I think this is a good example of stats being very misleading
66 occurences w/final score difference of 7 happened about 8% of the time (66/800) but to really find how it applies to this situation you would need to :
1)eliminate all games where the dog won by 7 from the 66(dog ML winner means the 1/2 pt wasnt factor)
2)plus eliminate all games where the closing point spread wasnt 7 from the 66(0, -1,-2,-3 ,-4,-5,-6,-8+)
leaving you just the games where the -7 pt fav won by 7- i am guessing there would only be a couple instances of this or "X"
3) find the total number of games with pt spreads of -7 for the yr
Then divide "x"(# times -7 fav won by 7) /number of all games with spread of -7
if there were 3 times -7Fav won by 7 and 80 of the 800 games had spread of -7
3/80= 3.75%
odds of occurrence
if it were 5x out of 60 games 5/60=8.3%
the actual numbers are probably closer to 2/35=5.7% or less
if 8% of all games randomly fall on 7, one would expect a specific event of -7 falling on 7 to occur at much lower rate