Would a random market kill switch reduce market volitility?

[sengsational]

In these 2016 market threads, I've seen comments that involve laments about how a cascade can start because of a tiny disturbance. This post centers around a thought experiment concerning a possible option to tame market volatility. It relies on the premise that the volatility is caused in part by professional traders' actions and that those destabilizing actions would be less frequent if the reliability of the market was reduced.

If the market was suddenly and unexpectedly shut-down*, how would that affect professional traders? If a 'normal investor' buys one hour and sells the next (or later), I don't think it would matter. But isn't it true that there's a lot of transactions the professionals use that are only low risk because they're "guaranteed" that some set of transactions will succeed as expected within a few seconds with very high reliability? If the certainty of having the full set of transactions execute very close to the same time was decreased, wouldn't that involve more risk for the trader? It seems that they could get caught with only part of their strategy executed, which might be very costly, thus making those types of transactions ill advised.

After typing this, I realized there's probably academic papers on this topic, but I'm not going to look for them...yet.


* You'd have a completely isolated computer, inside a vault at Fort Knox, with a true random (not psudo random) source, like a video camera pointed at a lava-lamp (lol!). The simple open-source program would be configured to "fire", say, once per 50 trading days on average, so it would have minimal impact on the real business of the market. The only connection to the computer would be an optical link through a small hole in the vault (the program would turn on an LED when it fired). The external sensor would be connected to the existing "circuit breakers" that were implemented years ago.

 

[marko]

I could be wrong but my sense is that the real culprits are machines doing these trades with very little human interaction outside of someone initiating a set of parameters, perhaps months ago.

Wasn't the 1000 point flash crash similar to this in reverse?

Having said that, I wonder if such a random shutdown would 1) be legal and 2) only add more uncertainty to an already jittery world.

 

[Texas Proud]

There are circuit breakers that do prevent program trading if things happen... but we are a LONG way from that happening... we are having swings of 1% to 3%.... not 10% or more like in the past...

Adding a random shutdown (you did not say how long it would be shut down, so I would assume minutes not days) would do nothing....

Think about the proposition you made... they trade in seconds or milliseconds.... your proposal is to stop that on average every 50 days... that is just background noise.... IOW, for 49.9999 days I can trade all I want and that .0001 day you close me down.... I still win...

 

[Closet_Gamer]

You should suggest that to the Chinese govt...they appear to be willing to try anything to stop their markets implosion. They've already told big shareholders they can't sell and told other people they must buy...flipping the off switch for a few minutes? They'd try that just to see what happens at this point!!

 

[DrRoy]

Some of the Chinese market selloff was because there were circuit breakers that stop trading after some point and the big players were determined to be the ones that got out. Some of their stabilization came after the circuit breaker was removed, traders did not have fear of not being able to sell, and the selling quieted down.

 

[photoguy]

Do we want less market volatility? I'm not sure that I do.

Suppose your plans succeeds and you got rid for extreme price swings -- what would that do for the equity risk premium?

 

[Rustic23]

Not a market historian. But, is there really more volatility now vs the past, or does it just seem that way. Is there better information, more rapidly available? I remember a friend in the 80's that lost a boat load of money on one day. By the time he got hold of his broker the worst was done. How bad was volatility during the stock market crash of the 30's? It would seem that program trading would be faster, but is it? I look forward to see what this thread generates.
Thanks.

 

[HadEnuff]

Would these market volatility measures be implemented when the market is rising "too fast" or only during sell offs?

 

[sengsational]

Think about the proposition you made... they trade in seconds or milliseconds.... your proposal is to stop that on average every 50 days... that is just background noise.... IOW, for 49.9999 days I can trade all I want and that .0001 day you close me down.... I still win...

You're probably right about my choice of tuning parameters. And just to be clear, it's not a penalty to the program traders while the system is down. It's an unknown that, during trading, everything might not go as planned, and so increase the chance that the set of trades don't execute "right" and increase the chances that money could be lost.

The idea would be to tune it such that the market is still able to act efficiently for investors while preventing professionals (yes, you can read that as "computers with programs running that execute trades") from gaming the system.

As to the duration, it would be long enough to disrupt the players gaming the system, but not long enough to mess-up investors. Also, the duration of the shut-down would be variable (to prevent the gamers from leveraging that knowledge to their advantage).

As to if we really want less volatility, I was taught that the best markets are ones that reflect the value of the thing being traded, and not some value that came about by entities that game the system. I wouldn't worry about too little volatility, since after all, we still have most people trading on emotion.

So if tuning the system so that it halted trading more often, but for shorter periods, then let's say that's the proposal.

Or maybe we should ask the question whether any tuning parameters would be successful at disrupting those that profit from this kind of program trading. I was under the impression that the trades are about half profitable and half unprofitable, but the profitable ones edge out the unprofitable ones. I obviously don't know much about the mechanics of this type of trading, but if a short shutdown would tilt the balance, that's all we need.

As with most things, if those invested in the status quo don't want it, it ain't happening. But it's fun to dream.

 

[Senator]

All it would take is to have each trade take 10 seconds to clear, not sub-second. Like the old days. Or immediate settlement, not 3-days.

If you do not have the liquidity, you have to stay home.

 

[ERD50]

First, I agree with others - is volatility worse? Is it bad?

I don't get the once every 50 days thing at all. The big boys have done a bazillion trades in that time. So they take a break, or lose the advantage once every few months?

If this would have any effect, I'd think it would be something that happens many times a minute, for a second or two. I would not notice a one or two second delay, so no effect on me.

But if everyone gets frozen for two seconds, does that even effect the big boys? Wouldn't they be right back where they were when the clock starts again?

Random small delays for random orders?

IIRC, and if I understand correctly - the flash traders were able to 'see' a bid or ask at one point along the line, and that trade would be sent out to other floors for execution. They could run an order fast enough to beat the other offer to that floor.

Seems to me no one should have access to that information along the way. I'm not sure how some random clock stalls change this.

-ERD50

 

[photoguy]

The idea would be to tune it such that the market is still able to act efficiently for investors while preventing professionals (yes, you can read that as "computers with programs running that execute trades") from gaming the system.

Random small delays for random orders?

IIRC, and if I understand correctly - the flash traders were able to 'see' a bid or ask at one point along the line, and that trade would be sent out to other floors for execution. They could run an order fast enough to beat the other offer to that floor.

Seems to me no one should have access to that information along the way. I'm not sure how some random clock stalls change this.

There's already an exchange to tries to eliminate the ability of traders to game the system:

https://en.wikipedia.org/wiki/IEX

I believe their primary method of doing this is ensuring that no trader gets information ahead of others. Supposedly they use a 38mile long spool of wire to delay everything. I believe IEX was covered in one of Michael Lewis' books and in various news stories.

I'm not sure if this covers the types of issues that Sengsational is looking to address

As to if we really want less volatility, I was taught that the best markets are ones that reflect the value of the thing being traded, and not some value that came about by entities that game the system. I wouldn't worry about too little volatility, since after all, we still have most people trading on emotion.

Some volatility is due to gaming (which I agree should be removed). But some volatility is just due events being binary and information traveling instantly. E.g. a company reporting that it failed to meet earnings can sink its stock instantly -- I think this volatility is good.

 

[sengsational]

After reading the IEX link, and the posts here,it looks like the problem wouldn't be addresses by implementing random shut-downs, no matter how frequent. The high frequency traders are crying foul with a 0.0007 second delay...I can't imagine what they'd say about applying short delays to a random set of orders. We're doomed to let the leeches do their front running

 

[ERD50]

After reading the IEX link, and the posts here,it looks like the problem wouldn't be addresses by implementing random shut-downs, no matter how frequent. The high frequency traders are crying foul with a 0.0007 second delay...I can't imagine what they'd say about applying short delays to a random set of orders. We're doomed to let the leeches do their front running

A little side note - about a year ago, I ran into a sales rep that I did a lot of business with at MegaCorp (expensive lab and production electronic measurement equipment). I asked what he was doing now that MegaCorp was a shadow of it's former self.

He said a lot of his business was supplying high accuracy time bases to these HFT groups. I can't even imagine why they would need anything high tech (or so many?), timing down to fractions of a micro-second is child's play (and cheap) these days. Do they need pico-second timing? I don't get it.

-ERD50

 

[FUEGO]

This sounds horrible. Sometimes I'll sell one ETF and buy another within a few minutes so I can rebalance. Not gaming any system, just maintaining my target asset allocation. I'll also tax loss harvest by selling one security and buying something similar (but not substantially similar ). I'd hate to be forced out of the market halfway through a 2 part trade.

 

[sengsational]

...buy another within a few minutes....

That's an eternity in HFT terms; short frequent pauses probably wouldn't affect individual investors. But, as I've said, I don't think they'd affect HFT's so the point is moot.

 

[sengsational]

He said a lot of his business was supplying high accuracy time bases to these HFT groups. I can't even imagine why they would need anything high tech (or so many?), timing down to fractions of a micro-second is child's play (and cheap) these days. Do they need pico-second timing? I don't get it.

Could it be that they need it to work no matter what the volume of transactions? When the sh.. stuff hits the fan, you've got to have a system that's guaranteed to do it precisely on time. A regular program running in a regular system (not system designed for real-time processing), will try to do things in the millisecond you ask, but would be vulnerable to a larger load on the system causing things not to happen on time. Just a guess.