I will say it again. There really is nothing quite like a British by-election! The speculation and rumours are made for betting drama and the experts are frequently wrong.
Remember Chesham and Amersham, where the Tories were 1.11/10 before losing by a 20% margin? Or Batley and Spen, where they were 1.251/4 before Labour held the seat? Throughout the years, we have come to expect such betting chaos.
What's the explanation? Polling firms rarely survey these contests, leaving us reliant on party propaganda and rumours. The dynamics are completely different to general elections. Voters are happier to register a mid-term protest. Turnout is often low and motivation highly differential.
With those caveats in mind, let's try to navigate North Shropshire. A race that has already seen plenty of betting drama.
Lib Dem odds-on after monster gamble
A month ago, I earmarked a potential upset for the Lib Dems at 4.3100/30. 48 hours out, they are red-hot favourites at 1.594/7 to pull off an upset that could seriously destabilise Boris Johnson.
The obvious explanation is Johnson's implosion. It started with the corruption scandal that triggered the resignation of Owen Paterson and this by-election. It worsened with the PM's 'Peppa Pig speech' to the CBI and then 'PartyGate' pushed the Tories into full-blown crisis.
National polls have swung against the Tories, with Labour now leading by up to 9%, across the range of firms. As discussed last week, the odds about an early Johnson exit have crashed.
All of which builds a strong argument that Tory voters in this rural constituency will either be prepared to defect to the Lib Dems - past masters of capitalising in such mid-term scenarios - or stay at home.
Huge turnaround required on 2019
This, however, is surely factored into the transformed betting and it would be wise at this juncture to remind ourselves of the scale of the Lib Dems' task. At the 2019 General Election, they finished third on a meagre 10%.
Paterson was re-elected with 62.7%. In 2017, he won 60.5%. He held the seat in all three Labour victories from 1997-2005, never falling below 40%. The highest Lib Dem share was 20.9% in 2010, and they finished below Labour in all the rest.
Looking back further, the race for second was frequently close. The Lib Dems were second during the Thatcher/Major years, peaking with 31.6% in 1983.
Lib/Lab rivalry could thwart tactical voting
That sideshow may be more relevant to Thursday's outcome than first appears. The theory behind a Lib Dem upset here involves the Labour vote collapsing in their favour, as occurred in Chesham and Amersham. Using the sort of dubious spin for which their constituency campaigns are famous, the Lib Dems released these internal numbers that seemed to confirm that tactical trend.
Labour, however, hit back with their own internal polling, telling a very different story.
Only yesterday, they were very clearly not standing aside or encouraging voters to lend theirs to the Lib Dems.
Who to believe? Betfair markets make the Lib Dems odds-on, Labour a 150/1 chance. No doubts there, but can we trust it? As explained above, by-election betting signals have a dismal record.
The likeliest explanation is that two local parties with a long rivalry are refusing to stand aside for the other, and pushing whatever spin they can.
Reform likely to advance at Tory expense
We also should consider the effect of Reform UK. This latest hard-right incarnation of Richard Tice has made opposing lockdown measures its central theme. Without the 'Brexit Party' label, they are predictably failing to get the same cut-through but are polling up to 7% nationally, mostly at Tory expense.
They won 6.6% in Old Bexley and Sidcup and are entitled - given how the Tories have sunk since and the Omnicrom variant, vaccine passports and other anti-Covid measures have taken centre stage - to expect improvement. Their supporters are bound to be motivated and, as with the Lib Dems, are a relatively risk-free recipient for protest voters who don't like Labour.
35% could be enough to win
The upshot is that this seems unlikely to be another Chesham and Amersham, where non-Tories united around the Lib Dems. I suspect the vote will split substantially four ways, making 40% a winning total. Perhaps even, as in Batley and Spen, 35%.
In that scenario the Tories are very much in this. In our Conservative Vote Share market, 35-39.99% is favourite at 2.56/4. That would mean they'd lost over a third of their 2019 share (compared to a quarter on national figures) and set a competitive target. Any lower would represent a catastrophe.
One would expect their primarily elderly contingent to have sent in postal votes early. Whereas the Lib Dems are likely to rely more on the campaign, and voting on the day. The latest worrying Covid scare could further dampen turnout.
Ultimately, I think the Lib Dems will find it very hard to unite all the anti-Tory factions and reach the high thirties, given it appears they are all campaigning. Laying them at odds-on appeals, keeping all other options onside. Anything could happen. I'm not entirely ruling Labour out at 150/1.
Follow Paul on Twitter and check out his website, Political Gambler.