The map is getting quite interesting
October 31, 2020
It looks like Biden has New Hampshire, Nevada, and all four Electoral Votes in Maine in the bag. On the flip side of the coin, Florida and Arizona polls have shifted in Trumps favor. If this election is anything
like 2016, those five states can be moved out of the tossup category. Under these conditions, the map appears to favor Trump, as he's only 11 EVs away from a win, with four close states to make it happen. However, two of those states are only worth ten EVs. 269 is not a win, so he has to pick up two out of the four to secure a win with a buffer against faithless electors.
If Trump were to only pick up one of the ten pointers in this scenario, we'd end up with a 269/269 tie and an already messy election will get ten times worse. Faithless electors could throw the results one direction or the other. Even worse, an unexpected single EV flip in Maine or Nebraska (the two states that break out their EVs) could result in an exact 270 win. In that case, a single faithless elector could undermine a legitimate win in the election. If either Biden or Trump were to hit 270 exactly and then have it stolen from them by a faithless elector, massive unrest is almost a guarantee.
Neither candidate crossing the 270 threshold is essentially the same as a tie. The decision for President would then go to the newly elected House, where each state's delegation casts one vote - majority required to win. The newly seated Senate would pick the Vice President. But the Senate stands a great chance of ending up tied as well. Yeah, huge mess. Pretty much the only person in the country who could be pulling for that kind of disaster is Nancy Pelosi, who becomes the President on Jan. 20th if the Congress hasn't settled on a winner.
No, the polls were not mostly right in 2016.
October 15, 2020
There's a common refrain you've heard since the 2016 election, that the polls that year were actually mostly right. It's not true, and if you pay close attention to what is being written, even by reputable sources, you can pick apart the bad analysis. Some deflect with stories of down-ballot accuracy. Some insist the polls were accurate when taken, and blame the error on late-deciders. Most trying to make the case that the polls were good conflate state-by-state polling with nationwide polling, and once they've blurred the lines, you'll see the two used almost interchangeably. Although there was national polling that predicted the outcome of the popular vote quite nicely, it's irrelevant. The state-by-state polling was a disaster - and that's all that counts.
The United States does not hold a nationwide vote for President.
Rather, the outcome of the Presidential race is determined by the Electoral College, which is determined by the individual elections in each state and D.C.. In 49 out 51 of these elections it's winner-take-all. This changes the strategy for an overall win, and it changes the way we try to predict that win. You can win the national popular vote and still lose the election, so we must forecast most of the individual states correctly. A good forecast for the national popular vote makes for an interesting piece of trivia, and nothing more. To correctly predict who will walk away with the Presidency, you need to have a good idea of how the vote will land state-by-state. Bad polling in just a handful of states can ruin the entire forecast. Missing key states is
missing the election. Bad polling in a few key states is
bad polling in toto.
The pollsters know this, and the forecasters who rely on poll numbers know this too, but very few will admit how bad it was in 2016. The site you're reading right now
missed 2016. FiveThirtyEight did too
, by substantially more electoral votes. They, however, are hesitant to call the polling bad.
Not only did 2016's state-by-state polling miss critical states, it missed them all in favor of Clinton. In fact, virtually all of the closely watched swing states had polling error in favor of Clinton. The error in Wisconsin was absolutely terrible, being off by over 7%. Iowa was just as bad at 6.5% Pennsylvania, Michigan and Ohio all had error in Clinton's favor between 3% and 5%. North Carolina clocked in just under 3% error. These numbers are not just outside a reasonable margin of error, they are horrendously outside - and horrendous in nature. One of the primary benefits of poll-aggregation, not just for the overall election but for each individual state's polls, is that biases and errors tend to cancel each other out. However, when every single miss is in one direction, the aggregation is going to be garbage as well.
|State ||2016 RCP avg ||2016 Actual ||Error/Bias||2020 RCP avg
|Wisconsin ||Clinton +6.5 ||Trump +0.8 ||Clinton +7.3||Biden +6.3
|Iowa ||Trump +3.0 ||Trump +9.5 ||Clinton +6.5||Biden +1.2
|Ohio ||Trump +3.5 ||Trump +8.0 ||Clinton +4.5||Biden +0.6
|Michigan ||Clinton +3.4 ||Trump +0.2 ||Clinton +3.6||Biden +7.2
|Pennsylvania ||Clinton +1.9 ||Trump +1.1 ||Clinton +3.0||Biden +6.5
|North Carolina ||Trump +1.0 ||Trump +3.7 ||Clinton +2.7||Biden +3.3
|Florida ||Trump +0.2 ||Trump +1.1 ||Clinton +0.9||Biden +2.7
|Georgia ||Trump +4.8 ||Trump +5.2 ||Clinton +0.4||Biden +0.4
|New Hampshire ||Clinton +0.6 ||Clinton +0.3 ||Clinton +0.3||Biden +11.0
|Arizona ||Trump +4.0 ||Trump +3.5 ||Trump +0.5||Biden +3.5
So why are forecasters hesitant to admit how bad the polling was in 2016? All election models are simple garbage-in-garbage-out propositions, and an admission that the polls can be garbage is an admission that the forecast might very well be garbage too. The bottom line is that they rely financially on you believing that the next
forecast they make will be accurate. Their bread and butter is people believing in the predictive power of their models, and if you don't have faith in their raw data then you might not tune in for their interpretation of it. That's the generous take. The not-so-generous take would be that they know that there is psychology in play, and that their forecasting doesn't just predict the vote - it can move it. When your team is hopelessly behind, you leave the stadium to beat traffic. The Daily Kos just told its left-wing readers that "people are motivated by winning, not losing."
How many Trump voters stayed home in Wisconsin that night because Hillary had a six point lead and an 84% chance of winning?
The million dollar question, of course, is "will the polling be more accurate in 2020?" Shown in the rightmost column is the current polling for the same set of critical swing states. If the pollsters learned their lesson and corrected the flaws in their samples and methods, then the only
interpretation is that Trump will unequivocally
lose the election. Even if you subtract out 2016's bias from the current figures, Trump is facing an uphill battle with little hope. There is, however, one more possibility: that the polling now is even worse than it was in 2016. In that case, it's Biden that is in for a rough night. Could this possibly be the case? The fact that most forecasters and prediction markets are calling it a 70/30 race right now tells you that they either don't believe these polls, or believe that the polls are about to tighten substantially in the coming weeks.
Will Arizona Flip?
September 29, 2020
Since the familiar red-state/blue-state walls were drawn in 2000's Bush V. Gore, Arizona has voted Republican in every Presidential election. The Grand Canyon State invariably gets marked as a swing battleground every cycle, and every cycle it lands comfortably in the GOP's corner. Once again the polling suggests that Democrats will carry the day in Arizona in this year's showdown.
Arizona is the only state that was not in the tossup category already
that has shown Biden polling ahead of Trump by at least three points. For this reason, it's been added to the tally of close states in the forecast map. The forecasters who think Arizona is going blue this year nearly unanimously cite a singular factor: demographics. Hillary Clinton won roughly eight out of ten Latino votes in Arizona in 2016, and the Democrats are hoping that ever-shifting demographics away from "old and White" to "young and Latino" will translate into a blue Arizona.
However, given Arizona's history of the actual vote coming in considerably more Republican than early polling would indicate, the GOP has a strong case to make for once again carrying the state. Nationwide, Donald Trump won 28% of the Latino vote in 2016, and his numbers with the demographic have only improved over the last four years. If the historical early polling inaccuracies repeat, and Trump wins at least a quarter of the Latino vote in Arizona, then it will be another easy win for the GOP.
Which way will Arizona break? Watch for quickly narrowing polls in late October. If Trump moves rapidly to match Biden's polling right near Halloween then you can assume 2016 is repeating itself and the GOP had nothing to worry about once again. If Biden's current three point lead widens, however, it could finally be the flip the Democratic Party has seen on the horizon for the last twenty years. If Biden carries Arizona, odds are he is the next President.
Biden Goes Anti-Riot
September 9, 2020
Three weeks ago, Joe Biden had a commanding lead in the polls and was keeping his mouth shut, and largely staying out of public view -- the traditional strategy of a frontrunner with a wide margin. However, public sentiment rapidly shifted against the ongoing violence taking place in America's urban areas. ANTIFA rioting in Portland entered its 100th consecutive day, there were high profile killings in Kenosha, Wisconsin and Portland, Oregon, and the overall death toll from BLM and ANTIFA rioting surpassed 30, with injuries in the thousands, hundreds of businesses burned to the ground, and property losses projected to top $2 billion. As riot-fatigue set in, Joe Biden's poll numbers began to plummet, and the RCP betting-odds average fell right along with them. Just as the betting odds fell to coin-toss territory, Biden finally made public statements against the rioting -- and it seems to have helped.
Biden's public statements that he condemns violence across-the-board "no matter who it is" appear, for now, to have stopped his precipitous fall in the polls. The betting odds are now bouncing back in Biden's favor -- sitting at around 53/47 as of today. There remains a large disconnect between state-by-state wagering on PredictIt and the contract for overall-winner, which technically represents an arbitrage opportunity, but one that is very hard to exploit because market participants can't be sure which
states are wrong -- but the current contract on the overall winner signals that they know some of the state-by-state contracts surely are. For example, PredictIt participants believe that Biden has a 65% chance of Winning PA and WI (and AZ!), and a 70% chance of winning MI -- but only a a 59% chance of winning the election. If those state contracts are accurate, Joe Biden's chances of winning the election are closer to 85%. Talk about serious 2016 Déjà vu vibes.
August 25, 2020
Historically, the opposition party typically sees a bump in the polls after a favorable VP pick, and during the party convention. We're now several days post Democratic National Convention and it's been more than a week since the Kamala Harris pick, but there hasn't been a discernable bump for Biden in the RCP average betting odds. In fact, Trump's numbers have been steadily improving over the past 30 days. So why are the Biden bumps missing?
The explanation kindest to the Biden campaign is that betting markets had already priced in a Kamala Harris pick, and that COVID driving the conventions online leads to lower viewership, less engagement, and less energy. Scott Adams, the Dilbert
cartoonist with an uncanny insight into Trump-era politics, called a Kamala Harris pick for the Dem ticket back in December of 2019. On the other side of the aisle, Politico
leaked an article naming Harris as the VP pick on July 28, later chalking it up to a website error (complete with graphics and a congratulatory quote from Biden). There didn't seem to be any widespread whisperings of anyone other
than Harris getting the nod. It's probably safe to assume that people betting real money on the election had already priced in Harris as the VP pick.
Even with disabled chatrooms and YouTube comments on convention videos, online viewership appears to be much higher than 2016. Television viewership, however, is down as much as 48% compared to the DNC ratings from four years ago. Don't assume that is a bad thing for Joe Biden. He's the frontrunner in the polls and the betting markets, and, if the numbers are to be believed, then it's his race to lose -- and keeping a low profile is a perfectly valid strategy. A lack of a convention-bump for Biden may end up being fairly meaningless if his 14 point lead in the RCP betting average is real.
Are We Seeing a Repeat of the Polling from Clinton V. Trump?
July 22, 2020
Everyone remembers when Hillary Clinton had a commanding lead in the polls in Michigan, Pennsylvania, and Wisconsin -- with Wisconsin's election day RCP poll average showing a dominating 6.5 point advantage over Donald Trump. She would lose all three states, with the polls in Wisconsin being off by a staggering 7.3 points. Joe Biden currently shows a 6.0 point lead in Wisconsin in the Real Clear Politics polling average. This raises the question: is Biden that far ahead, or are we seeing a repeat of 2016's forecasting disaster?
To answer that question accurately would require inside pollster information that few have access to, but we can at least assume that the polling outfits are going to conduct more
polling in the rust belt than they did in 2016. Some of 2016's inaccuracy was from a simple lack of polling in the (presumed-to-be-safe) states where the largest upsets were, and that is unlikely to be repeated this cycle. The polling error resulting from bad sampling, however, is likely to repeat - as the polls that disclose their sampling methodology show little change from 2016, and tend to oversample Democrats and generally do a poor job of distinguishing registered and likely voters.
The other question that is circulating in the forecasting community is "does Trump have secret voters?" These are voters who do not indicate support for Trump publicly or to pollsters, while fully intending to vote for the President. Nate Silver's 538
definitively states that this is an impossibility, and also stated that conclusion almost immediately after the 2016 election.
For a forecaster who missed 2016 by five key states, and over 90 electoral votes
they seem a tad overconfident in drawing this conclusion. Amidst a peaking cancel-culture, riots, doxxing, and targeted threats it's probably safe to assume that there are more than a few people who play their Trump cards close to the chest. If you lived in a deep-blue urban area, would you put a Trump sign in your yard? Even if it's just a point or two in the polls, that's enough to draw some of these races within the margin of error and dead-heat territory. Speaking of dead heats...
ElectoralMap.net's Favorite Tied Electoral College Scenario
Suppose that today's polls in the rust belt states are corrected for 2016's polling error. Currently, Trump would still lose Pennsylvania and Michigan, but would win Wisconsin. This presents an interesting possibility for an Electoral College tie, if President Trump were to also lose his single electoral vote from Maine's split vote. This scenario, however unlikely, could get extremely interesting...and messy. The issue of faithless electors (electors who do not cast their vote as the state did) would go from a historical curiosity to a profound historical event. In 2016 there were seven
electors who cast their votes for someone other than the winner in their state - more than enough to throw a close, let alone tied, election into chaos. Supposing that neither candidate crossed the 270 threshold, the election would then turn to the newly elected House Of Representatives which would convene immediately and cast one vote per state, with a majority of states being required to declare a winner. The process from that point forward generally favors Republicans, but this would definitely be in the realm of the strange where anything could happen.
Civil Unrest Helps Biden
June 23, 2020
The protests, rioting, and looting of the last few weeks has been a boon for Joe Biden on most prediction and betting markets. Before the unrest, Trump held a roughly 8 point advantage over Biden. Biden now holds a massive 16 point lead over Trump, representing a nearly 25 point swing in the betting market averages. This data runs counter to what many conservative pundits have asserted, that Trump would be the beneficiary of the lawlessness and destruction.
There are, however, some silver linings for the GOP in the data. Most of the wagering on the election takes the form of overall-winner contracts. When you drill down into the state-by-state polling, Biden's lead in the polls narrows substantially -- and is weak in the places he needs it most. In Wisconsin and Pennsylvania he holds a five point advantage, and in Michigan it's eight points. These leads are even smaller than the polling suggested for Hillary Clinton, and Trump upset for the win in all three states. The national election will probably hinge on those three states once again, assuming Trump can hold Florida and Ohio. Those wagering on overall-winner contracts may be underestimating Trumps position in the Electoral College.
Republicans also point to record-breaking gun sales figures coming out all across the country as proof that the "silent majority" is less than thrilled about the ongoing unrest that has now left over twenty people dead and hundreds of businesses burned to the ground. Crime is also surging in major cities across America, with Chicago seeing one of its all-time worst weekends with over 100 people shot
over the Father's Day weekend. Historically, these indicators point to strength in turnout for the GOP, but the election is still a long way off. Four and a half months is an eternity in an election cycle, just look how far the pendulum swung in one.
Prediction Markets Defy Polls
May 11, 2020
Polling data is still thin, but there have been three recently-conducted polls for a general election faceoff between Biden and Trump. Monmouth, The Economist, and CNBC all have published results as recently as May 6th -- and all three have Biden showing a lead over Trump. As of May 11th, prediction markets Betfair, PredictIt and PredictWise remain unbudging with Trump showing a lead over Biden. This is the first unmistakable disconnect of the season between polling and prediction markets.
This disconnect seems to hold even if you zoom out and look at a wider sample than just these three polls and three prediction markets. As of May 11, 2020, the Real Clear Politics national average of polls shows Biden with a +4.4 spread, while their Betting Odds
average shows Trump with an 8.1 point lead for Trump. Even this early in the season, one has to wonder why there is a healthy 12 point gap between what the polls are saying and how people are actually betting with real money. Nate Silver's fivethirtyeight.com site is not currently showing a spread on their main politics page, opting instead to run Trump's Presidential approval number, currently 44%. We already know that metric has marginal utility, as Barack Obama's approval percentage in May of 2012 hovered around 47%, and he ended up sailing to an easy victory that November.
Sanders Out, Betting Odds Unchanged
April 16, 2020
A great deal has happened in the last two weeks. Bernie Sanders suspended his campaign, much to the surprise of his supporters, making Joe Biden the presumptive nominee for the Democrats. The Democratic National Convention was moved to August 17th to give the planners more time to monitor what happens with Coronavirus. The first stimulus payments started getting to taxpayers with direct deposit information on file with the IRS. The US stock market indices have recovered a bit of their losses, and there has been good news on the numbers front in the battle against COVID19. Through all of this, the prediction markets have basically signaled that this information was already priced into most forecasts - barely budging at all. Trump maintains a 5 point edge on PredictIt, Betfair gives him a 54.6% chance of winning, and PredictWise has Trump holding steady at 54%.
Of note is that none of these figures registered even a small blip upon two prominent Biden endorsements, those of Sanders himself and former President Barack Obama. Also notable is that Trump's chances of re-election have barely budged despite a bounce in his approval numbers. All of this suggests that forecasters are still starved for data to make a meaningful prediction, and despite the busy appearance of the last two weeks, most of it was fully anticipated.
Trump Holds On To 54% Chance of Win
April 2, 2020
We're a solid two to three weeks into the Corona Virus shutdown and there has been some back-and-forth on a few of the prediction markets -- but not as much as you'd think. Donald Trump has maintained a 54% chance of re-election on Predictwise, a 49% chance on PredictIt, a 52% chance on BetFair, and HyperMind gives him 48%.
2020 Season Kickoff
March 18, 2020
ElectoralMap.net is now live for the 2020 election season. As usual, the House, Senate, and Governor's races will be tracked in addition to the battle for the Presidency. The forecasts will be based on an amalgam of curated and weighted polls and prediction markets. Find out more about which data sources will be used by following regular posts here.