WANT TO WIN THE REVOLUTION? VOTE!

After a trio of landslide wins in Washington, Alaska and Hawaii on Saturday — the best single day of his campaign — Bernie Sanders narrowed his delegate deficit with Hillary Clinton. But he still has a lot of work to do. Sanders trails Clinton by 228 pledged delegates and will need 988 more — a bit under 57 percent of those available — to finish with the majority.

That alone wouldn’t be enough to assure Sanders of the nomination because superdelegates could stillswing things Hillary Clinton’s way in a close race, but put aside that not-so-small complication for now. The much bigger problem is that it isn’t easy to see where Sanders gets those 988 delegates.

If you’re a Sanders supporter, you might look at the map and see some states — Oregon, Rhode Island, West Virginia, Montana and so forth — that look pretty good for Sanders, a lot like the ones that gave Sanders landslide wins earlier in the campaign. But those states have relatively few delegates. Instead, about 65 percent of the remaining delegates are in California, New York, Pennsylvania, New Jersey and Maryland — all states where Sanders trails Clinton in the polls and sometimes trails her by a lot.

To reach a pledged delegate majority, Sanders will have to win most of the delegates from those big states. A major loss in any of them could be fatal to his chances. He could afford to lose one or two of them narrowly, but then he’d need to make up ground elsewhere — he’d probably have to win California by double digits, for example.

Sanders will also need to gain ground on Clinton in a series of medium-sized states such as Wisconsin, Indiana, Kentucky and New Mexico. Demographics suggest that these states could be close, but close won’t be enough for Sanders. He’ll need to win several of them easily.

None of this is all that likely. Frankly, none of it is at all likely. If the remaining states vote based on the same demographic patterns established by the previous ones, Clinton will probably gain further ground on Sanders. If they vote as state-by-state polling suggests they will, Clinton could roughly double her current advantage over Sanders and wind up winning the nomination by 400 to 500 pledged delegates.

But things can change, and polls can be wrong — and so it’s worth doing the math to see what winning 988 more delegates would look like for Sanders. Call it a path-of-least-implausibility. If you think Sanders can meet or exceed these targets, then you can say with a straight face that you think he’ll win the nomination. If you think they’re too good to be true, then you can’t. Here’s the Bernie-miracle path I came up with:

Sanders’s unlikely path to a pledged delegate majority

		SANDERS DELEGATE TARGET
STATE OR TERRITORY	NO. ELECTED DELEGATES	ORIGINAL	REVISED	POPULAR VOTE MARGIN NEEDED TO REACH TARGET
California	475	239	274	Sanders	+15
New York	247	125	128	Sanders	+4
Pennsylvania	189	96	101	Sanders	+7
New Jersey	126	61	67	Sanders	+6
Maryland	95	42	43	Clinton	+9
Wisconsin	86	48	50	Sanders	+16
Indiana	83	44	48	Sanders	+16
Oregon	61	37	45	Sanders	+48
Puerto Rico	60	30	33	Sanders	+10
Connecticut	55	28	31	Sanders	+13
Kentucky	55	28	33	Sanders	+20
New Mexico	34	18	18	Sanders	+6
West Virginia	29	17	19	Sanders	+31
Rhode Island	24	13	16	Sanders	+33
Delaware	21	10	10	Clinton	+5
Montana	21	13	16	Sanders	+52
South Dakota	20	12	14	Sanders	+40
District of Columbia	20	8	9	Clinton	+10
North Dakota	18	11	14	Sanders	+56
Wyoming	14	9	11	Sanders	+57
Guam	7	3.5	4	Sanders	+14
Virgin Islands	7	3.5	4	Sanders	+14
Total	1,747	896	988	Sanders	+13

To repeat, these are not predictions. On the contrary, they describe a rose-colored-glasses scenario for Sanders that I consider to be very unlikely. To develop them, I started with our original pledged delegate targets for Sanders. Those already look optimistic for Sanders, who has underperformed his delegate targets in most primaries (he’s beaten them in most caucuses, but there aren’t many caucuses left on the calendar).

But for Sanders to get a pledged delegate majority, even our original targets aren’t enough now — they’d leave him 92 delegates short. So I kept tweaking these numbers upward until I got Sanders to 988 delegates. I was a bit more conservative about giving him extra delegates in states with substantial black or Hispanic populations, since Sanders has tended to underperform our original projections in those states. But mostly, I had to be very liberal about those extra delegates.

I assumed Sanders would narrowly win New York, for instance, even though he’s trailed Clinton by margins ranging from 21 to 48 percentage points in recent polls there. Likewise, I had him winningPennsylvania and New Jersey, where polls also have him down by 20-something points. And I had Sanders winning by a landslide, 15-point margin in California, even though he’s behind in polls there also. (Because Democrats’ delegate allocations are highly proportional, it’s easy to approximate the popular vote from the delegate count and vice versa.)

I assumed Sanders would win Oregon by the same enormous margin that he won Washington, even though Oregon is a primary while Washington held caucuses. I gave a blowout win to Sanders in Kentucky, even though neighboring Ohio and Tennessee easily went for Clinton.

The most recent poll of Wisconsin, which votes next week, has Clinton winning there. I ignored it and assumed Sanders will win by 16 percentage points instead. The demographics do look pretty good for Sanders in the Badger State.

But is Connecticut a good state for Sanders? I’m not so sure: Its demographics are more Ralph Lauren than L.L. Bean. I gave it to Sanders anyway.

I assumed Sanders would win Puerto Rico because it’s a caucus, even though Clinton has much of the party establishment behind her. New Mexico? Nearby Arizona and Texas went overwhelmingly for Clinton, but let’s give it to Sanders.

You get the picture. It’s not hard to imagine Sanders meeting these super-optimistic projections in a few of the states. But he’ll have to do so in all of the states, or else he’ll have even more ground to make up elsewhere. If he loses Wisconsin, for instance, or only narrowly wins it, that’s more votes he’ll need to win in New York or California.

The good news for Sanders is that this scenario would represent such a massive sea-change that superdelegates really might have to reconsider their positions. You might even say it would require a revolution, a profound rejection of Clinton and the status quo.

Originally published at FiveThirtyEight.