* Pollsters failed to predict 2015 election result
* Bookmakers made big losses while gamblers won
* Polls suggest EU vote close, bookmakers back UK to stay in
By Michael Holden
LONDON, Jan 21 (Reuters) - To predict the result ofBritain's upcoming referendum on European Union membership youmight do well to ignore the opinion polls and watch what thegamblers are doing instead.
The fact that pollsters failed to call last year's nationalelection winner while gamblers won big has given an additionaltwist to the knife-edge EU vote.
Political betting has become a serious business in recentyears for Britain's 6.3-billion-pound ($8.9-billion) gamblingindustry, and about 100 million pounds was estimated to havebeen staked on the outcome of last year's national election.
For months up until the very last moment, polls indicatedthat the May election would result in a hung parliament with thetwo major parties securing a similar number of seats.
Instead, Prime Minister David Cameron's Conservative partysecured victory with a comfortable, albeit slim, majority thepsephologists who study voting trends had failed to predict.
It was not just pollsters who got it wrong. Bookmakersseriously lost out as their punters correctly called the result.
"We lost an awful lot of money on the general election lastyear and that was because we relied too much on the polls,"Matthew Shaddick, Head of Political Odds at bookmaker Ladbrokes, told Reuters.
"We probably weren't taking enough notice of what peoplewere betting on. Lots of people were quite happy to ignore thepolls last year and bet on the Tory (Conservative) majoritywhich looked all but impossible to people who pay a lot ofattention to the polls."
A report published on Tuesday concluded that samplerecruitment methods and possible unintended herd behaviour bypollsters had led to last year's fiasco which angered many inpolitics and the media.
Only an exit poll that jolted the political establishment onelection night itself was close to being right.
The next major test for the polling companies will be thereferendum on whether Britain should remain in the EuropeanUnion, a vote to be held before the end of 2017.
Already there is a clear divergence between opinion pollsand the betting industry. A current poll of polls has "Remain"on 51 percent and "Leave" on 49 percent while the most recentsurvey put the number of Britons wishing to leave the bloc at 53percent, with 47 percent in favour of remaining.
However, bookmakers are unanimous in making the Remain campfirm favourites with nearly all offering the same odds, 2/5.
"It's almost exactly a two-thirds chance according to ourclients that Britain will choose to stay in Europe," PeterHetherington, chief executive of online brokerage IG Group, toldReuters. "That is interesting because that's actually quitediverging from the latest opinion polls which suggest it's muchcloser than that."
CASH DRIVES THE ODDS
So why the discrepancy?
"I think more and more people are getting more convincedthat that's the way it's going to go irrespective of the factthe polls are showing it very, very close on average overall,"said Ladbrokes' Shaddick.
"We spend a lot of our time looking at polls and thepolitical science analysis of those polls. But ultimately it'sthe money we're taking from our customers that drives the oddsmore than our particular opinion."
So far, not much money has been staked on the referendumoutcome - Ladbrokes says it has taken about 100,000 pounds -although the bookmakers expect that to change once Cameron hasannounced a date for the vote.
They expect the interest will then match that shown bygamblers during the 2014 Scottish independence vote when Scotsrejected a breakaway, an outcome pollsters correctly forecast. Up to 12 million pounds was said to have been bet on the vote.
Graham Sharpe, from bookmaker William Hill, said itwas no surprise pollsters got it wrong sometimes.
"If the opinion pollsters were always going to be right,we'd have been out of business long ago," he told Reuters.
"We update the odds on a regular basis depending a. on whatwe think, b. what money we've taken, and c. as a factor we lookat the opinion polls because we are aware that some of ourcustomers may be influenced by them as well. But they'recertainly not the main factor."
The fact William Hill had lost about 500,000 pounds at thelast election showed their customers did not believe the polls,he said. However, he pointed out that anyone predicting elections can get it wrong.
An unidentified man who won 193,000 pounds on a900,000-pound bet on the result of the Scottish independencevote then lost 205,000 pounds wrongly predicting the 2015election would see a hung parliament, William Hill said.
In the wake of their disastrous showing last year, pollstershave now promised to make changes, although the head of theinquiry into the failures said there was no silver bullet.
Shaddick said his team would be more sceptical but surveysstill remained the best way of gauging public opinion.
Sharpe was less convinced. "It's always been my policy todeliberately lie to any opinion pollsters so how you factor thatinto the equation when they're speaking to people I really don'tknow," Sharpe said.
($1 = 0.7058 pounds) (Additional reporting by Noor Zainab Hussain; Editing by GuyFaulconbridge)