« A look back at 2007 and Prediction Markets
How to interpret prediction market results on elections »

Presidential Decision Markets

January 15th, 2008

About a week and a half ago, Robin Hanson announced on Overcoming Bias and on the Prediction Markets Google Group that InTrade is now hosting Presidential Decision Markets. Users can now trade on what will happen to the US Debt and the number of troops in Iraq if either a Democrat or non-Democrat wins the upcoming election for the presidency. Additionally, users can also trade on what will happen to oil futures and T-bond interest rates on the day of the election.

Fortunately, Peter McCluskey has funded a market-maker for these contracts, which gives them liquidity that hasn’t necessarily been available on other specialised InTrade markets. I think this is a great model for future contracts that are a little theoretical for average traders. The results have already proven to be interesting!

Shock Response Futures

As expected, there has been very little trading on the contracts that compare the movements of T-bond interest rates and oil prices on election day. Both markets will likely develop significantly in the coming months, so trading now means committing funds with very little information. I’m not sure what will happen to these markets even close to election day. If the race is competitive until the very end (like it was in 2000 and 2004) than these could be interesting markets. But if the winner of the election is largely expected in the days and weeks before election day (like 1984), the ramifications of that decision will be priced into the various markets. Then the results will be based on the very short-term daily price movement on election day.

Currently only a total of 32 contracts have being traded between these two securities, and I don’t expect many more for quite some time.

Conditional Estimates

There has been significantly more interest on the contracts that compare what will happen to the US Debt and Iraq Troop Levels if a Democrat or non-Democrat is elected to the presidency.

As I write this post, these are the current last traded prices (volume in brackets):

DEM.PRES-GOVT.DEBT = 52.4 (141)
NONDEM.PRES-GOVT.DEBT = 39.5 (586)
DEM.PRES-TROOPS.IRAQ = 49.9 (62)
NONDEM.PRES-TROOPS.IRAQ = 37.5 (625)

PRESIDENT.DEM2008 = 62.0
PRESIDENT.REP2008 + PRESIDENT.FIELD2008 = 38.0

which yield:

Democrat & Debt = 84.5
Non-Democrat & Debt = 104.0

Democrat & Troop Levels = 80.5
Non-Democrat & Troop Levels = 98.7

What can we learn from these (early) results?

  • There is clearly an interest in these markets, despite the fact they will not be settled for at least three years. I highly doubt this would have been possible without Peter’s market-maker and InTrade’s suspension of trading fees.
  • There is a clear early difference between Democrats and Republicans. The results show that a Democratic President will result in less federal debt and fewer troops in Iraq by the middle of their first term in office.
  • Contracts on the Non-Democrat side are traded much more than the Democratic contracts.

Summary

There is clearly quite some time before the election, and these prices should still move around quite a bit. But it’s quite encouraging to see that with a subsidised market-maker (and no fees!) that the results show a clear difference between the two parties. I don’t expect that this will make it into the general consciousness of the popular press anytime soon, but is encouraging news.

This entry was posted on Tuesday, January 15th, 2008 at 4:45 pm and is filed under General, Real-money Markets.

  • Yes, it is encouraging to see some results already. But I think the shock response asset values should depend only moderately on how close is the election. And whether those values are above or below 50 should not depend at all.
  • I may be completely off-base here, but I guess I see the oil price movement on election day as not terribly related to the presidential election if the election results are largely expected, such as a huge blowout for one party. (If the election becomes a strong contest, then absolutely I think the oil prices will respond to estimated election returns.)

    Then again, I don't track the volatility on oil markets, so the day-to-day news may have a bigger effect than I estimate, even when largely known events occur.
  • It's not obvious that the prices show effects of the election on federal debt and Iraqi troop levels yet.
    A simple explanation for what those markets show is that traders have not yet devoted enough money to reflect the relative chances of a Republican versus a Democrat. The market maker started with all prices centered on 50, and it takes a fair amount of money for traders to move the market maker's prices enough to reflect the fact the difference between PRESIDENT.DEM2008 and PRESIDENT.REP2008 + PRESIDENT.FIELD2008. A large fraction of the trades in NONDEM.PRES-GOVT.DEBT and NONDEM.PRES-TROOPS.IRAQ fit a pattern of driving the price down.
    Before I start claiming that the prices have become useful to voters, I want to see something like 30% of trades be something other than trades that continue to push the market maker in the same direction. And ideally we should see some correlation with PRESIDENT.DEM2008 price changes before we're confident that the prices have become valuable.
    In case it isn't clear to some readers, the market maker is software that knows nothing about elections. I set its initial prices to be near 50 because that was the simplest way to implement it, and to the extent that that was close to a sensible price, it was because the contracts are designed in hopes that they will trade close enough to 50 that asymmetries in the money needed for each side of the market aren't too large.
    In hindsight, maybe I should have set the initial market maker prices to reflect the information provided by PRESIDENT.DEM2008.
  • Hello, Peter.

    I see your point that these markets are still in development, and once prices have stabilised they may show less of a difference between the two parties. And I think what you mentioned about setting the initial prices to reflect the PRESIDENT.DEM2008 prices would probably be useful in the future.

    That said, I think there's still some interesting data here, and am perhaps most encouraged that people are actually trading on these markets! So many times they are set up with the best of intentions, but in the end are so illiquid as to provide no information at all.

    Do you happen to have a script or a way of doing these calculations and showing the results above automatically? I'd be interested in tracking this, but don't necessarily want to take each days data and calculate it manually, and I am not a scripting-saavy guy.
  • I've created web pages at http://www.bayesianinvestor.com/amm/implied.html and http://www.bayesianinvestor.com/amm/implied4.html (which are currently being updated 4 times a day) which show implied prices for the conditional contracts that I'm subsidizing and for the contracts sponsored by Politimetrics.
blog comments powered by Disqus
Clicky Web Analytics