Democrats’ Chances are Overrated: The 2018 Midterm Elections

By: Tsahi Halyo

Democrats are favorites to win though a wave is far from certain

This coming November, voters across the country will have the chance to show their support (or anger) for the current White House administration through the ballot box. Given the Mueller investigation and the possibility of an open Supreme Court seat, control of Congress is paramount for both parties to achieve their political ends. With such a tense situation, one could reasonably assume that most voters would turn up on election day and vote. Unfortunately that isn’t the case – on average, only 39% of eligible voters come out to vote on midterm election years. This represents a marked difference to the presidential elections when the average turnout rate is 56%, nearly 1.5 times greater than the midterm turnout. However, despite this trend, the President’s low approval rating means that many assume that there will be a large Democratic wave in the 2018 midterm that will flip the House and perhaps the Senate. Unlike cable news pundits, let’s actually examine some of the claims that can possibly affect the Democrats’ chances of retaking Congress. These include: the claim that midterms usually go against the party controlling the White House; that the GOP has a midterm turnout advantage and that the generic ballot accurately predicts midterm outcomes.

Preliminarily, for the purposes of establishing a baseline, it’s always worth looking at online betting markets in which people can gamble on the results of future elections to see what people think the odds for a Democrats flipping control of a house of Congress are. Presently, the betting market  gives Democrats 2:1 odds of retaking the House while only giving them 1:2 odds of retaking the Senate. For some context, consider that Intrade, a popular betting market, gave the GOP a 55% chance of winning the House in 2010 and a different betting market gave Hillary a 89% chance of winning the presidential election on the night before the election. Thus, it is wise to take betting odds with a grain of salt. The difference in the odds given for the House and Senate races is grounded in the difficult Senate map this year in which Democrats will have to defend 26 seats while the GOP will only have to defend 9 (and, as we know from past elections, the incumbent normally has an advantage).

Now, let’s examine the first claim: midterms usually go against the party in control of the White house. Of the 18 midterms since the end of World War II only two have gone in favor of the president’s party and both of those elections occurred in unique circumstances. The first was in 1998 after the House unpopularly impeached Clinton and the second was in 2002 after 9/11 when the president’s approval rating was greater than 70%. On the other hand, the usual result of midterm elections is a loss for the president’s party. As seen in figure 1 below, the loss of seats can range anywhere between 4 seats and 63 seats (the 2010 midterm really was a disaster for democrats). On average though, the president’s party loses 25 seats in the House in a midterm election. The House has 435 seats meaning that a party needs to control at least 218 of them to control the House. Presently, republicans control 238 seats; if they lost the average 25 seats in the midterms, their 238 seat majority would be reduced to a 213 seat minority. This obviously portends ominous news for the GOP retaining their 8 year House majority, and is in line with the odds given by betting markets.

Figure 1: Midterm Election Swings with Regard to the President’s Party Affiliation

The second claim argues that the GOP has a midterm turnout advantage. As it turns out, Republicans are more likely to vote in a midterm election than Democrats. In races since 1978 the GOP has on average had a 3% midterm turnout advantage. That is, the GOP did 3% better with midterm voters compared to their margin among registered voters. A closer look at the data reveals an insight in line with our first claim. Under a Democrat president, the mean Republican turnout advantage was 5% irrespective of the Democratic president’s national popularity whereas under a Republican president the mean turnout advantage was a paltry 1%. And, when their president was unpopular with an approval rating below 40%, as was the case with George Bush in 2006, they had no turnout advantage. Clearly the magnitude of their turnout advantage is correlated with their party being out of the White House and the popularity of the incumbent president. Given that Donald Trump’s popularity is on par with Bush’s in 2006 (they both have a ~40% approval rating) it’s quite possible that the Republican turnout advantage won’t materialize.

The third claim concerns the accuracy of the generic ballot in predicting the midterm margin of victory (as a percent of the vote, not in seats) this far out from the election. The generic ballot is based on polls that ask respondents to name which party they plan on voting for in the coming midterm election. The polls are averaged (it’s a weighted average) to produce a smoother prediction. Historically, predictions made by the generic ballot a year out from the midterm elections were highly correlated with the results of that year’s midterms. When considering the cases of every midterm since World War II, the correlation is r = 0.9 (if r = 1, then the predictions would be perfectly accurate; 0<| r |<1), an extraordinarily high correlation.

Figure 2: Correlation Between the Generic Ballot Margin and the Midterm Margin

As is seen in figure 2 above, the relation between the generic ballot margin and midterm results margin is linear. It’s trend is approximately –Picture4

The current generic ballot margin is approximately 7% in favor of the Democrats. The general trend would suggest that the GOP is going to lose the midterms popular vote by a margin of 8.5%. How that will translate into a loss of seats is a difficult question to answer. If Democrats win every House seat won by a Republican in 2016 by a margin of less than 8.5% then they’re only going to win ~15 seats, too few to retake the House. If they win every district that Clinton lost by a margin of less than 8.5% then they’re on track to win ~30 seats and flip the House. The reality will likely be an outcome in between those too.

In light of the above, it’s clear that Democrats are the favorites to win the House though their victory isn’t assured. It’s still possible, albeit unlikely, that the GOP will narrowly hang on to their majority. Also, given that the GOP controls the White House, Democratic gains in Congress are to be expected, though their magnitude may be shy of securing a majority in the House and almost certainly too small to secure the Senate. Due to Trump’s unpopularity, Republicans can’t hope rely on their midterm turnout advantage to stem their losses as, historically unpopular GOP incumbents negate their midterm advantage. The generic ballot is usually a good bellwether for ascertaining the magnitude of an electoral wave. However, in this case, the range of likely outcomes can allow either party to control the House in 2019, with Democrats as slight favorites; Republicans have much to fear but, Democrats shouldn’t be too arrogant – the election could still hold surprises for both sides.


Sides, J. (2014, November 03). The 2014 midterm election fundamentals (in 4 graphs). Retrieved April 15, 2018, from

F. (2018, January 09). Do Republicans Really Have A Big Turnout Advantage In Midterms? Retrieved April 15, 2018, from

Enten, H. (2017, November 10). There’s No Reason To Think Republicans Will Be In Better Shape A Year From Now. Retrieved from

Wall Street Journal. (n.d.). Retrieved from

Political Polls:The False Oracles of D.C

By: Tsahi Halyo

Even with information and data collection the easiest it has ever been, our collective society should not consider polling as an exact predictor of election results.

Few things are misunderstood in the political sphere as much as polls. While their numbers tempt us with the siren call of mathematical certainty, a close examination reveals them to be far more limited and down-to-earth in their claims. Polls suffer from many flaws that should condition one’s faith in them so generally speaking, no news organization should pretend any polling result exists with 99% percent confidence. This article will characterize the efficacy of political polls by describing three facts that stymie perfect polling today.

Fact 1: Polls aren’t as accurate as most people think

                                                               Average Polling Error

  Presidential State-Level  
Cycle Primary General Governor U.S. Senate U.S. House Combined
2017-18 5.2 6.0 4.1 5.1
2015-16 10.1 4.8 5.4 5.0 5.5 6.8
2013-14 4.4 5.4 6.7 5.4
2011-12 8.9 3.6 4.8 4.7 4.7 5.1
2009-10 4.9 4.8 6.9 5.7
2007-08 7.4 3.6 4.1 4.7 5.7 5.4
2005-06 5.0 4.2 6.5 5.3
2003-04 7.1 3.2 6.1 5.6 5.4 4.8
2001-02 5.2 4.9 5.4 5.2
1999-2000 7.6 4.4 4.9 6.1 4.4 5.5
1998 8.1 7.4 6.8 7.5
All years 8.7 4.0 5.4 5.4 6.2 5.9

Figure courtesy of FiveThirtyEight:

There is a widespread belief among members of the media and the general public that polling is usually quite accurate. This led many of them to feel betrayed when polls predicted a Clinton presidency and were in turn granted a Trump one. What was supposedly meant to begin a new era of Democratic political hegemony instead ushered in an era in which the GOP would cement its control of all three branches of government. However, this shouldn’t have been such surprise. While the media reports polling numbers it rarely reports the polling error. The above table presents the average deviation polls conducted in the final 3 weeks of the campaign had from election results. It shows that on average, polling of Presidential elections is off by 4%; that of Senate races if off by 5.4%; and that that of U.S House elections is off by a whopping 6.2%. Trump’s winning the election should not have been as surprising as many perceived it to be. In the key states of Ohio, Florida and North Carolina his polling numbers trailed Clinton’s by numbers smaller than pollsters’ average error in Presidential races. The same can be said about that election cycle’s Senate races. Polls aren’t a crystal ball and the public should not expect them to act like one.

Fact 2: Republicans are underrepresented in poll results.

Pollster Average GOP Bias
ABC -1.4%
CBS -1.75%
CNN -2.6%
FOX -4.1%
Pew -4%

Another reason to take polling results with a grain of salt is the fact that polls systemically undercount GOP support. When comparing polls, conducted the week before congressional elections and with election results a consistent, it becomes clear that Republicans have consistently outperformed the polls on election night. Though many reasons have been offered to explain this phenomena this article will only explore one. It’s conjectured that some Republican voters are embarrassed to admit they back a Republican to pollsters who, practically speaking, are strangers over the phone. They may feel that there is a stigma attached to being a Republican (e.g as a bigot; racist; homophobe, etc.) which leads them to tell pollsters that they back a 3rd party or remain undecided. Evidence for this theory can be found in the fact that undecided voters in Congressional elections disproportionately vote for Republicans. Further evidence for this can be found in the case of David Duke’s, a Klan leader’s, Senate bid. Though pollsters found that less than 5% backed him, he won 20% of the vote. Who could admit they planned to vote for a despicable man like Duke over the phone?

Fact 3: Polls temporarily dip in accuracy 10-6 months before the election.


The plot to the left displays the correlation polls have with election results (r-value) as a function of time. (Note: Higher correlation doesn’t mean that the polls are more accurate themselves but rather that they have a stronger linear relationship with that year’s election results.) Basically, it tells us how well we can predict election results based on current polling. In the case of midterm elections there is an astounding dip in predictive power 10 months before voting day that is only ameliorated 3 months prior to the vote. It is difficult to explain this phenomena. While it’s possible that it occurs due to the natural fluctuations of the political pendulum the polling doesn’t suggest that a clear oscillation against the leading party is the answer. It may be that before our hyper-news era of the Trump presidency, most voters had made up their minds who to back in the year before the election far before intense political news coverage began to cover political campaigns. Whether this trend will survive in our current environment remains to be seen.

It’s tempting after seeing polling’s flaws to discredit the entire enterprise. In fact, after the 2016 election many news services published articles expressing such a view. However, it’s better to see political polls as a tool of limited use. It can predict a blowouts going either way, but sometimes  can only tell you that an election will be close. There has been much hype surrounding Democrats’ chances of winning control of a chamber of Congress. Hopefully this fall we won’t repeat the mistake again of treating polls like a crystal ball.



(Polling data was retrieved from, analyses of such data are my own)

TABLE OF CONTENTS (n.d.) Retrieved from

(n.d.). (2018,May,31). The Polls Are All Right. Retrieved from

(n.d.). Retrieved from


Ethereum Daytrading: Like Tossing a Coin?

By: Tsahi Halyo

Many have been drawn to Ethereum trading by its promise of getting rich quickly but is that promise legitimate or just a product of a “bubble” styled craze?

Given the rapid rise in cryptocurrency prices and their extreme volatility, interest has grown in understanding cryptocurrencies price behavior. This article examines the short term price fluctuations of the second largest cryptocurrency, Ethereum, and through such explains the pitfalls of daytrading that currency.

Before we delve into short term Ethereum price behavior a few points must be touched. Firstly, a cryptocurrency generally has little intrinsic value. Traditional currencies, which are minted by nations, derive their value through the necessity of paying one’s taxes through them and their ability to be used as a medium of exchange. Stocks depend on the company’s performance and commodities have intrinsic value. Unlike those mentioned before, a cryptocurrency almost completely derives its value from speculative behavior. The little intrinsic value it does possess is usually derived either through its use in illicit Dark Web transactions or, in the case of Ethereum, through the use of Smart Contracts, those being contracts that use Ethereum to ensure payment. The fact that cryptocurrency prices are mostly derived from speculative behavior leaves them extremely sensitive to swings in public opinion, some of which reinforce themselves in a positive feedback loop in order to create massive price shifts. While examples of bubble-like price increases are plentiful (consider that one year ago Ethereum was worth less than $15 whereas now it’s worth more than $900), that isn’t to say that examples of large price decreases can’t be found. Recently, the price of Bitcoin dropped more than 20% after the South Korean government announced that it would ban cryptocurrency trading.

Secondly, as a result of this extreme volatility; long term predictions are difficult. The sensitivity of the price to its initial conditions, meaning that any minor deviation from the initial conditions quickly snowballs into a far larger deviation when employed for predictions, renders accurate long term prediction futile. This is comparable in some ways to the behavior of the weather. No model can predict whether it will be rainy or sunny on a specific day a year from now. This chaotic behavior bars precise long term measurements. The only prediction we can make is that the crypto-bubble will eventually pop. We don’t know when this will occur; however, due to the nature of Ethereum’s rapid rise in a manner reminiscent of other bubbles it is assumed that it will pop just like the others did.

Thirdly, it is important to note that cryptocurrency prices are moderately correlated. The correlation coefficient for percent changes in Bitcoin price vs percent changes in Ethereum price is r=0.46 with a p value of approximately 1 x 10^-68.  This means that a downturn for Ethereum likely reflects a downturn for Bitcoin and vice versa.

Ethereum price changes every 6 hours were recorded over a ten month period. All comments made henceforth concern price changes with that resolution. The short window ensures that the prices remain roughly the same as they were six hours prior. To normalize the price changes for their magnitude, the analysis was conducted on percent price changes (this meaning that the analysis checked if there was a 1% increase in price not a $9 increase in price). The analysis found that the percent price changes follow a Cauchy distribution with location 0.0025 and scale factor 0.014. A variable will have a Cauchy distribution if said variable is the ratio of two Normal random variables. The location of the distribution is the location of the spike and the scale factor is equal to half the size of the interquartile range.Picture2

The graph to the left shows the normalized distribution of percent price changes. The red line shows the Cauchy distribution and the blue rectangles are the raw data. Unfortunately, the Cauchy distribution is one of the statistical distributions that lends itself least to analysis.  Its cumulative distribution function (the function that returns the probability of a value less than the input occurring) is given by
Picture3.pngin which x_0 is the location and gamma is the scale. This function however results in the distribution to having any moments, a  specific quantitative measure. The nth moment of a distribution is defined by Picture4in which f(x) is the probability density function and c is the axis the moment is to be evaluated about. If c is 0, the first moment is the mean; if c is the mean, then the second moment is the variance. Since this integral is undefined for the Cauchy distribution’s density function the distribution doesn’t have a mean or a variance. Its mode is the axis of symmetry and the range between -γ and γ covers 50% of the occurrences.

Despite this lack of several key statistical properties, conclusions can be drawn given that Ethereum’s short term price behavior is well described by a Cauchy distribution. Firstly, the distribution of percent price changes has fat tails This corresponds to the large shifts in price that occur far more often than a Gaussian distribution would predict. This behavior can perhaps be understood as a result of the speculative nature of Ethereum trading and the positive feedback effect people have an price changes. This allows large shifts in price to happen more regularly since a minor sell-off can quickly escalate into a panic and a surge in a buying can lead to tulip-craze level increases in prices. Under a Gaussian regime, the large changes in price observed lie more than ten standard deviations away from the mean. This would mean that the probability of large price shifts occurring would be less than that of being struck by lightning thrice after having won the lottery; something which is clearly not the case.

In short, Ethereum trading is little more than a technologically “cool” way of betting. Given that short term price shifts have a nearly equal chance of increasing or decreasing means that betting on Ethereum is similar to betting on a coin toss. This effect combined with its tendency to reinforce price changes through a positive feedback mechanism allows for large price changes. Finally, the chaotic nature of Ethereum price changes renders long term prediction moot. Many people have made thousands off the Ethereum bubble and their gains have prompted many to test their skills in investing in the market. Unfortunately the data shows that being an Ethereum investor is nothing more than being a glorified gambler.


The Rational Statesmen: Donald Trump and Kim Jong Un

By: Tsahi Halyo

Though tensions are high between the United States and North Korea, through an analysis using game theory, the potential for a nuclear fallout may not nearly be as high as made out to be.

It is a commonly held assumption by those in the media that the recent war of words between President Trump and Kim Jong Un represents a dangerous turn away from diplomacy that vastly increases the chances of nuclear war. However, a game theory analysis of the current situation can show how the threat of nuclear war is exaggerated and why North Korea’s acquisition of nuclear weapons may stabilize the situation in the Korean peninsula. A full analysis is beyond the purview of this article; this article will restrict itself to merely analyzing the reason behind North Korea’s nuclear programme, why nuclear weapons act to stabilize a situation and the introduction of anti-missile technology works to undermine it, and why fears of a nuclear war are overblown.

Quick Game Theory Introduction:

To illustrate key concepts in game theory that will make the rest of the piece clearer, let us begin with a famous problem in game theory known as The Prisoner’s Dilemma. Consider the following scenario: two men are arrested at the scene of a murder (punishable by life in prison); however, the district attorney lacks the evidence to convict them of murder and instead only has enough evidence to convict them of trespassing (punishable by 1 year in prison). In an attempt to get at least one murder conviction the DA offers each prisoner a plea deal: if he confesses and betrays the other prisoner then he’ll walk free. If both accept the plea deal then they’ll both be sentenced to 10 years in prison. The two prisoners can’t communicate. What should each prisoner do?

(Player 1, Player 2)

Accept Deal Refuse Deal

Accept Deal

(10, 10)

(0, Life)

Refuse Deal (Life, 0)

(1, 1)

The above table represents the game, with the vertical axis representing Player 1 and the horizontal representing Player 2. In the parentheses the results are written as:  (Player 1 sentence, Player 2 sentence).

To find the optimal strategy for each player let us consider a Nash Equilibrium, the situation in which any unilateral change of strategy (changing from refusing the deal to accepting it) will impact the player negatively. In this case there is only one equilibrium point at which both prisoners confess. If either player decides to unilaterally refuse the deal he’ll be sentenced to life in prison. This is the optimal strategy despite the fact that if both prisoners were to cooperate they would both have lighter sentences.

The Rationale Behind North Korea Acquiring Nuclear Weapons:

The rationale behind North Korea’s nuclear programme is simple. Only by creating a nuclear arsenal will Kim Jong Un be assured that the U.S. will not invade and depose him or launch any nuclear strike against him.

This behavior can be explained using a game called Nuclear War. In this game we will make a few assumptions. First, that if a nuclear strike is launched at any one of the players a retaliatory strike will be ordered launched back. Second, that both players are rational in that both will pick the optimal strategy. If both players choose to launch their nuke then both players are destroyed; if either player chooses to nuke his opponent, given the surety of the retaliatory strike, both players are destroyed; if both players choose to not nuke the other the status quo remains.

U.S.        \       North Korea


Doesn’t Nuke


Nuclear Annihilation

Nuclear Annihilation

Doesn’t Nuke

Nuclear Annihilation

Status Quo

Once again, to find the optimal strategy for both players let us consider the Nash Equilibrium. In this game, the Nash Equilibrium is in the case that neither player chooses to fire his missiles. As seen in the table above, a player’s deciding to preemptively fire missiles dooms both to nuclear annihilation. Since firing first causes one’s own destruction, it’s never advantageous to fire first. Consequently, the winning strategy for both players is to hold off and allow the favorable outcome of the status quo.

A nuclear armed North Korea is rightly denounced as a danger to the West. Their government may sell nuclear material to hostile actors (e.g: terrorists) and their successful acquisition of atomic bombs in the face sanctions may embolden other rogue powers to develop nuclear weaponry as well. Nevertheless, the threat of nuclear war is blown out of proportion in the public discourse. As shown above, both rational players will choose to maintain the status quo.

Why ICBMs are better for peace than anti-missile technology:

Given the development of the THAAD and Patriot anti-missile systems it is natural to ask whether these systems increase the chances of peace or whether both sides increasing their nuclear stockpiles would prove better. Common sense would dictate that defensive weaponry should help the cause of peace while the construction of ICBMs should cause the opposite; however, game theory can show that the reverse is actually true. Much like how a knight’s armour gives him an advantage over an armed peasant in allowing him to attack the peasant with impunity (the armour protects the knight from the peasant’s sword); defensive weaponry allows the nation possessing it to attack countries that lack it without any fear of consequences. By taking the danger out of war, defensive weaponry makes war a much more palatable, and in effect, probable option.

Let us now recreate the Nuclear War game shown above with the additional caveat that THAAD is capable of intercepting all North Korean missiles, meaning that the U.S. has a successful nuclear defensive shield. The game would now appear as so:

U.S.        \       North Korea


Doesn’t Nuke


North Korea Annihilated

North Korea Annihilated

Doesn’t Nuke

North Korea Annihilated

Status Quo

Where in the original game both parties would be annihilated in the case either party launched their missiles, in the modified game only North Korea would be destroyed. In this game there are two Nash equilibria, both being the case that U.S. chooses to nuke North Korea. The optimal strategy would then be to destabilize the peninsula and launch a first strike. If the U.S. were to only build additional ICBMs, the original balance of power created by the original game would remain unchanged. It is then clear that developing anti-missile technology is more destabilizing than increasing one’s nuclear stockpiles.


The threat of nuclear war is much exaggerated despite the recent spike in North Korean-American tensions. Using Game Theory, it is possible to show that given that the leaders of the U.S. and North Korea are rational, both nations will inevitably choose to de-escalate rather than cause a devastating nuclear war. In addition, it is possible to demonstrate that an arms race is safer than the development of anti-missile technology. If you’ve been staying up at night because the North Korean threat you should get some rest.


  1. Nuclear Weapons and Proliferation: A Cheat Sheet. (n.d.). Retrieved October 15, 2017, from
  2. Manea, M. (2016). Game Theory. Retrieved October 15, 2017, from
  3. Game Theory and Nuclear Weapons. (n.d.). Retrieved October 14, 2017, from