The Risk Of Algos

Authored by Lance Roberts via RealInvestmentAdvice.com,

Mike ‘Wags’ Wagner: ‘You studied the Flash Crash of 2010 and you know that Quant is another word for wild f***ing guess with math.’

Taylor Mason: ‘Quant is another word for systemized ordered thinking represented in an algorithmic approach to trading.’

Mike ‘Wags’ Wagner: ‘Just remember Billy Beane never won a World Series .’ – Billions, A Generation Too Late

My friend Doug Kass made a great point on Wednesday this week:

“General trading activity is now dominated by passive strategies (ETFs) and quant strategies and products (risk parity, volatility trending, etc.).

Active managers (especially of a hedge fund kind) are going the way of dodo birds – they are an endangered species. Failing hedge funds like Bill Ackman’s Pershing Square is becoming more the rule than the exception – and in a lower return market backdrop (accompanied by lower interest rates), the trend from active to passive managers will likely continue and may even accelerate this year.”

He’s right, and there is a huge risk to individual investors embedded in that statement. As JPMorgan noted previously:

Quantitative investing based on computer formulas and trading by machines directly are leaving the traditional stock picker in the dust and now dominating the equity markets.

While fundamental narratives explaining the price action abound, the majority of equity investors today don’t buy or sell stocks based on stock specific fundamentals. Fundamental discretionary traders’ account for only about 10 percent of trading volume in stocks. Passive and quantitative investing accounts for about 60 percent, more than double the share a decade ago.

As long as the algorithms are all trading in a positive direction, there is little to worry about. But the risk happens when something breaks. With derivatives, quantitative fund flows, central bank policy and political developments all contributing to low market volatility, the reversal of any of those dynamics will be problematic.

There are two other problems currently being dismissed to support the “bullish bias.”

The first, is that while investors have been chasing returns in the “can’t lose” market, they have also been piling on leverage in order to increase their return. Negative free cash balances are now at their highest levels in market history.

Yes, margin debt does increase as asset prices rise. However, just as the “leverage” provides the liquidity to push asset prices higher, the reverse is also true.

The second problem, which will be greatly impacted by the leverage issue, is liquidity of ETF’s themselves. As I noted previously:

“The head of the BOE Mark Carney himself has warned about the risk of ‘disorderly unwinding of portfolios’ due to the lack of market liquidity.

‘Market adjustments to date have occurred without significant stress. However, the risk of a sharp and disorderly reversal remains given the compressed credit and liquidity risk premia. As a result, market participants need to be mindful of the risks of diminished market liquidity, asset price discontinuities and contagion across asset markets.’”

When the “robot trading algorithms”  begin to reverse, it will not be a slow and methodical process but rather a stampede with little regard to price, valuation or fundamental measures as the exit will become very narrow.

Importantly, as prices decline it will trigger margin calls which will induce more indiscriminate selling. The forced redemption cycle will cause large spreads between the current bid and ask pricing for ETF’s. As investors are forced to dump positions to meet margin calls, the lack of buyers will form a vacuum causing rapid price declines which leave investors helpless on the sidelines watching years of capital appreciation vanish in moments.

Algo’s were not a predominant part of the market prior to 2008 and, so far, they have behaved themselves by continually “buying the dips.” That support has kept investors complacent and has built the inherent belief “this time is different.”

But therein lies the “risk of the robots.”

What happens when these algo’s reverse course and begin to “sell the rallies” in unison?

I don’t want to be around to find out.

Comments

Keltner Channel Surf Fri, 04/13/2018 - 12:22 Permalink

Yet another old school whiner, no longer competitive, who foolishly believes machines have more risk than panicking humans, who were actually the ones creating the last couple v-bottoms (fed 'hawk', trump follies) that the machines simply reverted back, quite calmly, to the mean for profit.  Computers aren't going away, not in cars, TVs or the market ...

MedicalQuack Fri, 04/13/2018 - 12:26 Permalink

Indeed the world belongs to the programmers...quants in healthcare video, I exposed a ton of it..

http://ducknetweb.blogspot.com/2017/11/algorithms-scoring-metrics-priva…

And you know 10 years ago I wrote about asking if we needed some kind of department of algorithms to help stop algorithmic fraud..this tweet about says it all...nobody's taking an oath last I looked..and how could a mediocre programmer like me see this and everyone else miss it?  By plan of course as they make money doing it.  

10 Years ago..The Modelers Hippocratic Oath-by Paul @wilmott & @EmanuelDerman-read it bit.ly/2EeRDcd Inspired me to post this in 2009: Does the US need some kind of Dept of Algorithms to stop algorithmic fraud? bit.ly/b7eLVa

Salmo trutta Fri, 04/13/2018 - 12:58 Permalink

B.S.

 

This how I denigrated Nassim Nicholas Taleb’s “Black Swan” theory (unforeseeaable event), 6 months in advance and within one day:

[1] To: anderson@stls.frb.org
Subject: As the economy will shortly change, I wanted to show this to you again – forecast:
Date: Wed, 24 Mar 2010 17:22:50 -0500
Dr. Anderson:
It’s my discovery. Contrary to economic theory and Nobel Laureate Milton Friedman, monetary lags are not “long & variable”. The lags for monetary flows (MVt), i.e., the proxies for (1) real-growth, and for (2) inflation indices, are historically, always, fixed in length.
Assuming no quick countervailing stimulus:
2010
jan….. 0.54…. 0.25 top
feb….. 0.50…. 0.10
mar…. 0.54…. 0.08
apr….. 0.46…. 0.09 top
may…. 0.41…. 0.01 stocks fall
Should see shortly. Stock market makes a double top in Jan & Apr. Then real-output falls from (9) to (1) from Apr to May. Recent history indicates that this will be a marked, short, one month drop, in rate-of-change for real-output (-8). So stocks follow the economy down.
And:
flow5 Message #10 – 05/03/10 07:30 PM
The markets usually turn (pivot) on May 5th (+ or – 1 day).
I.e., the May 6th “flash crash”, viz., the second-largest intraday point swing (difference between intraday high and intraday low) up to that point, at 1,010.14 points.

 

--– Michel de Nostradame

CoCosAB Fri, 04/13/2018 - 13:33 Permalink

" What happens when these algo’s reverse course and begin to “sell the rallies” in unison? "

 

NOTHING!

 

The fucking financial terrorists simply CLOSE THE MARKETS!

Downtoolong Fri, 04/13/2018 - 14:20 Permalink

Q: What’s dumber than a computer algo that thinks it’s OK to buy a $40 stock for $0.01 after it flash crashes?

A: A computer algo that buys a $40 stock for $0.01, then turns around and sells it for $0.02 on its way back to $40 (which is precisely  what many market making algos will  do).

 

SRV Fri, 04/13/2018 - 15:07 Permalink

What's amazing is the silence on the subject the past decade... after completely taking over the equity trading... not so much as a whimper.