In fact the radiation balance becomes negative earlier at high latitudes than low ones. The statistical end of the melting season is usually characterized by a balance of freezing of the leads and open seas that have formed at higher latitudes, even as some melting is still happening at lower latitudes. You could make an argument on that basis that the melt season should end *earlier* as the Arctic manages to melt some ice at higher latitudes each season, creating a stronger ice regrowth mechanism at earlier dates in the now-exposed high-latitude water. Maybe that's still a possibility once we really hit a seasonally ice-free state but it sounds as though we're not there.
But what ever arguments we make must be constrained by observation.
Current least squares fit 1979 to 2012: -0.0376x +255.72 with an R2 of 0.0029.
Trend in the residuals from that trend: 1E-05x - 0.0034 with an R2 of 4E-10.
In other words, there is very little trend in the date of minimum, the trend itself is too small to be statistically significant and describes a negligible fraction of variance (that variance is random and can be assumed to be due to weather, not any underlying process). And the difference from the trend (residuals) also show no behaviour that is suggestive of a long term trend.
So base on the data so far, despite people's desire to believe otherwise there is no trend in the date of min in CT Area data. Given the massive changes we have seen this is telling us that it is highly unlikely that we will see such a change in the years to come.
It might seem that with a seasonally sea ice free state we will see a later date from which ice growth will recommence (the equivalent of date of minimum). This remains a popular meme amongst some amateurs studying the Arctic.
However we have dropped from minimum area of 5.3M km^2 in 1979 to 2.3M km^2 in 2012, a drop of 130%. This has entailed a much earlier melt profile, and much more open water warming. Despite this massive change in thermodynamics (much more heat to lose): the date of minimum in 2012 was half the standard deviation of the residuals away from the trend line (which is virtually flat). And the preceding years had negative residuals while 2012 was positive.
When we hit a seasonally sea ice free state (short period in Sept with no ice) it's hard to see how the date of ice regrowth will change given the lack of change so far. However once we get months of open water a later date of regrowth may start to occur.