history is not a good predictor of future events.
the bookmaker is beaten in the short term, as they are typically slow to 'catch up' to the current day-to-day values.
using values from over 3 years ago is not a good predictor AT ALL. the game has changed, the personell has changed, the rules have changed, the playbook has changed, the athletic prowess has changed, even the vig has changed as does the value of bookmaker's opinions etc, etc, etc.
even the value of a half point is constantly changing. it is different by team, it is different in the beginning of a season than later on, it is different off of different key numbers, it is different off of different totals, it is different by year.
If you are trying to beat the oddsmaker, one of the best things to do is handicap every game that you possibly can. Formulate an opinion, record this opinion, track it over time, and review the results to see if you had the right side but a random occurrence caused a loss, or if you missed something or didnt put enough emphasis on it. The more lines and results that you see and remember the better. After a while you will start to see correlations that lead you into identifying value. A well trained human mind is much more powerful and effective than any computer.
If you want to line games in order to book them to the public, use the above and shade the lines to match your opinion. It also helps to know your customers. A lot depends on what kind of action you are getting or want, and limits versus financial backing. Just my opinion.