Well, it is a gradient. 3.5+ is really good, 1 is still profitable but just. The cut-off is maybe 2.5 points before you really start getting the upper hand so let's call that "really good" . So if you want a set of rules let's go with this:

1) play for sure if there are 3 games that have a score of 2.5 or more,
2) if there are only one or two like that, play only if you are feel like it (but never play less than 3 games/ticket whatever you do!).
3) if there are zero games at 2.5 or more stay home that week
4) padding out the ticket to 4,5 or more games is optional but should be done if you are thinking long-term


Whether you pad out the ticket with others that are 0.5-2 is up to you - more games is much more profit and is mathematically in your favour but you will statistically have to wait longer before you cash a ticket.

Examples:

3 game ticket, all games 2.5+

Chance of winning: 0.64*0.64*0.64 = 26.2%
Profit-needed-to-break-even: 381%
Actual profit: 500%
Profit per bet: ~31%

5 game ticket, 3 games 2.5+, remainder 1+

Chance of winning: 0.64*0.64*0.64*0.59*0.59 = 9.1%
Profit-needed-to-break-even: 1099%
Actual profit: 2000%
Profit per bet: ~82%

The real problem with this system is that example two won't come up that frequently in a season, or even example one for that matter. But by throwing these up on twitter automatically I hope to avoid the boredom of calculating each week only to find you should not bet but instead all that is now needed is a quick glance each Sunday morning to see if it is worth trudging out to the convenience store to play.

The historic percentages are at the bottom for your reference.

POINT SPREAD winners as shown by points diverged 2004-2009:

Percentage
0.5-55.1%
1-59.7%
1.5-58.4%
2-59.0%
2.5-63.8%
3-64.2%
3.5-67.3%