As promised last time, here’s an introduction to my favourite US election market, involving the margin of victory. The presidency is decided via an electoral college, for which each state awards a set number of votes (ECVs) to their winning candidate. For example, Florida has 29 ECVs among the overall total of 538. The winning target is 270.

Complex market provides the ultimate election night challenge

In past elections, I’ve found this market to be a treasure trove of value. On the night itself, the betting will update live with numerous firms, after the polls have closed. The betting can go wild. It is challenging because tallying up the ECVs requires fast maths and being able to weigh up the implications of each result as it emerges, from different time zones.
 
The majority of states are regarded as either safe Republican or Democrat, so the result always turns on five or six swing states. Here’s a list to all the states deemed at least vaguely competitive by the ABC News projection - chosen for their high grade polling. Also listed are each state’s number of ECVs, above both the 2016 result and 2020 position based on the RealClearPolitics average.

New Hampshire (4 electoral college votes)
(2016 Clinton +0.3, 2020 Biden +9.3)

MINNESOTA (10)
(2016 Clinton +1.5, 2020 Biden +9) 

Michigan (16)
(2016 Trump +0.3, 2020 Biden +7.8) 

FLORIDA (29)
(2016 Trump +1.2, 2020 Biden +6.2) 

Pennsylvania (20)
(2016 Trump +0.7, 2020 Biden +6) 

WISCONSIN (10)
(2016 Trump +0.3, 2020 Biden +5) 

Nevada (6)
(2016 Clinton +2.4, 2020 Biden +4)

North Carolina (15)
(2016 Trump +3.7, 2020 Biden +4.5) 

ARIZONA (11)
(2016 Trump +3.5, 2020 Biden +3.7) 

OHIO (18)
(2016 Trump +8.1, 2020 Biden +2.3) 

Texas (38)
(2016 Trump +8.1, 2020 Trump +0.2)

GEORGIA (16)
(2016 Trump +5.1, 2020 Trump +1.6)

Iowa (6)
(2016 Trump +9.5, 2020 Trump +1.7)

ALASKA (3)
(2016 Trump +15.2, 2020 Trump +3)

Missouri (10)
(2016 Trump +18.5, 2020 Trump +6.4) 

Montana (3)
(2016 Trump +20.5, 2020 Trump +10)

Biden must gain 37 electoral college votes on Clinton tally

Note ABC have awarded Maine’s Congressional District to Biden, which amounts to one ECV. Therefore we begin our calculations with the 2016 shares minus one. Trump starts on 305. The winning line is 270 so he afford a net loss of 35. 36 would mean a tie. 37 plus, defeat. 

The list shows every competitive state besides New Hampshire, Minnesota and Nevada was won by Trump in 2016. Therefore unless causing a big upset in any of those states, every state he loses subtracts from the 305 tally. 

Better value for Trump backers in individual states

These maths are precisely why it pays to keep up with the state-by-state picture, and to understand the relation to Trump’s outright odds. No serious analyst can envisage a path to victory that does not entail Florida’s 29 ECVs. He trails by 6.2% there and is trading around 2.7 on Betfair’s state market to win the Sunshine State. 

Yet Trump is only 2.8 to win nationally. Surely it makes better sense to back him for FL than the presidency. Because he could easily win FL, yet still lose those 37 ECVs elsewhere. 

Michigan looks a certain Democrat gain so, without that, he only has 21 in hand. There are 20 up for grabs in Pennsylvania - Biden’s home state, for which Trump is trading around 3.5. Higher odds than for the presidency, as is the case in Wisconsin at 3.25.

Therefore, the best pro-Trump betting plan is to spread the stake around Florida, Pennsylvania, Wisconsin and North Carolina. It isn’t a bet I’ve had yet, but if and when a cover is required for my strongly pro-Biden position, that will be the strategy.

How to play Betfair’s margin markets

Rather for now, my focus is on the total ECVs. Betfair has a market for each candidate, offering bands - eg 240-269, 270-299. If these polling averages prove correct, Biden will win 351 ECVs to Trump’s 187. The winning band on each side is trading around 6.6.

I like those odds, as both an outcome and a trading vehicle. My range of Biden victory margins span from a low of 318-220 (losing Ohio and North Carolina) to a peak of 411-127 (flipping Iowa, Texas and Georgia). 

Those alternative winning bands would pay and 6.4 and 8.8 respectively. For now, I’d rather be in the middle, backing the current state of play - Biden 330-359 at 6.6. If my projection moves up or down, I will cover the position accordingly, either via this market or by backing particular states.  

Can we rely upon state polls?

This is a particularly valid question after 2016. The final RCP average was 7.2% wrong in Winconsin, 3.8% in Michigan, 2.8% in Pennsylvania and North Carolina. On current ratings they would need to be wrong on a greater scale to stop Biden but it shows that, if Trump can close by a few points he’d be competitive.

However it is important to note the consistency of the swing away from Trump since 2016. Of course there are outliers. Biden has only advanced on Clinton’s lead in Nevada by 1.6%. He’s only fallen back by 3.5% in Georgia but at the other end of the scale, by 10.3% in Texas and over 12% in rarely polled Alaska and Montana. 

In the decisive swing states, however, the shift is in a small range - 5.3% in WI to 8.2% in NC. It broadly mirrors the national polling and approval ratings picture. To be wrong in these states, I think the numbers would need to be wrong pretty much everywhere.

How much these projections matter three months out depends how many persuadables there are. I have never been convinced there are many such people, either way, with regards to Trump. Right now, he’s on course to lose the electoral college by a 200-plus margin.