I just don't see how he arrives at those figures.
Currently, the Conservatives hold 288 seats in the Commons (after expelling the 21 rebels).
The Lib Dems currently have 18 seats, and the SNP 35, so the first projection involves the Lib Dems winning 14 seats and the SNP 6.
Both those projections look on the low side to me, particularly the ones for Scotland, but almost all, if not all, of the projected Lib Dem and SNP gains will be at the expense of the Conservative Party, thus implying a net loss to the Conservatives of 20-odd seats they currently hold.
So somehow the Conservatives are going to end up with 35 more seats than they have at present, which means they'll somehow have to take more than 50 seats off Labour, which doesn't sound right to me -- where does he say these 50-odd constituencies are?
The problem with using national opinion poll figures to predict General Election results is that it's not a national poll -- it's constituency by constituency.
Under normal circumstances, we can examine the national percentage votes cast in the previous general election (since the way people voted last time is generally the best predictor of how they'll vote in the next election) and see how that translated into seats, but these aren't anything like normal circumstances, since we've now got many seats in serious contention between 3 or more candidates, both because the Brexit Party (BXP, as it's being abbreviated to) are polling so much better than did UKIP and because one of the main election issues -- Brexit -- cuts right across normal party loyalties.
This means that it's at least highly likely that the raw national figures conceal huge local variations.
For example, I would put good money on much of the increased Lib Dem support in national polling and the fall in Labour support representing the collapse of the Labour vote in
Lib Dem target seats, almost all of which are held by the Conservatives (at least in England), as Labour supporters, whose candidate stands no chance of winning in those seats, vote for the Remain candidate most likely to beat the Leave-supporting Tories, as will Remain-supporting Tory voters.
However, I think the Labour vote is likely to hold up much better in Labour-held seats, where Labour supporters are motivated to keep the Tories out, and most Labour defectors will either not vote at all or will switch to BXP rather than the Conservatives.
Under those circumstances, and particularly given the normal margin for error in polling and the effect that's going to have on predicting the outcome in constituencies where, for the first time, there are 3 or 4 serious contenders for the seat, I just don't see how those figures work out.
Plus, of course, there's so much still to happen before any general election that any prognostications -- even those based on the results of the various parties' private polling, which is much more granular -- are no more than speculation.