contract which was not chosen yesterday (i.e. not the one we were trading) and this a different interpretation, which I find less convincing. The "July to December Multiplier", which when applied to [2018-06-13 price of] 376 yields 397 is 1.055851. When this multiplier is applied to the July price of 2018-06-12 namely 377.5 we get 398.5837766 which perfectly agrees with the Stevens figure. So again I agree with Stevens. The multiplier is computed on the day the new contract comes in, namely the 13th, and applied on the days before this.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.