When we come up with percentages on the Democratic odds of retaining control of the Senate, and individual odds in each race, it may see all very sterile and abstract, maybe even arbitrary as to what comes out of the black box. However, it isn't arbitrary at all; each day we run the model, it means we run 10,000 Monte Carlo simulations of what might happen, and add up all the many ways those imaginary elections play out.
There's no one "right" way for the election to happen according to the model, but rather there are thousands of possible permutations. Of course, only a few of those permutations happen regularly, which is why there's a pretty reliable distribution of results between 45 and 54 Democratic-held seats. Unfortunately, those scenarios where the Democrats hold all their current seats and pick up Mississippi for good measure are pretty uncommon!
So what are the most common permutations? Here's a look behind the curtain. For our purposes, all of the 20 most common permutations involve the Democrats holding all their seats in the 99 percent range (like Minnesota, Oregon, and even Michigan), and losing all their seats in the 1 percent range (including their currently-held but open Montana, South Dakota, and West Virginia). All the variation happens within the ten truly competitive seats:
1) Dems hold NH, lose AK, AR, CO, IA, LA, NC, don't pick up GA, KS, KY (net 46) 6.7 percent odds
2) Dems lose AK, AR, CO, IA, LA, NH, NC, don't pick up GA, KS, KY (net 45) 6.6 percent odds
3) Dems hold NH, NC, lose AK, AR, CO, IA, LA, don't pick up GA, KS, KY (net 47) 6.6 percent odds
4) Dems hold NC, lose AK, AR, CO, IA, LA, NH, don't pick up GA, KS, KY (net 46) 3.9 percent odds
5) Dems hold IA, NH, NC, lose AK, AR, CO, LA, don't pick up GA, KS, KY (net 48) 1.8 percent odds
6) Dems hold AK, NH, NC, lose AR, CO, IA, LA, don't pick up GA, KS, KY (net 48) 1.8 percent odds
7) Dems hold IA, NH, lose AK, AR, CO, LA, don't pick up GA, KS, KY (net 47) 1.6 percent odds
8) Dems hold AK, NH, lose AR, CO, IA, LA, don't pick up GA, KS, KY (net 47) 1.6 percent odds
9) Dems hold CO, NH, NC, lose AK, AR, IA, LA, don't pick up GA, KS, KY (net 48) 1.3 percent odds
10) Dems hold LA, NH, NC, lose AK, AR, CO, IA, don't pick up GA, KS, KY (net 48) 1.2 percent odds
11) Dems hold CO, NH, lose AK, AR, IA, LA, NC, don't pick up GA, KS, KY (net 47) 1.1 percent odds
12) Dems hold IA, lose AK, AR, LA, NH, NC, don't pick up GA, KS, KY (net 46) 1.0 percent odds
13) Dems hold IA, NC, lose AK, AR, CO, LA, NH, don't pick up GA, KS, KY (net 47) 1.0 percent odds
14) Dems hold AK, lose AR, IA, LA, NH, NC, don't pick up GA, KS, KY (net 46) 1.0 percent odds
15) Dems hold AK, IA, NH, NC, lose AR, CO, LA, pick up KS, don't pick up GA, KY (net 50) 1.0 percent odds
16) Dems hold AK, NC, lose AR, CO, IA, LA, NH, don't pick up GA, KS, KY (net 47) 1.0 percent odds
17) Dems hold LA, NH, lose AK, AR, CO, IA, NC, don't pick up GA, KS, KY (net 47) 1.0 percent odds
18) Dems hold IA, NH, NC, lose AK, AR, CO, LA, pick up KS, don't pick up GA, KY (net 49) 0.8 percent odds
19) Dems hold CO, IA, NH, NC, lose AK, AR, LA, pick up KS, don't pick up GA, KY (net 50) 0.8 percent odds
20) Dems hold AK, CO, NH, NC, lose AR, IA, LA, pick up KS, don't pick up GA, KY (net 50) 0.8 percent odds
We'll explain what this means, as well as looking at the new polling changes, over the fold:
At first glance, this looks pretty bad for the Democrats. The likeliest pathway leads to only 46 Dem-held seats, and only 3 of the 20 likeliest pathways lead to Democratic victories (50 seats or more). In fact, the second likeliest pathway is pretty much the worst possible scenario: 45 Dem-held seats, which would be a net loss of 10.
However, when you look at the rest of the tools on the Senate page, it's not quite that bad. The median number of seats in all the simulations is still 48; if you look at the histogram, the modal number (the one that occurs most often) is 47. And Democratic odds of holding the Senate are 34 percent, which is much better than 3 in 20 (which would be only 15 percent).
That's because there are so many pathways to 47, 48, 50, or more. Some of them are pretty obscure, and they don't occur enough times to make it onto our top 20 list; these are the ones that, for instance, might involve a victory in Arkansas, Georgia, or Kentucky, which is something that doesn't happen in any of the top 20 permutations. Or it might involve a weird combo, like winning in Louisiana but not North Carolina. These kinds of examples happen less, but they add up. Meanwhile, there's really only one way for the Dems to land at 45 (unless you delve into really weird scenarios, i.e. losing Michigan).
At any rate, you can see the most plausible pathway for the Dems, if you look closely: Hold New Hampshire and North Carolina (as they're currently on track to do), then hold two out of the three of Alaska, Colorado, and Iowa (as they're currently not on track to do), and then hope that Greg Orman not only wins in Kansas (which he's on track to do) but decides to caucus with them (who the hell knows). It'll take a couple lucky bounces, but it's not far-fetched.
Since our last look at the Outlook last Thursday, the Democrats' odds of holding the Senate have actually improved a bit, rising from 30 percent to 34 percent. The state where the Democratic odds improved the most — and that had a big impact on the overall odds — was Alaska, where the most 'wtf?' poll of the month popped up on Friday. It was a poll from local pollster Hellenthal, giving Mark Begich not just the lead (something we haven't seen from any pollster in a month) but a lead of 10 points over GOP challenger Dan Sullivan.
That boosted Begich's odds from 22 percent to 35 percent, and would have had an even bigger impact if it weren't for two other polls in the same period, a YouGov poll giving Sullivan a 4-point lead and a Dem internal from Harstad showing a tied race (which, after compensating for its partisan status, pencils out to a 3-point Sullivan lead anyway). Those other polls are much more consistent with the recent trend, so either Hellenthal rolled a pretty wild outlier or they're the first to catch a bold new trend. (They're standing by it, saying that Begich's unprecedented ground game is finally changing the game.)
There was movement toward the Democrats in several other key races too, though. In Colorado, for instance, Mark Udall's chances went up from 23 percent to 30 percent on the strength of a 1-point lead in that same wave of YouGov polls, and a 1-point deficit in a new Marist poll. Mary Landrieu's odds in Louisiana improved from 21 to 29 percent, also based on the new YouGov poll (although that still had her trailing in the runoff against Bill Cassidy, just by less than average), and Jeanne Shaheen's odds in New Hampshire rose from 63 percent to 73 percent, thanks to four different polls since last Thursday all giving her single-digit leads.
On the gubernatorial front, however, Democratic odds of gaining seats went down a bit, from 60 percent last Thursday to 55 percent now. That's largely because of unexpected drops in two New England states, especially Maine, where Mike Michaud's odds of defeating Paul LePage plummeted from 52 to 35 percent mostly because of one odd poll from the always-wacky Univ. of New Hampshire, who found Michaud falling from a 2-point lead to a 10-point deficit over the course of the last month (although the League of Conservation Voters' release of an internal poll from PPP showing the race tied didn't help the averages either).
Meanwhile, Martha Coakley continues to flail away in Massachusetts, falling from 52 to 35 percent odds, after this week's Boston Globe poll saw her fall from a tie the week before to a 9-point deficit now (and, much like Maine, the release of an internal poll showing her leading by only 1 didn't help matters, either). Pat Quinn also fell slightly below water in Illinois, from 69 to 48 percent odds, after three different polls showed him trailing Bruce Rauner, though in each one by only one or two points.
On the plus side (if electing Gina Raimondo can be considered a 'plus'), the Rhode Island race is about to move off the table; Democratic odds here moved from 83 to 97 percent, thanks to a Brown Univ. poll giving Raimondo an 11-point lead. And after some troublesome tightening in the last few weeks, Paul Davis seems to have more breathing room in Kansas again, with his odds increasing from 55 to 81 percent, thanks mostly to a 7-point lead in the newest Rasmussen Reports poll.