1989issue C091-6
Seasonal dates, confirmation, and reconstructed spread bars
Desktop commodity-spread research treated both legs and the difference series as one problem. This case study follows a historical workflow in which a listed seasonal window, a live confirmation day, and a reconstructed spread bar could each start a trade or force abstention.
- Commodity-spread work treated both futures legs and the difference series together, combining contract-level data with a spreadsheet, a charting program, and a system tester.
- Seasonal candidates were first read from a monthly catalog of listed windows and risk fields, then narrowed on past graphs, with further review limited to calendar spreads whose three prior years each showed a favorable run-up.
- Listed dates were not treated as fills. A confirmation-day rule required at least one later session to move the spread in the anticipated direction, and an immediate entry at extreme width was recorded as a violation.
- Spread-reconstruction differenced the highs, lows, and closes of the two legs into a synthetic bar series so RSI, directional movement, stochastics, and oscillators could run on the relationship itself.
Two legs plus the difference series
Desktop commodity-spread work treated the problem as both legs plus the difference series. Contract-level futures data were combined with a spreadsheet, a charting program, and a system tester.
That two-leg review kept width, direction, and timing in one view. Each commodity spread was a long-short pairing of related futures contracts. A calendar spread used different delivery months in the same market or in closely related markets.
Catalog windows and graph filters
Seasonal candidates were first read from a monthly historical catalog. The catalog supplied listed entry and exit dates plus risk fields such as margin, stop-loss levels, share of profitable years, average drawdown, and worst loss. A smaller subset was then inspected on past graphs.
Further review was limited to calendar spreads whose three prior years each showed a favorable run-up on the historical spread graph. The listed entry-to-exit range was used as a seasonal window: a hypothesis for when a spread relationship tends to change, not a complete execution plan.
A sale of a wide calendar spread
A September Eurodollar versus March T-bill calendar spread was specified as a sale of a wide relationship expected to narrow. The listed window ran from December 31, 1987 through February 7, 1988, with horizontal entry and exit marks on both legs and on the spread.
An early exit rule closed a spread once a profit objective had been reached and price then moved against the position. That rule was later replaced with technical trend criteria meant to stay with the dominant move.
When later years fade
A coffee spread with listed opens on February 7 and/or February 15 and a February 22 close was judged less decisive after one strong year was followed by two years that faded by the exit date. That multi-year fade is an abstention rule: a planned stand-aside when the live chart or the recent pattern fails the seasonal hypothesis.
The current chart was then overlaid with 38 percent, 50 percent, and 62 percent retracements and with speed resistance lines at one-third and two-thirds. A speed resistance line is a fractional-slope overlay drawn from a swing high and low.
Confirmation before initiation
Listed seasonal dates were treated as incomplete for execution. An entry rule required at least one later trading day to move the spread in the anticipated direction before initiation. That confirmation-day rule withholds entry until a session after the listed date agrees with the hypothesis.
On a Minneapolis May wheat versus Chicago July wheat spread, the listed entry date arrived while the difference was still widening and at an extreme width. Initiating immediately was recorded as a violation of the one-day confirmation rule.
A fair seasonal history
A May corn versus September wheat spread whose three-year seasonal history was only fair, with a listed open as late as February 28, was treated as a case where a two-day upturn in the current downtrend could justify deeper tests. Those tests included moving averages, trading bands, trendlines, percentage retracements, and cycle analysis.
Editorial note: a fair multi-year record did not cancel the window. It raised the burden on live confirmation and on the other overlays before the spread could be treated as ready to test.
Rebuilding the spread bar
To run bar-based studies on the relationship itself, highs, lows, and closes of the two legs were differenced in a spreadsheet into a synthetic high-low-close series. That series was returned to the charting program so RSI, directional movement, stochastics, and oscillators could be calculated on the spread.
Spread-reconstruction made the difference series inspectable with the same bar tools used on a single contract. Editorial reading: the reconstructed bar is a third gate, not a decoration. If the synthetic series fails the hypothesis, the procedure can stand aside even when the catalog date has arrived.
All readings on this track · 18 readings
- 1982Constructing commodity spreads, calendar legs, and option straps
- 1987Seasonal calendar spreads as financed capital slots
- 1989Seasonal dates, confirmation, and reconstructed spread bars
- 1991Calendar-spread divergence as a demand clock for commodity reversals
- 1992Expiration-forced convergence for basis, calendar, and option-income trades
- 1993Write the soybean intercrop calendar as one procedure
- 1996A commodity-basket redesign read as a proxy and a calendar-spread
- 2003Calendar spread construction by volatility skew
- 2003Livestock calendar spreads during a commodity-index roll
- 2004Same-strike calendar construction, decay income, and account gates
- 2004Range-bound calendar spreads and precommitted exits
- 2007Construct futures size from the good-faith deposit and the spread
- 2007Critiquing calendar spreads when reversals overrun
- 2010Name the spread, then construct the legs
- 2012Monthly calendar income under a volatility-regime process
- 2016A VIX calendar-spread case after the 2015 volatility spike
- 2017Yield curve kinks as spread laboratories
- 2018Bitcoin futures contract choice, percentage margin and calendar spreads