My starting point is that after the 8th lowest minimum extent in 2017, winter freezing was so low as to result in a near as dammit record low extent in March 2018,and a 2nd lowest PIOMAS volume. Sea ice in the peripheral seas looks ready to disappear, and albedo warming potential is therefore high at the moment.
Then, looking at the melt that happened from now to minimum over the last 10 years, 5 would result in an extent minimum above 4 million km2, and 5 below 4 million km2.
In 2017 melt from now was a measly 9 million, giving a minimum in 2018 of 4.6 million, while in 2016 the melt was 9.8 million, giving a 2018 minimum of 3.6 million km2.
I discount much of the extreme 11.5 million melt in 2012, as extent on 25 March in 2012 was 0.9 million km2 greater than it is today, and I understand this was largely due to vast amounts of ice shoved into the N. Atlantic causing a very high extent maximum followed by a late very strong melt (in addition to perfect melting conditions late in the melting season)
Global Ocean Heat dropped in 2016 (El Nino) and came back strongly to a record high by December 2017 (ENSO neutral and La Nina). La Nina continues, and is expected to continue until very late in the melting season, so I am expecting Global Ocean heat to rise strongly at least until the end of the melting season. Eventually (but when) some of that additional heat must reach the Arctic? Already in the N. Pacific?
So, even if there is a cool and cloudy Arctic Summer, it seems to me the balance of probabilities is for enhanced melting compared with 2017 due to ocean heat. I plump for less than 4 million, and a record (extremely?) low maximum in March 2019, but will go no further than that. Hostage to fortune again.