Category Archives: Prediction Markets

Health Care Reform, Politics, Prediction Markets

Prediction Markets' take on the 2010 United States Senate special election in Massachusetts

Leave a comment

Intrade.com maintains an actively traded market for futures contracts which pay 100 points (where 1 point = $.10) in the event that a specific contingent event occurs and 0 points otherwise. Thus, prices represent “risk neutral” event probabilities. I have previously blogged about how useful prediction markets can be in assessing political events such as election outcomes (e.g., see Prediction markets assessment of the (2004) Presidential Election and Preliminary assessment of the accuracy of the Intrade State-by-State contracts (for the 2008 Presidential Election)); furthermore, an article entitled “Prediction Markets“ by Justin Wolfers and Eric Zitzewitz, Journal of Economic Perspectives (Vol. 18, No. 2 (Spring 2004), pp. 107-126) provides a particularly informative technical survey concerning prediction markets in general.

Lately, the upcoming (January 19) United States Senate special election in Massachusetts has become particularly interesting because up until the past couple of weeks, the Democratic candidate (Martha Coakley) was heavily favored to defeat the Republican candidate (Scott Brown).  Below, I list the time series graphs for the MA.SPEC.SENATE.DEM and the MA.SPEC.SENATE.REP contracts that are currently trading on intrade.com.  The MA.SPEC.SENATE.DEM contract is a (100,0) bet on Martha Coakley winning this special election, whereas the MA.SPEC.SENATE.REP contract is a (100,0) bet on Scott Brown prevailing.  Both contracts were initially offered on this exchange in September 2009, but only since the beginning of 2010 has there been much interest in trading these contracts.  


MA.SPEC.SENATE.DEM

 
MA.SPEC.SENATE.REP

Lately there has been strong momentum in Brown’s favor, and as I write this, the MA.SPEC.SENATE.REP contract is up 13 points from its previous (1/15) close of 41, whereas the MA.SPEC.SENATE.DEM contract is down 12.5 points from its previous close of 61 (note that that the contract prices can sum to slightly more or less than 100 because trades are non-synchronous).  The price change since earlier this week has been quite dramatic, going from a roughly 80/20 advantage for Coakley to the present situation which indicates a slight advantage probabilistically for Brown.  Anyway, it will be obviously very interesting to see how this election plays out!

]]>

Health Care Reform, Politics, Prediction Markets

Prediction Markets’ take on the 2010 United States Senate special election in Massachusetts

Leave a comment

Intrade.com maintains an actively traded market for futures contracts which pay 100 points (where 1 point = $.10) in the event that a specific contingent event occurs and 0 points otherwise. Thus, prices represent “risk neutral” event probabilities. I have previously blogged about how useful prediction markets can be in assessing political events such as election outcomes (e.g., see Prediction markets assessment of the (2004) Presidential Election and Preliminary assessment of the accuracy of the Intrade State-by-State contracts (for the 2008 Presidential Election)); furthermore, an article entitled “Prediction Markets“ by Justin Wolfers and Eric Zitzewitz, Journal of Economic Perspectives (Vol. 18, No. 2 (Spring 2004), pp. 107-126) provides a particularly informative technical survey concerning prediction markets in general.

Lately, the upcoming (January 19) United States Senate special election in Massachusetts has become particularly interesting because up until the past couple of weeks, the Democratic candidate (Martha Coakley) was heavily favored to defeat the Republican candidate (Scott Brown).  Below, I list the time series graphs for the MA.SPEC.SENATE.DEM and the MA.SPEC.SENATE.REP contracts that are currently trading on intrade.com.  The MA.SPEC.SENATE.DEM contract is a (100,0) bet on Martha Coakley winning this special election, whereas the MA.SPEC.SENATE.REP contract is a (100,0) bet on Scott Brown prevailing.  Both contracts were initially offered on this exchange in September 2009, but only since the beginning of 2010 has there been much interest in trading these contracts.  


MA.SPEC.SENATE.DEM

 
MA.SPEC.SENATE.REP

Lately there has been strong momentum in Brown’s favor, and as I write this, the MA.SPEC.SENATE.REP contract is up 13 points from its previous (1/15) close of 41, whereas the MA.SPEC.SENATE.DEM contract is down 12.5 points from its previous close of 61 (note that that the contract prices can sum to slightly more or less than 100 because trades are non-synchronous).  The price change since earlier this week has been quite dramatic, going from a roughly 80/20 advantage for Coakley to the present situation which indicates a slight advantage probabilistically for Brown.  Anyway, it will be obviously very interesting to see how this election plays out!

Economics, Health Care Reform, Prediction Markets, Risk and Uncertainty

What are the odds that the Public Health Option will be passed into law by the end of this year?

Leave a comment

Intrade.com maintains an actively traded market for futures contracts which pay 100 points (where 1 point = $.10) in the event that a specific contingent event occurs and 0 points otherwise. Thus, prices represent “risk neutral” event probabilities. I have written previously about how useful prediction markets can be in assessing political events such as election outcomes (e.g., see Prediction markets assessment of the Presidential Election from October 26, 2004 and Preliminary assessment of the accuracy of the Intrade State-by-State contracts from November 5, 2008) and Supreme Court confirmations (e.g., see SC.CONFIRM.ALITO from November 6, 2005), as well as other kinds of contingent events such as the state of the economy (e.g., see What are the prediction markets saying about the economy? from April 10, 2009), etc.

Currently (as of August 19, 2009), the Intrade market gives the Public Health Option a 36% chance of being passed into law by the end of this year (the closing price yesterday, on August 18, was 32 Intrade points).  The contract is called US.GOVT.HEALTHPLAN.DEC09, and it has been trading ever since June 12, 2009. The highest price recorded for this contract was 51.5 Intrade points (at the contract’s inception on June 12, 2009), and the lowest price recorded was 14.1 Intrade points (on August 17, 2009, after it appeared that the Obama administration might be willing to compromise on the public option in favor of non-profit health cooperatives).  Here’s the complete time series since contract inception on June 12:

For more information concerning the topic of “prediction markets”, I recommend an article entitled “Prediction Markets“ by Justin Wolfers and Eric Zitzewitz that appeared a few years ago in Journal of Economic Perspectives (Vol. 18, No. 2 (Spring 2004), pp. 107-126).

Prediction Markets

What are the prediction markets saying about the economy?

Leave a comment

I have had some interest during the past few years in studying the accuracy of so-called “prediction markets” such as Intrade.com. Markets such as Intrade.com offer bets on the outcomes of any number of future events, including politics, the economy, climate change, current events, entertainment, scientific discoveries, etc. The typical setup is that a contract pertaining to some type of uncertain future event is created which pays $10 if a certain outcome occurs (e.g., whether Obama wins the presidential election, or whether the U.S. economy falls into a depression within a given period of time), and $0 if this outcome does not occur. These markets first gained some notoriety back in the spring of 2003 in the build-up to the U. S. invasion of Iraq when so-called “Saddam Futures contracts” (which placed real money bets concerning whether Saddam Hussein would remain in power) became actively traded. For an assessment concerning prediction market accuracy pertaining to the 2008 election, see my “day after” (November 5) blog entry entitled “Preliminary assessment of the accuracy of the Intrade State-by-State contracts”.

Prices are quoted on Intrade.com in the form of Intrade “points” where 1 Intrade point = $0.10. This way, the market prices resemble probabilities (technically, my inner economist tells me that these are actually “Arrow-Debreu” prices), but for practical purposes, it is quite reasonable to interpret these prices as probabilities.

Anyway, now that I have explained prediction markets in a nutshell, I would like to call attention to the latest (email) summary, provided by Intrade.com’s CEO, Mr. John Delaney, concerning what prediction markets are indicating may lie ahead for the economy (what follows is a direct quote from an email that I received earlier today from Mr. Delaney):

______________________________________

“Gross Domestic Product:

GDP fell by 6.3% in the final quarter of 2008, the single worst drop in quarter of a century. The advance estimate of GDP for the first quarter of 2009 is due for release on April 29th, and the Intrade market on what that figure will be points to an easing in the contraction. There is an 83.0% chance GDP will contract by 4.0% or more but only a 25.0% chance the contraction will be 5.5% or more.

Our markets also indicate that talk of a depression appears unfounded. Although definitions vary, our market (see market specific definition) for a 10.0% decline in GDP from its peak value (currently Q3 of 2008) indicates only a 12.0% chance of this happening before the end of 2009. We have other markets listed here which give a similarly low probability of a depression.

Is there any light at the end of the tunnel? Our markets indicate that things could turn around by the end of 2009. They show a 50.0% probability that GDP growth will be positive in Q4 of 2009, and a 73.0% probability of positive growth in Q1 of 2010.

Unemployment:

The month of March saw the unemployment rate in the United States rise from 8.1% to 8.5%. How bad will it get? The president of the Federal Reserve Bank of Dallas and many others believe that unemployment could surpass 10.0% by the end of the year.

Do the Intrade markets agree? A glance at the figures indicates that an unemployment rate of 10.0% is a very real possibility. The market currently shows a 61.0% probability of the December 2009 unemployment rate topping 10.0%. If you think it will top 10.0% and it does, you would make a $39 profit for every $61 you invested in your opinion. It also shows a 45.0% chance the rate will be 10.5% or more and a 35.0% chance it will be 11.0% or more.

Automakers and Banks:

President Obama recently identified GM and Chrysler as candidates for bankruptcy unless they could quickly and successfully restructure.

Chrysler has been given until the end of April to sort out a partnership with Italian automaker Fiat. The market is uncertain if this can be managed, giving a 50.1% chance of a partnership agreement being successfully concluded by April 30th. The market is also still bullish on Chrysler having to file for bankruptcy by the end of 2009, giving a 79.9% probability of this happening.

The market clearly doubts the ability of GM to successfully manage a turnaround, giving GM a 74.3% chance of having to file for bankruptcy before the end of the year.

In contrast to the struggles of GM and Chrysler, the other member of the big-three, Ford, has fared better. Ford recently shed $9.9 billion in debt, which encouraged a 13.0% jump in their share price. It also reduced their probability of bankruptcy to 16.0%.

The President also forced GM CEO Rick Wagoner to stand down as part of the administration’s plans to restructure the company. This raises the question, will the President also force out those in charge of other companies that have received federal funds? Such as those heading up the large banks who received TARP funds? A market is available on whether the CEO’s of JP Morgan, Bank of America, AIG, Citigroup and Wells and Fargo will be forced out in a manner similar to Wagoner.”

Prediction Markets

Prediction markets, new year edition

1 Comment

For what it’s worth, here’s what the intrade markets are predicting for 2009 (as of 7am ET on 12/29/2008):

  1. GM to announce a merger with another major auto manufacturer: 35%
  2. More than US$25 billion to be injected into the big 3 auto-makers: 60%
  3. Caroline Kennedy to replace Hillary Clinton in the US Senate: 53%
  4. Guantanamo Bay Detention Camp to be closed in 2009: 84%
  5. The US in Recession in 2009: 85%
  6. An air strike against Iran before end of 2009: 21%
  7. US unemployment rate at or above 8% in December 2009: 50%
  8. Robert Mugabe to depart as President of Zimbabwe in 2009: 50%
  9. Slumdog Millionaire to win Academy Award for Best Picture: 52%
Prediction Markets

Will the U.S. Senate have a 60 seat Democratic Majority?

Leave a comment

At this point, it would appear (on the basis of prediction market trading this morning) that the answer will likely be no. CNN.com currently lists 56 Democratic Senators, 40 Republican Senators, and 4 undecided Senate seats in the states of Alaska, Georgia, Minnesota, and Oregon. Here are the last trades this morning on Intrade.com pertaining to the four undecided races:

1. Alaska: AK.SENATE08.REP: 75.5, AK.SENATE08.DEM: 15
2. Georgia: GA.SENATE08.REP: 85, GA.SENATE08.DEM: 25
3. Minnesota: MN.SENATE08.REP: 90.1, MN.SENATE08.DEM: 8.5
4. Oregon: OR.SENATE08.REP: 75, OR.SENATE08.DEM: 33.1

Prediction Markets

Preliminary assessment of the accuracy of the Intrade State-by-State contracts

Leave a comment

Now that the election is over, it is interesting to compare the actual outcome with my naive “prediction” method which allocated Electoral College votes based upon a cutoff price point of 55 for the 102 state-by-state Intrade contracts. On the eve of the election (see “Prediction Markets Update: November 3, 2008“), the state-by-state contracts implied a 353-174 Obama Electoral College “lead” over McCain, with Missouri being the only swing state. This morning, according to the Wall Street Journal‘s Interactive Map, Obama has officially won 349 Electoral College votes compared with 163 Electoral College votes for McCain. The actual state-by-state allocation matches mine exactly with the following exceptions:

1. I had Indiana going for McCain; apparently Indiana’s 11 Electoral College votes will instead end up in Obama’s column. This is why McCain’s actual Electoral College vote total is 11 less than what I predicted on November 3. In Indiana, Obama ended up with a 22,986 vote (0.86%) margin over McCain out of 2,681,726 votes cast.

2. North Carolina has yet to be called (although this morning, with 100% of the precincts reporting, Obama currently appears to have a 12,160 vote (0.29%) margin out of 4,191,812 votes cast). If you reduce Obama’s November 3rd total of 353 Electoral College votes by 15 (for North Carolina) and add 11 (for Indiana), this gives you 349 Electoral College votes for Obama.

This morning, the NTH.CAROLINA.DEM and NTH.CAROLINA.REP contracts are trading for 96 and 2 respectively, so it is almost certain that North Carolina’s 15 Electoral College votes will end up in Obama’s column. Furthermore, the MISSOURI.REP and MISSOURI.DEM contracts last traded this morning at 96.5 and 3.2 respectively, so it is almost certain that Missouri’s 11 Electoral College votes will end up in McCain’s column. Therefore, I expect the final Electoral College totals for the 2008 Presidential election to be 364 Electoral College votes cast for Obama, compared with 174 Electoral College votes cast for McCain.

Addendum: November 5, 2008 Electoral College Vote allocation (based upon actual voting data reported by the Wall Street Journal):

Barack Obama (349): California (55), Colorado (9), Connecticut (7), Delaware (3), District of Columbia (3), Florida (27), Hawaii (4), Illinois (21), Indiana (11), Iowa (7), Maine (4), Maryland (10), Massachusetts (12), Michigan (17), Minnesota (10), Nevada (5), New Hampshire (4), New Jersey (15), New Mexico (5), New York (31), Ohio (20), Oregon (7), Pennsylvania (21), Rhode Island (4), Vermont (3), Virginia (13), Washington (11), and Wisconsin (10)

John McCain (163): Alabama (9), Alaska (3), Arizona (10), Arkansas (6), Georgia (15), Idaho (4), Kansas (6), Kentucky (8), Louisiana (9), Mississippi (6), Montana (3), Nebraska (5), North Dakota (3), Oklahoma (7), South Carolina (8), South Dakota (3), Tennessee (11), Texas (34), Utah (5), West Virginia (5), and Wyoming (3)

Not yet called for either candidate: Missouri (11) and North Carolina (15)

Prediction Markets

Prediction Markets Update: November 3, 2008

Leave a comment

The 2008.PRES.OBAMA Intrade contract now trades at 90.5, whereas the 2008.PRES.McCAIN Intrade contract is currently trading at 10 (compared with 83.3 and 17.3 respectively as as reported in my last update on October 30).

On this, the eve of the 2008 election, the state-by-state contracts imply that Mr. Obama holds a 353-174 “lead” over Mr. McCain (based upon my cutoff price point of 55 for allocating Electoral College votes). Since my last update on October 30, Missouri has moved out of the Obama column and is now a “swing” state (indeed, the last recorded trades for MISSOURI.DEM and MISSOURI.REP were at prices of 50.1 for both contracts).

FiveThirtyEight.com currently gives Mr. Obama a 340.2 to 197.8 advantage over Mr. McCain.

Addendum: November 3, 2008 Electoral College Vote allocation

Barack Obama (353): California (55), Colorado (9), Connecticut (7), Delaware (3), District of Columbia (3), Florida (27), Hawaii (4), Illinois (21), Iowa (7), Maine (4), Maryland (10), Massachusetts (12), Michigan (17), Minnesota (10), Nevada (5), New Hampshire (4), New Jersey (15), New Mexico (5), New York (31), North Carolina (15), Ohio (20), Oregon (7), Pennsylvania (21), Rhode Island (4), Vermont (3), Virginia (13), Washington (11), and Wisconsin (10)

John McCain (174): Alabama (9), Alaska (3), Arizona (10), Arkansas (6), Georgia (15), Idaho (4), Indiana (11), Kansas (6), Kentucky (8), Louisiana (9), Mississippi (6), Montana (3), Nebraska (5), North Dakota (3), Oklahoma (7), South Carolina (8), South Dakota (3), Tennessee (11), Texas (34), Utah (5), West Virginia (5), and Wyoming (3)

Prediction Markets

Prediction Markets Update: October 30, 2008

Leave a comment

The 2008.PRES.OBAMA Intrade contract now trades at 83.3, whereas the 2008.PRES.McCAIN Intrade contract is currently trading at 17.3 (compared with 87.2 and 13.3 respectively as as reported in my last update on October 28).

The state-by-state contracts continue to imply that Mr. Obama holds a 364-174 “lead” over Mr. McCain (based upon my cutoff price point of 55 for allocating Electoral College votes), with none of the 50 states or the District of Columbia falling into the “swing” state category.

FiveThirtyEight.com currently gives Mr. Obama a 344.1 to 193.9 advantage over Mr. McCain.

Addendum: October 30, 2008 Electoral College Vote allocation

Barack Obama (364): California (55), Colorado (9), Connecticut (7), Delaware (3), District of Columbia (3), Florida (27), Hawaii (4), Illinois (21), Iowa (7), Maine (4), Maryland (10), Massachusetts (12), Michigan (17), Minnesota (10), Missouri (11), Nevada (5), New Hampshire (4), New Jersey (15), New Mexico (5), New York (31), North Carolina (15), Ohio (20), Oregon (7), Pennsylvania (21), Rhode Island (4), Vermont (3), Virginia (13), Washington (11), and Wisconsin (10)

John McCain (174): Alabama (9), Alaska (3), Arizona (10), Arkansas (6), Georgia (15), Idaho (4), Indiana (11), Kansas (6), Kentucky (8), Louisiana (9), Mississippi (6), Montana (3), Nebraska (5), North Dakota (3), Oklahoma (7), South Carolina (8), South Dakota (3), Tennessee (11), Texas (34), Utah (5), West Virginia (5), and Wyoming (3)

Prediction Markets

Prediction Markets Update: October 28, 2008

Leave a comment

The 2008.PRES.OBAMA Intrade contract now trades at 87.2, whereas the 2008.PRES.McCAIN Intrade contract is currently trading at 13.3 (compared with 88 and 12.8 respectively as as reported in my last update on October 26).

The state-by-state contracts now imply that Mr. Obama holds a 364-174 “lead” over Mr. McCain (based upon my cutoff price point of 55 for allocating Electoral College votes), with none of the 50 states or the District of Columbia falling into the “swing” state category.  Since my last update, Indiana (11 Electoral College votes) has moved from Mr. Obama’s column into Mr. McCain’s column.

FiveThirtyEight.com currently gives Mr. Obama a 348.2 to 189.8 advantage over Mr. McCain.

Addendum: October 28, 2008 Electoral College Vote allocation

Barack Obama (364): California (55), Colorado (9), Connecticut (7), Delaware (3), District of Columbia (3), Florida (27), Hawaii (4), Illinois (21), Iowa (7), Maine (4), Maryland (10), Massachusetts (12), Michigan (17), Minnesota (10), Missouri (11), Nevada (5), New Hampshire (4), New Jersey (15), New Mexico (5), New York (31), North Carolina (15), Ohio (20), Oregon (7), Pennsylvania (21), Rhode Island (4), Vermont (3), Virginia (13), Washington (11), and Wisconsin (10)

John McCain (174): Alabama (9), Alaska (3), Arizona (10), Arkansas (6), Georgia (15), Idaho (4), Indiana (11), Kansas (6), Kentucky (8), Louisiana (9), Mississippi (6), Montana (3), Nebraska (5), North Dakota (3), Oklahoma (7), South Carolina (8), South Dakota (3), Tennessee (11), Texas (34), Utah (5), West Virginia (5), and Wyoming (3)