◼ POLITICS Bloomberg Businessweek August 5, 201935DATA:
PREDICTIT,
AS
OF
JULY
31
THEBOTTOMLINE Predictionmarketsrunonmuchofthesame
informationaspollsandmediareports,sodon’tstopbelievingwhat
youseeandread.26¢0Latest
yes
priceJoe
BidenKamala
HarrisPete
ButtigiegElizabeth
WarrenAndrew
YangTulsi
Gabbard
Cory
BookerBernie
Sanders1.6m750k90-day
trading
volume● PredictIt’s odds
on “Who will win the
2020 Democratic
nomination?”PredictIt’smarketsareroutinelycitedbythepress
asevidenceofwhomthesmartmoneyfavors.
Bettingonpoliticsis illegalintheU.S.PredictIt
owesitsexistencetoanexemptionforacademic
researchgrantedfiveyearsagobytheCommodity
FuturesTradingCommission(CFTC)toVictoria
UniversityofWellingtoninNewZealand.Togetper-
missiontooperatethesite,theschoolsaidprofes-
sorswouldrunit forfreeandthedatawouldbeused
inacademicstudies.
PredictItis similartoa stockmarket:Gamblers
postoffersto buyand sell “shares”incandi-
datesatpricesupto$1.If youwantedtobetthat
MassachusettsSenatorElizabethWarrenwould
betheDemocrats’ 2020 nomineeafterhersecond
debateonJuly30, you could have bought a share the
next day for 21¢. If she wins, you’ll get $1. If she loses,
her stock goes to $0, and you lose your money. The
site collects a 10% fee on winnings and a 5% fee on
withdrawals. Individual bets are limited to $850, per
the agreement with the CFTC. But with so much pol-
itics to gamble on, that hasn’t put much of a damper
on the action. Political day traders try to capitalize
on swings in opinion long before votes get cast.
Aaron Reese, a 34-year-old MBA student at Rice
University, says he often makes enough money
on PredictIt to pay his rent. He says Trump’s sur-
prise victory created a group of bettors who don’t
trust polls, which means easy money for him.
“People don’t believe reality anymore,” Reese says.
“People think every poll is fake and nothing mat-
ters anymore.”
Victoria University owns a small percentage of
PredictIt and collects a monthly administration fee,
according to Anne Barnett, who runs a program at
the school that helps professors turn research proj-
ects into businesses. The site’s operations have
largely been farmed out to a political software com-
pany called Aristotle Inc. John Aristotle Phillips,
its founder, says the school was already running a
similar site and he helped it expand into the U.S.
But don’t call it gambling, he says. He prefers the
term “forecasting.” “It’s an antidote to fake news,”
Phillips says. “If a trader makes a buying or selling
decision on something that’s inaccurate or mislead-
ing, they’re going to lose their money.”
A century ago, bookies openly took illegal polit-
ical bets at New York City’s stock exchanges and
hotels. With no public polls to report on, newspa-
pers wrote about big bets placed on candidates.
Election wagering fell out of favor once Gallup
introduced a poll that accurately predicted Franklin
Delano Roosevelt’s 1936 presidential victory and
New York legalized betting on horse races, accord-
ing to economists Paul Rhode and Koleman Strumpf.Chances are that at some point this election
season someone will tell you that prediction mar-
kets are a better gauge than polls. They developed
thisreputationafterbettorsona similarsitecalled
InTradecorrectlycalledthepresidentialelectionin
49 of 50 states in 2012 and 47 states in 2008. But one
2012 analysis showed that the prediction market had
done no better than pundits who relied on polling
averages—unsurprising, given most traders lean on
those indicators, too. InTrade was based in Ireland
and took bets from Americans until it became too
well-knownduringthe 2012 electionandU.S.reg-
ulatorssuedit.Thesiteclosedthenextyear.The
IowaElectronicMarkets,anotherpolitical-gambling
site, got an exemption from antigambling laws in
1993, but its website is outdated and difficult to use.
It offers just a couple of events to bet on and only
allows gamblers to deposit $500.
To justify the academic exemption that allows
it to ignore gambling laws, PredictIt provides its
data at no cost to researchers. Phillips says they
study the data “to understand why markets are as
predictive as they are.” A spokesman for Aristotle
declined to say how much the company has made
from PredictIt, but he said that the site has yet to
earn back its startup costs.
When Bloomberg Businessweek contacted doz-
ens of the people listed on PredictIt’s site as “our
research partners,” none said they had done any
work of the sort Phillips described, though a hand-
ful had made use of the odds as a control variable
in other studies. Several were surprised to see
their names listed. “That is a bit misleading,” says
Paul Armstrong-Taylor, an economics professor at
the Hopkins-Nanjing Center in China, who says he
asked for the data on behalf of a student who never
usedit.TheAristotlespokesmansaidArmstrong-
Taylor’snamewouldberemovedandprovideda
listof 15 papers and presentations that made use
of PredictIt data over the past few years. (PredictIt
says all the researchers listed signed an agreement
to receive data.)
Reese, the trader, says the best bet on PredictIt
has been selling shares of Andrew Yang, the long-
shot candidate who’s an internet favorite. His fans
are driving up his stock price to show their support—
up to 8¢, double U.S. Senator Cory Booker’s—and
there’s not enough smart money to balance them
out. “When prices start moving, people will assume
other people know something they don’t,” he says.
“You have to figure out what you think the real prob-
ability is.” �Zeke Faux