FYI, my remarks related to SPY, not $SPX. There are tiny differences.
There is what I call a noise difference which is the daily tracking error. This is usually quite small, yet can be fairly large during significant rare events such as the flash crash of a couple weeks ago.
A long-term difference is caused (I think) by the fact that SPY pays dividends while $SPX does not. This can usually be seen as a singular event on the daily chart every three months, then its effect is averaged into the data over time by the fund managers in order to produce a smoother correlation between SPY and $SPX. Another source of long-term difference may be due to a small management fee to produce SPY.
The long-term effect (and noise) can be seen in the first chart below. It is a 1-year chart -- I think the upward slope over a year reflects the gain in SPY due to dividends minus fees. But that might just be my fantasy. The second chart uses only daily Closing prices instead of the daily OHLC values of the first chart.
News 
Market Data 
Discover 
