Evaluating five-day soybean open-to-close rules

A locked open-to-close procedure

A soybean sample from 1970 through 1988 was used to test every combination of four successive closes plus a higher or lower open on the following session, producing 32 five-day sequences. A five-day sequence is those four relative closes plus the next open, with that open as both the last symbol and the entry.

Entry was taken at the fifth session's open and the position was closed at that same session's close, so the holding period was a single open-to-close interval. Pattern recognition coded the ordered up and down closes, then the following open, as a forecast of that session's open-to-close direction. Opening range breakout used the trade day's open as the executable entry print that starts the same-session test. Rule-based entry joined the sequence, the open, and a same-day close exit into one complete buy or sell procedure.

How a sequence was judged

Buy versus sell for each sequence was assigned only by which side produced a positive gross-profit total, so every tabulated sequence was shown on its selected side. A sequence was to be judged from percent profitable trades, the average-win-to-average-loss ratio, and gross profit together, not from any one of those readings alone.

A cross-market check

The same sequences were repeated in Treasury bonds to see whether buy and sell orientation stays put. Eight of the 32 sequences showed significantly different directional biases in soybeans than in Treasury bonds, including sequences 3, 4, 15, and 17, which reversed buy versus sell orientation across the two markets.

In the soybean table the largest gross-profit totals clustered on short sequences, with no long sequence above a ten-thousand-dollar gross-profit mark, while the bond comparison was described as more even on that measure.

A range split on sequence 30

Sequence 30 was specified as three lower closes, then a higher close, then a lower open, with a sale at that open. A range split keeps only an expanding or only a narrowing range on the day before entry.

Restricting sequence 30 to an expanding day-4 range versus day 3 kept 63 of 94 cases and lowered the percent-profitable reading to 51 percent, while a narrowing day-4 range kept 30 cases and raised that reading to 67 percent.

A starting template

Small changes to the base sequences, especially one-day range differences used to mark momentum, were presented as more useful development steps than the unfiltered five-day templates alone. The five-day sequences were treated as a starting template for later rule work rather than as finished high-profit procedures, and this style of development was described as less promising in soybeans than in other markets.