I'm going with 4.5 - 5.0
Just in case this prediction comes true, let's walk through the thought process which leads to it.
1- 2019 finished with 4.1M in volume and has about a 750K lead on 2020 through 6/30. If that difference holds...we're in business.
2 - 2020 did something I didn't expect by beating the trend of leading years in the last decade and setting a CAB volume loss record for the second half of June. Gotta respect the ability of this season to find a gear above and beyond that we've seen in the past.
3 - 2020 really kicked ass in the period from 6/16 - 6/21 with incremental volume loss. Looking back, the explanation that makes sense to me is high surface temps. It's not a common take, but I'm rolling with that as a big indicator of melt. I expect the first few days of July to be blockbuster volume losses and make up a few hundred km3 on 2019 which was absolutely not a slouch in early July.
4 - The surface temperature edge over the ice covered areas is getting lost in the forecast going forward. If the 850hpa temp crowd is right, I'm wrong.
5 - The Beaufort Sea may escape with a volume absolutely unheard of in recent years. The AWP deficit is just piling up and may reach 200 MJ / m2 by seasons end. The high pressure is causing the CAB to rotate some thick ice into the arm and the open water may not make it past 72N this year. The 2D Beaufort at the minimum might not look very different than it does today !!
6 - History informs that not much export is expected in the late season and this season might have almost none. The compaction has moved the Cablantic ice farther from Fram.
7 - I fully expect the Laptev to go super DEEP into the CAB. Past 85N. But that's it. The Atlantic side should wind up near the usual 82-83N with thick ice at the border. The ESS has a different identity than the Laptev and may not dent the CAB and the Chukchi shouldn't either. We may even get a little left in the northern CAA which is currently taking a mini break.
Over 50 people are guessing a lower volume minimum than my bucket and one guess is higher. Obviously an outlier and set up to potentially fail spectacularly. Just in case it works out, I want to leave a trail of crumbs that helps explain the hunch.