A closing point spread is the predicted scoring margin after wagers have been placed and the betting line has closed. It is the final market-adjusted spread available before an NBA game’s result is known. Each game has opposing road and home values: if one team’s closing spread is positive, the other team’s is negative by the same amount. section 17.3 section 17.0
The spread changes from its opening value as bets arrive. For example, Indiana moved from -7 to -7.5 against Memphis; the Pacers then won 111–83. A negative number marks the side expected to win by more than that margin, while the opposing positive number identifies the underdog’s allowance. section 17.0 section 17.3
To evaluate a closing spread, compare it with the final scoring margin rather than simply asking whether a team won. Using the home-team fields, calculate margin as road points minus home points, then calculate the absolute error between that margin and `closespreadH`. A smaller error means the closing spread was nearer the final margin. section 17.4.3
oddsdf4 %>% mutate(diff_margin_closespreadH = abs(margin - closespreadH)) -> oddsdf4
This R step creates `diff_margin_closespreadH`, the absolute distance between the final road-minus-home margin and the home team’s closing spread. section 17.4.3
A practical comparison uses three derived measures:
- `diff_margin_openspreadH`: the opening spread’s absolute error from the final margin. section 17.4.3
- `diff_margin_closespreadH`: the closing spread’s absolute error from the final margin. section 17.4.3
- `spreadmove`: whether the absolute closing spread was greater than (`up`), less than (`down`), or equal to (`same`) the absolute opening spread. section 17.4.3
Comparing opening and closing spreads
Across the 2018–19 NBA regular season, closing spreads were nearer to final margins more often than opening spreads. Mean absolute error was 9.92 for closing spreads versus 10.10 for opening spreads. The closing spread also outperformed after both downward and upward movement, with reported advantages of about 13% and 8%, respectively. section 17.5.2 section 17.5
This result treats the closing spread as evidence of collective intelligence: bettors who risk money can add information beyond the initial oddsmaker estimate. The finding is limited to the analyzed 2018–19 NBA regular-season data; it compares nearness to game results, not a guarantee of future betting success. section 17.5 section 17.6
