Dec 7, 2014

Final Update
Mary Landrieu has been defeated handily in the runoff election in Louisiana giving the Republicans their 9th seat gain in the U.S. Senate. Despite a large bias against the GOP in this year's polling data, correctly forecasted the outcome in every state in the "tossups go Republican" scenario, and missed only one state (South Carolina) in the official forecast.

Contrast this year's polling performance with that of 2012: in September of 2012 many forecasts were exactly right in all 50 states for Barack Obama's re-election bid. In September of this year pollsters were still talking about a 2 seat gain for the GOP, when in reality, more than four times that number of seats would be gained. Just something to think about heading into the big race in 2016, which should be interesting.

See you then,

Nov 14, 2014

Alaska goes to the GOP
Alaska will be the 8th seat gain for the Senate Republicans, and Louisiana will have a runoff in early December. The potential for nine seats is there, which would mean the "tossups go Republican" forecast was dead-on.

Nov 8, 2014

Results finalizing...
Ballots are still being counted in Alaska, and Louisiana has headed to a runoff election. But the rest of the races in the Senate are settled, and the Republicans have picked up seven seats. The forecast was for eight seats total. If Alaska and Louisiana go Republican, the nine seat pickup will match the forecast plus North Carolina -- so far the only state which was wrong in the forecast. The GOP stands at +12 in the House, and +2 governorships as of today.

Nov 5, 2014

Forecast accurate, in spite of bad polling data
As Nate Silver of FiveThirtyEight noted, the polls had a significant Democratic skew this year, averaging six points in the ten most competitive races. In spite of that large skew in the polls, the forecast proved accurate. As of this writing, the forecast of +8 seats looks like it will be one shy of the actual result. The "tossups go Republican" forecast of +9 may be dead on. Louisiana will go to a runoff, which Landrieu will presumably lose. Alaska and Virginia are still pending as of this writing.If Virginia goes to the GOP it would be the only race that was dead wrong according to the forecast.

Given the skewed data, the prediction that there was an 82% chance that the GOP would take the Senate was probably low. The Washington Post seemed to see through the bad polling data and forecast a 97% chance that the Republicans would take the Senate.

Princeton Election Consortium65%
New York Times70%
Five Thirty Eight76%
Washington Post97%

Check back for more updates as races are finalized.

Nov 3, 2014

Final Forecast
The Monte Carlo Simulation stands at 82% for the overall probability that the GOP will capture control of the Senate. predicts the Republicans will gain eight Senate seats, picking up Montana, South Dakota, Iowa, Colorado, Alaska, Arkansas, Louisiana, and West Virginia. However, if the polling data proves to be inaccurate and the GOP fails to capture the Senate, still predicts a gain of +4 Senate seats for the Republicans. That would bring the balance of power in the Senate to 49/51. If every card falls into place for the GOP, on the other hand, they could walk away with a +9 seat gain, for a power balance of 54/46.

Forecast (GOP)Range (GOP)
Senate+8+4 to +9
House+3-7 to +19

Oct 31, 2014

Polls slide in North Carolina
Incumbent Senator Kay Hagan(D) slid a little further in the polls, officially putting North Carolina into tossup territory. If the GOP were to capture all tossup seats, the forecast right now yields a nine seat pickup for the Republicans. The election simulator hit a new high of 83% for the overall probability that the GOP will capture control of the Senate.

Oct 27, 2014

Polls in Kansas and Georgia offer Democrats a ray of hope
The bad news for the Democrats in Colorado was partially offset in today's update by an uptick in the likelihood that Georgia will go to a runoff and possibly end up in Democrat hands, as well as a big surge in Kansas for the independent candidate. In the "tossups go to the Democrats" scenario we see a 50/50 map, where Joe Biden would cast tie-breaking votes and control of the Senate remains Democratic. But this scenario also has Greg Orman (I-KS) deciding the fate of the Senate, and caucusing with the Democrats. If he were to caucus with the Republicans, the GOP tally would be a winning 51 seats.
However, the polls weakened slightly in North Carolina for incumbent Kay Hagan giving the Republicans a slim, but real chance at adding that seat to their pick-up list. This drove the Monte Carlo simulation to a new high of 77.9% for the probability that the Republicans will take the Senate. If everything were to go just right for the GOP on election day, and all of the seats considered contested fell their way, the Republicans could be looking at a nine seat pick-up.

Oct 23, 2014

Polls worsen for Democrats in Colorado, Arkansas, Louisiana
The aggregation is now showing a sharp uptick in the likelihood of Republican seat pick-ups in Colorado, Arkansas, and Louisiana. Republican support also moved upwards in Georgia, a seat one once seen as a possible pick-up for the Democrats. The nearly 10% increase in the probability of a Republican victory in Colorado sent the Monte Carlo simulation to a new high of 74% for the overall chances of the Republicans taking the Senate. Kansas continues to be a wildcard tossup, with independent Greg Orman polling dead even with the Republican incumbent. The table below summarizes various forecasting shops' estimation of the probability that the Republicans will take control of the Senate, as of today:

Huffington Post61%
Five Thirty Eight66%
New York Times67%
Daily Kos69%
Washington Post94%

Oct 16, 2014

Forecast hits +8 Republican seats, 68% chance of taking the Senate
For the first time this election season the forecast indicates a Republican pickup in Colorado, bringing the overall forecast to +8 seats. Combined with a drop in the polls for the independent in Kansas, the overall chances of the Republicans taking the Senate has also reached a new high of 68%. The nearly ten percentage point surge in the overall chances for Republican control of the Senate comes from the fact that taking eight seats without Kansas means that the independent challenger in that state could caucus with the Democrats and it wouldn't matter - the Republicans would still have a winning 51 votes. Also adding to the surge is polling in Alaska, once forecast to remain Democratic, which has also trended Republican.

As the likelihood of any "October surprises" drops off, and the polls become more frequent, we are starting to enter the final phase where the forecast tends to trend distinctly in one direction. In 2008 Barack Obama really took off in the forecast in late September after the Wall Street meltdown. In 2010, both the House and Senate forecast started a strong trend toward the actual results in mid October. In 2012 the forecast was dead-on in early September, then diverged until mid-October, and then started trending back towards accuracy on October 25. If 2014 repeats the past then the trending right now indicates a Republican takeover of the senate. If this is indeed the case, then the Washington Post's Election Lab deserves some special recognition for being way out ahead of other forecasts, having the Republicans' overall chances for a Senate takeover at over 90% for some time now.

Oct 10, 2014

Denver Post endorses Republican in dead heat Senate race in Colorado
The editorial board of Colorado's largest and most influential newspaper has just endorsed Republican Cory Gardner in the dead heat race against incumbent Mark Udall. The endorsement comes as a bit of a surprise from the traditionally left-leaning newspaper, and could make a difference in what is a true 50/50 race at this point. If Colorado went Republican, it could be the eighth pickup the Republicans need to neutralize the emerging independents in Kansas and South Dakota who would likely caucus with the Democrats should the Republicans fail to capture the Senate with a firm 52 or 53 seats.

Oct 1, 2014

Forecast hits +7 Republican seats in the Senate
For the first time, the aggregation performed by predicts that the Republicans would take the Senate if the elections were held today (see map). Polls have tipped slightly in Alaska and Iowa indicating narrow Republican victories in those states and leading to a 52 seat forecast. Getting to 52 seats is critical for the Republicans because it negates the wildcard factor in Kansas. Even if Greg Orman were to win in this scenario, and caucus with the Democrats, the Republicans would still hold 51 seats -- enough to take the Senate.

The Monte Carlo simulation now indicates that the Republicans have a 58.7% chance of taking the Senate.

Sep 25, 2014

Simulation now allows configuration of how Orman will caucus
If Greg Orman wins in Kansas, he will most likely caucus with the Democrats which would result in a loss of one seat for the Republicans. But he has made remarks to the effect that he might caucus with the Republicans if they achieve a majority. You can now configure the Monte Carlo simulation to make Orman chose a side according to various criteria, if he wins Kansas. The most likely (default setting of the simulation) scenario is that he caucuses with the Republicans only if they get to 52 seats. It's unlikely that he'd stay with the Republicans if he was the 51st vote. For this reason, it's statistically identical to simply counting him as a Democrat for determining the overall probability that the Republicans will take the Senate. All that is really affected is the number of seats involved in the most likely outcomes -- there is a gap at 51 Republican seats because Orman would most likely give the Senate back to the Democrats in that scenario.

Sep 23, 2014

Model updated to reflect uncertainties in Kansas
After the withdrawal of Democrat Chad Taylor, the race in Kansas is now between independent Greg Orman and Republican Pat Roberts, and is more or less a tossup at this point. When calculating the overall odds of the Republicans controlling the Senate, this creates a lot of uncertainty because we don't know how independent this independent actually is. Orman has made remarks about caucusing with whichever party has a clear majority - but this fall's election is more likely than not to result in a Senate that is within one seat of 50/50. How would Orman factor in Joe Biden's tie breaking vote?

Would Orman caucus with the Republicans if he was the 51st Republican seat? Probably not, based on his policy positions and the fact that he has run as a Democrat before. Being more of a left-centrist than a right-centrist, is now treating an Orman victory in Kansas as a seat for the Democrats. This reduces the Republicans' probability of controlling the Senate by about 7% overall.

Sep 20, 2014

Senate Monte Carlo Simulation
We have the polling data available for each state where there is a Senate race, but how do we combine them all to determine the probability that the Senate will change hands? The answer is, we simulate thousands of elections using the individual state probabilities. Then we analyze the results of those thousands of runs and see what percentage of them result in the Republicans taking at least six seats. This is called a Monte Carlo simulation.

The Republicans' overall chances of taking the Senate stand at 57%, according to today's Monte Carlo simulation run for this fall's election. The simulator is fully configurable -- you can adjust the probabilities in each individual state, the number of runs to perform, and the way the pollster data is mapped to the individual probabilities. Just his the "reset defaults" button to get back to the official forecast data. The simulation displays a bar for each number of Republican-seat-pickups, indicating the probability of that number of seats being gained at the top of the bar. The overall probability that the Republicans will gain at least six seats (needed to control the Senate) is the sum of all the bars for +6 seats and greater. Note that no two runs will be exactly the same, but the more iterations you allow the model to run through, the more precise the final numbers will converge.

Sep 11, 2014

Forecast remains stable
Impending U.S. military escalation in the middle east does not appear to be helping or hurting either party, and the forecast remains steady at a pickup of 2-9 Senate seats for the Republicans. The Republicans are banking on almost guaranteed pickups in Montana, South Dakota, and West Virginia. That means they would need to get three of the remaining six tossups to win control of the Senate. Two of those tossups, Louisiana and Arkansas, are leaning Republican at the moment. The most likely 51st seat in that scenario would be North Carolina, Alaska, or Iowa. Colorado will be the most difficult pickup for the Republicans. The bottom line is, there are a number of paths to 51 seats for the GOP.

Aug 21, 2014

Stirred, not shaken
With many primaries determining the opposition candidates, the forecast received a slight stirring with no real shake-ups. As we move toward an defined ballot, pollsters are telling us that the electorate is responding basically the same way they did to a generic ballot. The current forecast is that the Republicans will pick up from 3 to 9 Senate seats. Taking control of the Senate is looking likely for the Republicans, but is by no means guaranteed. They are almost assured a 3 seat pick-up, meaning that they only need to secure half of the remaining tossup races to get to the magic 51 seats.

The House forecast indicates almost no chance of a shift in power, with the current forecast indicating a possible swing of +5 seats for the Democrats to +15 seats for the Republicans -- neither outcome changing the current Republican hold on the House. The gubernatorial polling indicates anywhere from a pickup of six mansions for the Democrats to four for the Republicans.

May 13, 2014

Republican Senate Forecast Hits 50 Seats is now combining the output of over seven different aggregators to form its forecasts, and for the first time the Republicans are predicted to hit the 50 seat mark in the U.S. Senate. With two dead heats and four other races that are very close, the Senate could easily go either way. If the Senate did hit an exact 50/50 balance of power Vice President Joe Biden would cast tie-breaking votes, effectively meaning that the Democrats would retain control of the Senate. Vice President Dick Cheney cast eight tie-breaking votes in the Senate during George W. Bush's presidency. Al Gore cast four tie-breakers as V.P. under Bill Clinton.

Apr 9, 2014

Bias and error in incumbency-based maps
In forecasting this year's Senate races, many sites are using incumbency-based maps. This style of map shades a state red if the incumbent is a Republican and blue if the incumbent is a Democrat. A lighter shade of the base color is used if the polls show the seat to be competitive. This style of map quickly conveys which party holds a given seat, and whether or not that seat is threatened, but that is the limit of what it can depict. It is a poor map style for accurately depicting the true state of polling data. For example, shown below is today's polling aggregation shown on both an incumbency-based map and a map shaded without regard to incumbency.

The incumbency map inaccurately depicts the current state of the races in seven states. Polling data forecasts fairly certain Republican wins in Montana, South Dakota, Arkansas, and West Virginia. Additionally Alaska, Louisiana, and North Carolina are dead heat races. All seven of these states are shaded light blue on the incumbency map. When shaded according to actual polls, four of those states become light red and three become grey. believes this makes the incumbency map a poor, and possibly biased, choice for displaying a forecast.

To eliminate this bias uses shading according actual polling, not incumbency, and marks states where a Republican incumbent is forecast to be unseated with a "+D." States marked with "+R" are where a Democrat is forecast to be unseated. This conveys all the information that an incumbency map does while keeping the shading true to the actual polling data. The Senate map also has toggles at the top allowing the user to see the same map under an incumbent-wins scenario, as well as tossups being awarded to either party.

Mar 25, 2014

Nate Silver releases Senate forecast
FiveThirtyEight, the forecasting shop run by statistics whiz Nate Silver, has released a Senate 2014 prediction that "might be thought of as a Republican gain of six seats -- plus or minus five." FiveThirtyEight correctly forecast 50/50 states in the 2012 Presidential election garnering accolades, most strongly from Democrats who were happy to see the forecast come true. The present reaction from the DNC and DSCC to Silver's Senate forecast sounds more like his Republican detractors back in 2012. Everyone loves a crystal ball, as long as it's telling them what they want to hear. It would have been most interesting to see a Silver forecast for the recent Florida 13 race, and the political reaction to it. FiveThirtyEight did not make a call on that race, and probably wisely so. Other forecasters almost unanimously failed to make an accurate prediction based on the available polling data.

The FiveThirtyEight forecast has been incorporated into the aggregation, and slightly red-shifts a few extremely close states. Notably, Montana shifts out of tossup status while Alaska, Louisiana, and North Carolina all shift into dead heats. The overall forecast remains basically the same, at GOP+4 seats, with three 50/50 races and one slightly Democratic tossup.

Factoid: who will see the fewest political ads?
It's impossible to say for sure which state will have to endure the least campaigning, but there are five states with no Gubernatorial race and no Senate race for 2014. They are Indiana, Missouri, North Dakota, Utah, and Washington. Of those states, Missouri and North Dakota have the least competitive House races.

Mar 12, 2014

Forecasters miss first House race of 2014
The election in Florida's 13th Congressional District to replace Republican Bill Young, who died in office, has gone to Republican David Jolly despite election forecasters giving a slight edge to Democrat Alex Sink. When the votes were tallied, Libertarian Lucas Overby carried 4.8% of the vote - enough to deny a clear majority to Jolly who carried 48.5% of the vote to Sink's 46.6%. Back in 2013 the pollsters were giving Sink a lead upwards of 13 points, which narrowed going into 2014. By February the polls were forecasting a 2 point race, but which candidate was ahead depended on which pollster you asked. Conservative RedHorseRacing commissioned a poll showing Jolly ahead by 2%, whereas Public Policy Polling had Sink ahead by 3%. Jolly's actual margin of victory was 1.9%.

Republicans are calling the win a bellwether for November, and Democrats are quick to point out the limited meaning of one election - roles that would no doubt reverse with a reversal of the results. But the fact remains that the forecasters were wrong. They placed too much weight on the fact that Obama carried this district in both 2008 and 2012, and they were unable to filter the noise out of the polling.

Mar 2, 2014

Source of data for this year's forecast. based all its predictions on data from since 2008. The Irish company was the world's highest volume prediction market for U.S. political events and it performed amazingly well, correctly forecasting 49/50 states in 2012 and missing the Electoral College tally by a single vote in 2008. Unfortunately shut down in late November of 2012; more on why below. Although the company has indicated that "Intrade 2.0" is under development, we don't yet know when it will go online. There are no other real-money prediction markets with the kind of volume and the scope of contracts that Intrade offered, so a different approach for is required for the 2014 midterms.

Prediction markets are unique in that there is no single person, polling outfit, or think tank modeling the election, analyzing the polls, and determining the odds of a particular outcome. Rather, a multitude of individuals do the legwork - buying and selling contracts on the outcome of an event and the market clearing price of those contracts tells us the odds. It's a proven model, and quite remarkable in its simplicity. It's the shortest and most elegant route to an aggregation of all the polling data and analysis available to the public. In keeping with the philosophy of aggregating as much data as possible, will make its 2014 forecasts by attempting to build an amalgamation of other aggregators' data. It will be a simple, unweighted average of as many sources as possible so long as they are kept up to date.

So what happened to Intrade?
Intrade suffered a series of setbacks starting in 2010 when U.S. banking regulators prohibited U.S. citizens from funding their accounts with credit or debit cards. In May of 2011 CEO and founder John Delaney passed away on Mount Everest, just fifty meters from the summit. Then In 2012, just after the election, the U.S. Commodity Futures Trading Commission (CFTC) filed suit seeking an injunction against U.S. citizens trading on the site, claiming that Intrade contracts were "commodity options" and that the company was soliciting U.S. citizens to buy and sell unregulated securities. In the end, Intrade was forced to suspended the accounts of U.S. citizens which quickly decimated trading volume on the site. Shortly thereafter the site ran afoul of Irish banking regulations as some sizeable "financial irregularities" were discovered on the company's books, and the site closed down.

Although the company is far from blameless in its own downfall, the power and reach of U.S. government regulators is well illustrated in the story of Intrade, an Irish company located on Irish soil. Despite correctly forecasting two presidential elections, Intrade was treated by the U.S. government as just another illegal online casino. All the usual buzzwords (gambling, money laundering, terrorism) were bandied about, and the very thing that made it so successful (lots of American participants) is what ultimately painted the target on its back. It is unlikely we will ever see a prediction market with political contracts operate with the blessing of the U.S. government, as the official position of the CFTC remains that such contracts "involve gaming and are contrary to the public interest." Read more about the death of Intrade at

back to top

All Posts

Dec 7
Nov 14
Nov 8
Nov 5
Nov 3
Oct 31
Oct 27
Oct 23
Oct 16
Oct 10
Oct 1
Sep 25
Sep 23
Sep 20
Sep 11
Aug 21
May 13
Apr 9
Mar 25
Mar 12
Mar 2