From the journal

The 1994to 1995NBA season.

In 1995 the notebooks became software for the home PC, and for the first time two whole rosters could be compared in an evening. The 1994–95 NBA season was the first the author followed that way, game by game. This is what the record says, and what it leaves out.

The first season the software followed, game by game.

The season

The company page tells the story of the software. By the autumn of 1994 it could take a roster of names and birth dates and compute every player’s cycles for a given date. The author used it to look at both teams before each game he followed, and to keep a log of what he found and what happened.

Three bands, by how far apart the two teams were in positives

  1. Under 2:1

    01

    The two teams close in positives. Most games fell here.

  2. 2:1 to 3:1

    02

    One team with twice the positives of the other.

  3. 3:1 and above

    03

    One team with three or more positives for every one. The rarest band, and the one the author watched.

For each game the log counted the players in a positive phase on each side and sorted the game by how far the two counts were apart. The journal gives no number of games in any band.

The ratio

The measure he settled on was a ratio. For each team he counted the players whose cycles were positive and the players whose cycles were negative on the day, and compared the two teams’ counts. A game where one team had three positives for every one on the other side was a 3:1 game; a game where the two teams were close was under 2:1.

The log grouped the games by that ratio: under 2:1, between 2:1 and 3:1, and 3:1 or more. The author’s note on the season was that the games in the top band went more often the way the ratio leaned than the games in the bottom band did, and that the pattern looked similar in the seasons that followed.

What the record does not hold

The season log itself was not published, and the journal gives no counts for it: no number of games in each band and no number that went either way. What survives is the method and the author’s summary of it. We print that summary as his, not as a result.

The original page measured the games against a bookmaker’s point spread. This site does not: we make no claim that the cycles pick winners, and nothing here is offered as a reason to wager. The ratio is kept because it is the ancestor of the team figure in the Teambios spreadsheets, and because it shows how the author thought.

One rule from the season

What the season taught its author was a rule of restraint he kept from then on, the law of initial value: when two sides are far apart in ability, the cycles do not come into it. Only when the teams are close does a comparison of cycles mean anything at all, and even then, as the studies would later show, less than he hoped.

The records