It is fair to say that London Labour’s activists and wonkish elite are by and large behind Ed Miliband, unless they are ideological Blairites (like the Progress crowd) or careerist greasy pole climbers like most MPs, in which case they are behind David Miliband. The headbanging Tory hating activists go for Balls and lefties back Abbott with Burnham picking up a Northern token vote.

Supporters of Ed Miliband are adamant, despite polls showing David has more than twice as much support among Labour supporters, that he will win on second preferences.  Will Straw at LeftFootForward has even built a predictive model that forecasts Ed scraping through. Yet punters persist in making David Miliband the runaway favourite. Are the punters or the wonks right?

Guido accepts that Ed will probably win the union’s endorsements, though the Fabian’s Sunda Katawala argues that doesn’t necessarily mean he will win the votes of union members. Punters agree with Will and give Ed a 60% chance of winning the union votes.

Guido and punters make David the 80% favourite to win the MP/MEPs vote.  Will forecasts it will be closer than that but accepts that David will win this part of the electoral college.

It is over the membership vote that Will’s model and punters differ dramatically. Based on a non-representative, self-selecting poll of LabourList’s readers (DM 34.9% EM 30.8%), he predicts that the second preferences from the more left-wing candidates will switch to Ed and he will thus beat his older brother. The argument being that since Ed is positioned to the left of David, second preferences won’t tack right to the most centrist candidate. This is delusional.

Not all the voters will see the candidates in such finely calibrated positions on the centre-to-left spectrum, much of the electorate will vote on character and personality. Apart from Diane Abbott the policy positions of the candidates are in reality very closely bunched and Ed Balls’ new found tactical leftism is transparently risible.  Experience shows that second preferences tend to break in correlation with first preferences. The correlation isn’t perfect, but nor is it so weak as to be insignificant. Yet Labour sympathisers in the media and even more objective pundits like Toby Young believe Ed will come through.

So if it will be decided by the membership vote, is the LabourList poll accurate? It is unlikely that a self-selecting web poll will be. YouGov actually put Balls ahead of Ed Miliband but behind David with Labour voters (and another private poll by Survation put Diane Abbot within 5% of Ed among union members).  With the polling confusion Guido puts little reliance on the sampling and absent of clear polling evidence Guido opts to “follow the money”. David Miliband has raised more money than all the rest and he has the weight of  punter’s money backing him. It will be close, but Guido suspects Ed will lose to his big brother.