Post-crash market efficiency 1696-2020

WHAT DO past stock mar­ket crashes tell us about the Effi­cient-Mar­ket Hy­poth­e­sis? In a melt­down, do prices be­come more pre­dictable?

In the­ory, they in­stantly ad­just to all pub­lic in­for­ma­tion. Hence tech­ni­cal anal­y­sis – pre­dict­ing fu­ture stock prices from pat­terns in past ones – should not work. In par­tic­u­lar, there should be no us­able trends – ten­dency for prices to keep mov­ing up or down – nor pre­dictable re­ver­sals.

On the other hand, big crashes are so rare and ex­treme that they might be ex­cep­tions. In a crisis, per­haps enough traders act im­pul­sively, or at least do some­thing un­usual, that there’s a smidgeon of pre­dictabil­ity. Surely greed and fear don’t always ex­actly can­cel out?

So let’s take a look. Do big crashes on av­er­age pro­duce any kind of pat­tern or trend in prices?

I’ll con­sider all ma­jor melt­downs that have ever hap­pened in the US and UK, us­ing their broad in­dices, the S&P 500 and FTSE All-Share; and defin­ing a crash as a fall of 20% or more within 8 weeks.

UK crashes

Start­ing with the UK (as its stock mar­ket be­gan first), here are all crashes through­out his­tory:

Each coloured line is a crash – start­ing with a 20% fall – and its af­ter­math. I’ve even in­cluded the Bank of England crisis of 1696, and the still-fa­mous South Sea Bub­ble of ex­actly 300 years ago (which coined the term ‘bub­ble’), though both are omit­ted from the av­er­ages. There­after, de­spite slides and slumps, there were no more ac­tual crashes un­til World War 2, af­ter which they be­came more fre­quent.

The red line is the cur­rent crash, from mid-Jan­uary up to this week. The FTSE fell fur­ther, faster, than ever be­fore, then has partly re­cov­ered; though this may yet turn into a dead cat bounce.

Su­perfi­cially, all these crashes have lit­tle in com­mon: caused by spec­u­la­tion, oil, pro­gram trad­ing, ter­ror­ism, dis­ease; all kinds of shapes – L, V, U, W; they fall 20% or 70%, take days or years to hit rock bot­tom, and up to a decade to re­cover. Like spaghetti, the lines are all over the place.

But to find an un­der­ly­ing pat­tern of sorts, we can sim­ply take an av­er­age: the solid black line shows the mean of all crashes over the past cen­tury. And af­ter the ini­tial drop, it’s ba­si­cally flat for the rest of the year. On av­er­age, the mar­ket didn’t un­der- or over-re­act: prices nei­ther kept fal­ling, nor bounced or re­cov­ered. There is no sign of any trend. So the fact that there has been a crash pre­dicts noth­ing about whether stocks will go up or down, ei­ther days or months later; not even by a few per­cent.

This con­sis­tency is sur­pris­ing. Not least as half these crashes pre-date com­put­ers, so there was no algo trad­ing, far less data and num­ber-crunch­ing, and vast scope for bi­ases and iffy strate­gies.

If we ex­clude those bad old days, and av­er­age the three crashes of the com­puter era – 1987, 2001 and 2008 – the pic­ture stays much the same (see dashed black line). In week 8, you may as well toss a coin to de­cide whether to buy or sell; for by year end, prices will be al­most un­changed. And if it looks like you could make a few per­cent by week 42, con­sider the huge vari­a­tion be­tween these crashes, with high volatility: your risk-ad­justed re­turn would be mediocre.

Of course, this does not prove that the Effi­cient-Mar­ket Hy­poth­e­sis holds for crashes. We are look­ing at the whole UK stock mar­ket, not in­di­vi­d­ual shares, which may still have mis­pric­ings – though per­haps too small, rare, or er­ratic to make much money from. And we are only con­sid­er­ing price his­tory, not other in­for­ma­tion that might mat­ter (e.g. news of a virus). Once the mar­ket has fallen 20%, ev­ery­one knows why; but ear­lier on, traders may not see the storm clouds form­ing.

US crashes

Across the pond, stock trad­ing didn’t get go­ing un­til the 19th cen­tury. The first melt­down was the fa­mous one, the Wall Street Crash of 1929. This was fol­lowed by oth­ers, about one per decade – in­creas­ingly in sync with UK crashes, due to global­iza­tion:

This is similar to, but not quite the same as, the UK chart. For the solid black line shows a grad­ual rise through the year, some­what faster than usual (which is 6.6% p.a. for this in­dex). But again, there are prob­a­bly no ex­cess re­turns once you ad­just for risk. And in any case, the dashed black line sug­gests that this slope was steam­rol­lered by the ad­vent of com­put­ers.

There’s a bit of a post-crash bounce in the dashed line, but this looks un­likely to be a real effect as it wasn’t there be­fore 1987, and is similar to ran­dom me­an­der­ings in later months.

So, past US and UK melt­downs seem to con­firm the Effi­cient-Mar­ket Hy­poth­e­sis. Like a gold­fish bump­ing into the side of its bowl, in­dices in­stantly for­get they’ve crashed, and just carry on as nor­mal.


(P.S. Oh, but things seem to get oddly pre­dictable af­ter a year, as crashes re­cover. Maybe I’ll write that up, if there’s enough in­ter­est.)