Sterling was hit by a freak sell-off overnight as the currency crashed by six per cent seconds after the Asian markets opened for trading.
The pound fell from $1.26 to $1.18 almost instantly, before recovering slightly to stand at $1.24 against the dollar. It was the same story against the euro, where the pound sank to the eye-watering low of €1.1031 at around 12.30am this morning.
Societe Generale's Kit Juckes explained what happened:
Sterling was drunk yesterday, went completely mad late in the evening and probably has a terrible headache this morning.
But why did it happen? Currencies don't just drop six per cent without somebody doing something and a lot of money switching hands.
It was late at night in Europe. It was early in the morning in Asia.
Tiredness makes fingers fatter, traders lazier and mistakes more likely. The idea is simple really: Say, you want to buy £10m of sterling at $1.26. You punch on the buy order, and hit confirm. But, you actually ended up placing an order for £10bn at $1.16.
A multi-billion pound typo.
The fat finger theory is the go-to of the "I don't know what on earth just happened it must have been a mistake" school of thought. In part, it's because they do happen abd when there's no news it makes sense.
Read more: Is the UK economy now smaller than France?
In 2001 somebody tried to by £8bn of shares in tech company Autonomy in a fat-finger trade. It was cancelled, but it would have equalled four-times the firm's market capitalisation.
In 2002, Bear Stearns knocked 100 points off the Dow Jones stock index by selling $4bn of stock instead of $4m in a finger slip.
Last year a banker at Deutsche Bank transferred $6bn to a US hedge fund when he confused gross and net in his calculations.
The other reason the fat finger theory gains traction is because its a neat analysis that traders and non-traders can laugh at alike. It's always good to know those people in charge of trillions of pounds have bad days at the office.
But could it have caused the flash crash? It's possible. But nobody has owned up yet and, despite the embarrassment, you would think somebody would come forward pretty swiftly given the ongoing market fallout.
It happens, but those fingers would need to have been pretty chunky.
If something goes wrong, blame the French.
And so it was last night.
Comments from the French Prime Minister emerged in the early hours of this morning dismissing Theresa May's plans for Brexit. He told reporters the UK wanted to leave the EU "but doesn't want to pay" - something he said was "not possible".
In the current currency environment, where politics seems to be driving the value of everything, the initial reaction was to scour for any comments which could have knocked confidence. Finding Hollande's remarks, some believed the French PM must have been what sent Asian traders reeling.
As a standalone theory, it's unlikely. The comments barely go further than what Angela Merkel had said on Thursday morning, and sterling managed to keep its composure in the face of that.
Moreover, all he really said was that he is sticking to the hard line which is the official EU position and has been proposed on many occasions.
Plausibility: 2/10 (as a standalone theory ...)
This theory might be the only thing less useful than a £5 note this morning. However, it's only the start of the story.
Despite trillions of dollars swishing around the forex markets every day, you do get spells where liquidity is low.
For instance, before and immediately after the referendum, a lack of money in the market was blamed for the volatile swings up and down we saw as the UK approached polling day.
Liquidity - or, basically, how much money there is in the market at any one time - matters because it ensures the smooth functioning of the markets.
For instance, imagine there was £100bn of cash ready to buy sterling at $1.26, and £100bn ready to sell it at $1.28 - these marks are called "resistance points" and define the tight range within which sterling moves. It would take a massive shift in sentiment or news to push sterling below those levels, because the owners of at least £100bn would have to all change their mind in a relatively short space of time and move their cash.
However, if liquidity is low, and there was only £5bn sitting at $1.26 and £5bn at $1.28, a movement either side could trigger a big drop. Especially as once those levels are breached, there is significantly less money sitting on the market ready to be bought or sold. Why would you put an order out there at £5bn at $1.18 when you see absolutely no chance the market will get to that level?
That means falls can be steep and they can be quick, as markets gobble up any cash they can until new "resistance points" are put in place - i.e. people chuck money at the market.
CMC Market's Michael Hewson points out: "Since 2008 a lot of major liquidity providers no longer participate in the currency markets, due to increased regulation and higher barriers to entry."
However, weak liquidity itself isn't a cause of a big fall or rise. Something needs to happen in order to trigger the chain of events which sees sterling shoot up or crash down.
Dry your eyes, sterling. Liquidity was the fuel, but something else lit the fire.
The fat finger theory is nice. But these days so much trading is carried out by algorithms - or "algos" - not much actual typing and trading is done by the market makers.
That means those "algos" have come under the spotlight in the wake of the flash crash. Could one of them have broken? A glitch in one of the rogue robots which triggers a massive sell-off all at the same time?
Possibly. But, again, it would need a trigger.
Kathleen Brooks, research director at City Index explained how it might have happened. And it all comes back to our good friend, Francois:
"These days some algos trade on the back of news sites, and even what is trending on social media sites such as Twitter, so a deluge of negative Brexit headlines could have led to an algo taking that as a major sell signal for the pound.
"Once the pound started moving lower then more technical algos could have followed suit, compounding the short, sharp, selling pressure."
This algorithmic trading helps traders ensure they don't miss out on key market events while they aren't at their desks or they are trying to scour for the relevant information, but can lead to big movements.
The effect, to a much less extreme degree, can be seen around the time of key central bank meetings. The Bank of England publishes its interest rate decision at 12pm on a Thursday. The news is instantly communicated to the markets via a myriad of channels. The pound will have already moved - up or down - at 12:00:01. That's the algorithms scraping the data and seeing the Bank voted 9-0 in favour of an interest rate cut, for instance.
They can even get more complex. Where there isn't an actual shift in policy, the robots are trained to look for key words or phrases in central bank statements, with certain language programmed to trigger a certain reaction. Again, it's why the US dollar moves within a second of the Federal Reserve's decision being published, before traders have even refreshed their own screens.
The robots are taking over ...
The consensus, nine hours after the flash crash, seems to be that everything played a part.
Francois Hollande's comments were misinterpreted by algorithms, which instantly placed sell orders on sterling. At a time of weak liquidity, the sell-off was exaggerated, triggering a massive slide in the currency.
Once the humans took back control, cooler heads prevailed.
Sterling is now at $1.2435 - two cents down on its close from yesterday, not the eight cents it reached when the pound went bump in the night.
As Deutsche Bank's Craig Nicol points out:
"Much of the blame is being laid on algo-driven orders at a time of day when liquidity is at its thinnest and it's a stark reminder of just how fragile markets can be at times when the liquidity just isn't there."
However, a two cent slide on what those cooler heads have determined was a freak occurrence is still a cause for concern.