It was a late night last night with the results of the US Midterm Elections going into the wee hours.
If you stayed up then you will know it was a bit of a rollercoaster ride as things looked bad for the Democrats early on but gradually improved as the night went on.
I had given my thoughts on the betting for the Midterms last week, so how did my predictions turn out?
Well overall it was a good night, with my main bets on the Democrats to win the House and Republicans to hold the Senate landing comfortably.
Looking back I can still not quite believe the Democrats were 1.6 to win the House, but I guess people were spooked by the Presidential Election in 2016 and didn’t believe the numbers that strongly favoured the Democrats.
It was a shame there weren’t odds being offered on the Republicans increasing their Senate seats as that would have been a good bet to take.
In terms of the individual races, it was a bit more of a mixed picture. On the plus side, Texas did indeed go to Ted Cruz, but it was fairly close in the end as I expected. Jon Ralston’s early voting numbers in Nevada once again proved spot on so well done to Jon and a nice win landed there. And in Michigan Debbie Stabenow also won comfortably so that one went to plan too.
The main surprise though came in Florida, where Democrats Bill Nelson and Andrew Gillum both lost. The polls had them both in front – Gillum significantly so – so I guess the lesson here is to be very wary of state level polls which are notoriously unreliable. Whilst the national polls were reasonably accurate this time, some of the state-level polls were way off, so going forward I will probably not put much store in them at all.
Arizona is unlikely to be called for a few days as there are still over a million ballots to be counted but it looks like Republican Martha McSally will win there.
Overall then a good night but a few lessons to be learned going forward and hopefully next time will get things all right.
Betting on the US Midterms – Is Trump In For a Kicking?
30th October 2018
The US Midterm Elections take place on 6th November and there are a variety of betting opportunities available.
Perhaps because there aren’t many events to bet on, elections don’t seem to attract as much money from professionals as sports betting events and can often present some excellent value if you study them closely.
The Midterms this time are being seen as a referendum on President Donald Trump and there is a chance he could be in for a tough night – perhaps even a real proverbial kicking.
I’ve had quite a lot of success betting on politics over the years so thought I would share my ideas on the upcoming Midterm Elections.
I’ll take a look at the markets and where I see some of that value this time around.
What are the US Midterms?
In the US they have a chance to change government (or at least part of the government) every two years.
In between Presidential Elections, there are elections for all 435 seats in the House of Representatives (aka “The House”) and 35 of the 100 seats in the Senate, plus a slew of governors races and local elections across the country.
The Midterms have historically been a chance to bring some balance to the partisan make-up of government and the party that holds the Presidency usually does very badly.
Barack Obama suffered crushing losses in both of the Midterm Elections during his Presidency in 2010 and 2014, as did George W Bush in 2006. It has often been the case that the President’s party loses control of the House in the midterms and may well lose seats in the Senate as well. The historical pattern is that governing parties tend to lose 30 House seats on average.
This year looks set to be no different, with President Trump suffering poor approval ratings in the low-40s and Democrats looking highly motivated.
But how will his unpopularity play out and how are the betting markets estimating the outcome? Let’s take a look at the different races in turn.
The House of Representatives
Democrats need a net gain of 23 seats to gain a majority in the House. The race for control of the House is a little complicated by the fact that Republican gerrymandering (i.e. the drawing of voting districts that favour them) means the Democrats have to win the popular vote across the country by about five percentage points to reclaim the House.
Leaving aside the question of how this is allowed to happen (most other democracies have independent panels that draw their voting districts), it does create some uncertainty in a race where the Democrats currently hold a commanding average poll lead on the generic ballot of 8.2 points.
If the polls were off by just over 3 points for example, then Republicans would likely retain control of the House. And of course 3 points is about how far off they were in the last Presidential Election, when most polls gave Hillary Clinton a solid lead.
Having said all that, I still think the betting markets are currently underestimating Democrats’ chances in the House. On Betfair, the Democrats are around 1.6 at the time of writing, which equates to about a 63% chance of winning the House. You can view the market here.
However, on Fivethirtyeight’s House Forecast (the Classic version, which incorporates not just polls but fundamentals like fundraising, historical trends and other data), they currently give the Democrats an 86% chance of winning. So based on that model, the odds should actually be around 1.16.
That is a pretty phenomenal difference and if you saw that in a football match or horse race, you probably wouldn’t believe your luck. It would be like Man City being 1.6 at home in the Premier League to beat Cardiff, instead of the 1.16 they should be.
Now of course people will point out the Fivethirtyeight model got Trump’s chances wrong in 2016 and so it might be off again.
Their model was the closest to the actual result though out of all the mainstream models, giving Trump a 30% chance of winning, as against some other models which (quite laughably now with hindsight) gave Clinton a 95% chance of winning. And Fivethirtyeight seem to have learned from their mistakes last time and incorporated a lot more data into their model this time around, rather than focusing too much on the polls.
There are other reasons though why I think their model has things more or less correct this time and that Democrats have a really good chance of winning the House – or certainly better than the betting markets currently give them:
– Special Elections
These are the elections to fill the empty seat of a politician who dies or retires (we call them by-elections here in the UK). They give a useful guide to how the public mood is shaping up.
Since the 2016 Election the Democrats have outperformed the partisan lean of districts in Special Elections by an average of 16 points. If that was repeated in the Midterms, needless to say it would deliver them a thumping victory.
The Democrats have outraised the Republicans 2 to 1 in individual contributions. This is pretty much unprecedented and the models have struggled to know how to interpret this data as it’s such a strong trend.
What is important is not so much that the extra money gives one side an advantage – the Republicans will probably more than make up for it in big donations from Super-PACS (bodies that can raise unlimited amounts of money from corporations and billionaires and donate it to a campaign), but what it indicates in terms of the enthusiasm of the public. The fundraising data suggests Democrats are very enthusiastic indeed.
– Voter Registration
Although typically not a strong predictor of the final results, voter registration is another barometer of enthusiasm. And Democrats have seen a surge in voter registration in a range of states across the country, particularly amongst the young, a key demographic for them (more on that below).
– Youth Enthusiasm
The young vote heavily Democrat (nearly two thirds voted for Clinton against just a third for Trump) – the problem is they don’t vote very much. Whilst turnout for those aged over 65 tends to be around 60% in midterm elections, for those aged 18-29 it struggles to get over 25% – and was a paltry 20% in the last Midterms.
This year could be different however. Polls suggest as many as 40% of young voters may turn out this time, which would presage a “blue wave,” given that 72% of these voters disapprove of Trump.
– Other Forecasts
It is not just Fivethirtyeight predicting a Democrat takeover of the House. The Cook Political Report predicts the “likeliest outcome is a Democratic gain of between 25 and 35 seats.” Inside Elections concurs that the “Most likely outcome: Democratic gain of 25-35 seats, with larger gains possible,” whilst Sabato’s Crystal Ball states “the bare minimum for Democratic House gains is in the mid-to-high teens. The needed 23-seat net gain is not that far beyond that and there are many different paths Democrats can take to achieve it.”
So there seems to be solid consensus about the Democrats’ chances from the forecasters.
There are a variety of indicators suggesting the Democrats are strong favourites to retake the House and yet the betting markets don’t seem to take these into account. Unless something dramatic changes between now and the elections, I see 1.6 as excellent value on the Democrats.
There is uncertainty in terms of youth and minority turnout which could affect the result, but the evidence suggests they will turn out in higher numbers than in recent midterm elections. Even if they don’t, it is almost certain that suburban women will turn out strongly and they favour the Democrats by an estimated 30 points, so they could carry the Democrats over the line in any event.
The Senate is a completely different story and looks almost certain to remain Republican. I backed them at 1.61 back in September and their odds have now dropped to just 1.28.
The Fivethirtyeight model gives the Republicans an 82% chance of holding the Senate, which is a little higher than the Betfair odds would suggest.
This may still present value even at 1.28 as the Democrats have an extraordinarily difficult map – or in other words they have 26 seats up for election where as Republicans have just 9 this time round.
And whilst the Democrats have limited opportunities to pick up seats, a number of their own seats are in red states like North Dakota and Missouri which they may have problems defending.
So the Senate as a whole looks like a very heavy lift for the Dems, but let’s take a look at some of the individual Senate races.
Texas is shaping up to be one of the most exciting races of the election, to most people’s surprise. To be honest this shouldn’t even be close in a deep red state like Texas. But rising Democratic star Beto O’Rourke has surpassed all expectations and run a hugely successful grassroots campaign that has seen him visit all 254 counties in Texas and raise a phenomenal $38m in individual donations.
On the other hand, Republican Ted Cruz is perhaps not the most popular person on the planet.
Even his colleagues in the Republican party have been quoted as saying things like: “If someone shot Ted Cruz on the floor of the Senate, they wouldn’t be convicted,” of calling him “Lyin’ Ted” (that one of course came from Donald Trump who now wants Cruz to win so has taken to calling him “Texas Ted”). My personal favourite though has to be from former House Speaker John Boehner who described Cruz as “Lucifer in the flesh.”
Despite Cruz’s unpopularity and O’Rourke’s brave efforts however, polls have shown the Republican consistently ahead and Fivethirtyeight give him a 78% chance of winning.
I backed Cruz at 1.5 and he is currently around 1.2, which I think is about right. However I think this could be one to watch closely in-running. If O’Rourke can run up some big numbers in the urban districts and gets strong youth turnout, it’s not out of the question he could win.
Arizona is a race where again there is quite a disparity between the odds and the forecasters’ predictions. Whilst Fivethirtyeight give Democrat Krysten Sinema a 64% chance of winning, Betfair actually have her as the outsider. I have backed her at 2.16 and think this is perhaps the Democrats’ best chance of a Senate pick-up.
The bookies see this as a close race and whilst Florida is virtually always close, the 1.83 about Democrat Bill Nelson looks favourable. He is the incumbent and may get a boost from the Democrat candidate for governor Andrew Gillum, who looks to be well ahead in his race. Nelson is rated as a 72% chance with Fivethirtyeight and with the national environment favouring Democrats by around 8 points, they appear to have a good chance of holding Florida.
This looks as good as certain for the Democrats, with the Fivethirtyeight forecast giving Democrat incumbent Debbie Stabenow a 97% chance of winning and the polls giving her an average lead of 12 points. Yet Betfair is offering 1.21 on Stabenow, which suggests some uncertainty. I am happy to take those odds all day long.
Generally you want to ignore early voting (the voting that happens in the two weeks or so before Election Day) and certainly you want to ignore the media narrative around it. One exception to this however is Nevada and the excellent early voting blog from Jon Ralston. He gathers all the data on people who have voted so far in Nevada by party registration and calculates the position of the race. Last time he got it spot on and made me a tidy few quid. I will be watching Jon’s numbers closely again this time.
Sadly there is poor liquidity on Betfair for the Governors races and the bookies don’t seem to be offering odds at all.
If liquidity picks up though, I would be happy to take anything above 1.6 about Andrew Gillum in Florida and 1.3 or above on Tom Wolf in Pennsylvania.
In-Running Betting Can Present the Best Opportunity
There can be great opportunities to make money in-running on politics these days. In the cases of both Brexit and Trump, the betting markets were very slow to react to the early results coming in.
In both instances, it was clear from the early results which way the contests were going. In the case of Brexit, the first two results that were declared, from Sunderland and Newcastle, heavily favoured leave. The only way that remain still had a chance was if those results were massive “outliers.” It’s fairly unlikely that the first two results to come in would just happen to be huge outliers though. And the more results that came in, the more certain you could be they were not.
Yet the markets still saw it as roughly 50-50, even when a slew of strong results for leave had come in. Once you have around ten results come in from different parts of the country, the chance they are all big outliers is virtually nil.
The same thing happened with Trump. Early results from places like Florida and Ohio suggested he was outperforming expectations. Once more results came in from the Midwest, it then became likely he was actually going to win. The results tend to be strongly correlated in Presidential elections and it was highly unlikely Trump was going to win Ohio by 8 points as he did but then lose other rust belt states like Pennsylvania and Wisconsin.
Yet the betting markets still saw the race as close even after all these results were going one way. They may have been spooked by an early exit poll that showed a big win for Clinton, but exit polls aren’t always accurate so I don’t put too much store in them. What matters are results and you can always learn more from the early results than the polls.
So this time I will be watching the early results closely and be getting ready to take advantage. House races don’t tend to correlate as closely as Presidential ones so you need to be careful, but I will be comparing the early results with Fivethirtyeight’s forecasts to judge how things are going. If I see one party outperforming the forecasts significantly then I will look to act.
Conventional wisdom has it that the average of polls give the most likely prediction of an Election outcome, but recent history has suggested that outlier polls can actually prove more accurate.
These are the polls that seem outlandish and way out of whack with the others, like Trump winning, or Corbyn gaining seats in the 2017 UK Election for example.
This time there are some big outliers on each side – some with Democrats with a whopping 17 point lead and others seeing it as an even race.
So it could be a worth a small investment on these outliers, although at the moment seat spreads don’t appear to be available unfortunately. If they do become available I will update this post.
The current House odds on the Democrats look generous to me so unless something drastic changes between now and Election Day I would see 1.6 as good odds. In the Senate Republicans are favoured, but the odds aren’t as generous at 1.28 currently.
Whilst there are some good odds on individual races at the moment, I am most looking forward to the in-running opportunity as the markets are often very slow to react to results coming in.
I am also paying close attention to Jon Ralston’s early voting blog on Nevada.
So there are my thoughts – we will see how close I am on November 7th – or if I am way off!
If you are going to have a punt on the Midterms, please gamble responsibly and only risk money you can afford to lose.