Final Election PredictIt Positions (still betting against Nate Silver!)
For background into my strategy and the mechanics of the markets, read my initial betting primer here. For my final predictions in the election race by race, see my article here.
When I started modeling the elections, it was all just to have a better sense of the state of the race for myself and those around me, but by the time my friend sent me a link to PredictIt.org, the legal academic elections betting market, I already had a complete set of predictions, and a good bit of beef with the methodologies used by The 538 and Nate Silver. Their model tended to make assumptions that included relatively high likelihoods for extreme outliers, leading to some rather outlandish probabilities like South Dakota or Montana going for Hillary 10–15% of the time, or Minnesota flipping red a quarter of the time. I felt strongly that these were the sorts of results that should cause a re-evaluation of a model, but I didn’t have much recourse other than writing my results and waving my hands at the 538.
What I found was that the predictit.org betting markets tended to track the odds provided on the 538 almost exactly. When one outlier poll would come out that would swing their odds in a safe state, the markets would overreact and crash, giving excellent risk-adjusted odds for races that hadn’t really shown any signs of movement. Watching this behavior for a day or two, I had to commit.
So, it’s mostly an accident that I fell into betting heavily on the results of the election, and encouraging others to do so by publishing my advice on the best bets with the highest risk-adjusted returns. But I’m proud to be staking my statistical claim against the behavior of the 538 model this year, and I’m thrilled to be bringing both close friends and strangers with me on this journey.
I’ve dragged six of my friends in directly, and together we’ve placed more than $10k on the election, primarily on the overall presidential election outcome (for Clinton), and on individual state results, concentrated in states strongly leaning Dem or Rep. I’ve also dabbled in senate races, mostly in unlikely but undervalued contracts like republican wins for NV, NH, and PA. I’ve also taken positions in the electoral vote distribution market since the 538 model has exceptionally wide tails.
All in all, I’m in 28 different markets, with a bulk of that in the presidential overall (where 538 has been approximately 10–15% more Trumpish than me), and in strongly leaning states. These bets are primarily intended as specific rebuttals of probabilities from the 538 model. Bets in the senate or other categories are primarily made as opportunistic market plays, and not necessarily in opposition to the 538 model.
List of all states betting Democrat: ME, MI, NY, NM, NC, WI, FL, GA (Last two very cheap).
List of all states betting Republican: AK, ID, IN, IA, KS, MT, SD (all similarly valued)
List of Republican Senate Bets: NH, NV, PA
List of Democrat Senate Bets: MO, FL (~10–1 odds)
I’ll update one last time post-election for a full recap of results, but till then… good luck to you all, and GO VOTE!
