I'm not a physicist, but my guess is there is enough data to compute accurately the orbit of Apophis. That is, the proximity to Earth may be quite close (150 lunar distances). However, if they're pretty certain it won't deviate too far from that pathway then the x-axis value on the Torino Scale not just small, but pretty robust.
Those instances of NEOs where the orbit/pathway is erratic and the error bands around the central tendency are wide/grow over time (see below)...
...that's where we may see x-axis values further from the origin, producing Torino Scale values that are worrying.
For now, I'll keep setting my alarm clock.
I remember the heady days in 2004, when it was discovered, and the brief time it became (for a few days) the first actual impact threat, on both those scales. Prediscovery images were found in the archives which gave a longer data set, and poof! The threat vanished. The 2036 passing will be way off, because the 2029 "keyhole" will be missed. Oh well, there goes the excitement.