h1

Richard Nabavi on the US Senate elections

September 14th, 2018

On the 6th November, 33 of the 100 seats in the US Senate come up for election. The Democrats currently hold 47 seats and two independents caucus with them, so they need a net gain of two for them to get control1. Unfortunately for the Democrats, the starting position is rather difficult; of the 35 seats up for election, only 9 are Republican-held, and of those 4 are solid and 3 are fairly safe. On the other side, of the 26 seats the Democrats are defending, several are potential Republican gains.

The table shows the 13 main battlegrounds, approximately in increasing order of difficulty for the Democrats. Not shown in the table are the 41 Democrat-caucus seats, and the 46 Republican seats, which are either not up for election, or are considered safe. The second column shows the number of seats the Democrats would have (including the two independents who caucus with them) if they won that seat and every one above it in the table (plus their safe seats). The third column shows the probability of a Democrat win as calculated by Nate Silver’s ‘Classic’ model2 – this is broadly compatible with the qualitative assessments of Larry Sabato and the Cook Report. The last two columns show the best odds3 available at time of writing from Ladbrokes or Betfair Sportsbook on the two parties.

State Cumulative count 538 Dem prob. Cook Sabato Best odds Dem Best odds Rep
New Jersey 42 92.9% Likely D Likely D 1.2 6
Minnesota (Special) 43 91.9% Lean D Likely D
Montana 44 90.3% Likely D Lean D 1.36 9.5
West Virginia 45 88.8% Tossup Lean D 1.36 4.0
Indiana 46 76.7% Tossup Tossup 1.66 2.1
Missouri 47 71.3% Tossup Tossup 1.66 2.1
Arizona 48 66.6% Tossup Tossup 1.53 2.8
North Dakota 49 60.1% Tossup Tossup 2.5 1.9
Nevada 50 59.4% Tossup Tossup 1.5 2.5
Florida 51 55.8% Tossup Tossup 2.0 2.2
Texas 52 34.8% Lean R Lean R 3.75 1.44
Tennessee 53 30.3% Tossup Lean R 3.1 1.44
Mississippi (Special) 54 17.2% Likely R Likely R

As you can see, the Democrats are favourites in 10 of these contests, according to the 538 model. If they win all of those, they will just make the 51 seats needed, down to and including FL. However, that assumes they don’t mislay any along the way, which is very far from certain – Florida, Nevada, North Dakota and Arizona could each easily go either way, and it won’t be much of a surprise if the Republicans win Missouri and/or Indiana.

If national sentiment moves firmly towards the Democrats between now and the election, that might be sufficient to carry them over the line. However, Senate elections are subject to significant state-specific and candidate-specific variation. As things stand, to win control, the Dems need to stay lucky in all of the first 10 listed states, or perhaps stay lucky in 9 and pull off a surprise in Texas or Tennessee.

Note that a Democratic majority as defined by the Betfair market, where the two independents who caucus with the Democrats aren’t counted, corresponds to 53 seats or more in the table. That means winning at least 12 of the 13 contests shown – a very tough challenge.

My betting strategy on this is to take advantage of the uncertainty by betting on the Democrats in individual states where the odds look (relatively) favourable, but betting against them winning 53 seats or more (including the two independents) by laying Dem majority in the Betfair market, currently at around 8.0. That way, I hope to make some money on individual contests, but also on them tripping up in at least one or two.

Richard Nabavi

1. If it’s a dead heat, the Vice President has the casting vote, so the Republicans retain control.
2. As at Sept 13th
3. You might get better odds on the Betfair exchange, subject to charges