With a regional open-list MMP system such as the Law Commission of Canada recommended (but with smaller regions), and using the Scottish Parliament's "highest average" calculation method, if Manitoba voters voted as they did in 2007 my spreadsheet projects an NDP majority of three: 30 NDP, 23 PC, four Liberal.
See MMP Made Easy.
That's using a model with at least one-third of the MLAs elected regionally, in five regions. In most cases three local ridings would become two larger ones. You might have 36 local MLAs and 21 elected regionally.
One interesting difference would be in South West and Central Manitoba: instead of the sole Brandon New Democrat MLA, I project four. That would be the three regional NDP candidates who got the most votes across the region. Maybe Denise Harder from Ste. Rose, James Kostuchuk from Portage La Prairie, and Harvey Paterson from Minnedosa? Instead of the PC near-sweep of the region, when NDP votes count equally the PCs get six seats, not nine. The ten MLAs in that region would be six local, four regional. The PCs would no doubt have won five of the six local seats, so they even get one of the regional MLAs.
Of course, this projection simplistically assume voters would have cast the same ballots they did in 2007. The reality would be different. When every vote counts, we typically see around 8% higher turnout. And you would see different candidates. When any party's regional nomination process nominates five regional candidates at once, you can expect them to nominate a diverse slate.
Conversely, Conservative votes across Manitoba would also count equally. In the 18 ridings of north Winnipeg, instead of one lonely PC we'd see five, and two Liberals instead of only one. Maybe Linda West, Chris Kozier, Kelly de Groot and Brent Olynyk would be PC regional MLAs, and Wayne Helgason a Liberal regional MLA?
The 13 ridings of south Winnipeg were less skewed. Instead of three PCs we'd see four; perhaps Jack Reimer? Instead of only one Liberal, we'd see two: perhaps Paul Hesse?
The six Northern ridings would have a couple of regional Conservative MLAs along with four local New Democrats. That might be Maxine Plesiuk and David Harper?
The 10 southeast ridings around Winnipeg actually, by a fluke, saw a fair result over all. It would still be four New Democrats and six PCs.
The exact numbers might be different if Manitoba had four regions rather than five. And they would certainly be different if Manitoba used the old German "highest remainder" calculation (which Germany has just moved away from): maybe 28 NDP, 22 PC and seven Liberal. I'm assuming the Manitoba government would prefer the Scottish model.
As noted in previous posts, I prefer regional "top-up" MPs elected personally under the "open list" model. You would have two votes, and more choice. "Open list" means that voters can vote for whoever they like out of the regional candidates nominated by the party's regional nomination process. The party would win enough regional "top-up" seats to compensate for the disproportional local results we know all too well. Those regional seats would be filled by the party's regional candidates who got the highest vote on the regional ballot. Canadian voters have twice rejected models with closed province-wide lists. The open-regional-list model is used in the German province of Bavaria, and was recommended by Canada's Law Commission and by Scotland's Arbuthnott Commission.