Market Data Firm Spots the Tracks of Bizarre Robot Traders

[Triple Zero]

Mysterious and possibly nefarious trading algorithms are operating every minute of every day in the nation's stock exchanges.

What they do doesn't show up in Google Finance, let alone in the pages of the Wall Street Journal. No one really knows how they operate or why. But over the past few weeks, Nanex, a data services firm has dragged some of the odder algorithm specimens into the light.

The trading bots visualized in the stock charts in this story aren't doing anything that could be construed to help the market. Unknown entities for unknown reasons are sending thousands of orders a second through the electronic stock exchanges with no intent to actually trade. Often, the buy or sell prices that they are offering are so far from the market price that there's no way they'd ever be part of a trade. The bots sketch out odd patterns with their orders, leaving patterns in the data that are largely invisible to market participants.

In fact, it's hard to figure out exactly what they're up to or gauge their impact. Are they doing something illicit? If so, what? Or do the patterns emerge spontaneously, a kind of mechanical accident? If so, why? No matter what the answers to these questions turn out to be, we're witnessing a market phenomenon that is not easily explained. And it's really bizarre.

Read more >>

[Triple Zero]

Also check out the gallery of other "Crop Circle" quote sequences:

http://www.nanex.net/FlashCrash/CCircleDay.html

I have no idea what this all means, but the fact that thousands of bids and asks and quotes per second are being algorithmically generated, and interact with eachother, creating Emergent and Complex patterns, can only mean that finances are completely out of human control and basically run by machines.

Where I'm using the terms Emergent and Complex in somewhat of the LessWrong sense of the words, in that they are not an explanation for what is going on, and that we really don't know what's going on, and that this is ultimately a Bad Thing until we figure out what is going on. :-P

[Captain Utopia]

Honestly, when I read of this my first thought was that other trading programs would be reading those crazy values, incorporating them into averages for their own calculations, and thus make slightly worse decisions.  Maybe just enough to account for a few million dollars a day or more, but not enough to set off alarms, attract attention and warrant an immediate investigation likely to uncover the source of the error.

It's what I'd do if I wanted to give my AI trader an advantage over those from competing organisations.  And if I were a fucking banker.

[Captain Utopia]

I have no idea what this all means, but the fact that thousands of bids and asks and quotes per second are being algorithmically generated, and interact with eachother, creating Emergent and Complex patterns, can only mean that finances are completely out of human control and basically run by machines.

Say your systems can detect the pattern to these anomalous trades coming from a nefarious source trying to screw with your financial models.  The first thing to do is to change your model to account for the pattern, such that you're back where you started before the bad signal started.

The second thing to do is to introduce your own signal at a different frequency such that it modulates the first signal into higher amplitudes.  You assume that the original source is adjusting for the signals which it is generating, but unprepared for the amplification you provide.  Escalation and hilarity would ensue, as would funky emergent patterns.

[Cain]

Bot traders like these are the suspected cause of a very temporary stock market crash a couple of months back

http://ftalphaville.ft.com/blog/2010/05/07/222911/blame-the-trading-bots-again/

The blame for Thursday's Dow Jones rout has swiftly shifted from a fat-fingered Citigroup trader to the rise of the machine-trading bots.That's right, the bots took $1,000bn off the DJIA (at one point).

...

The Wall Street Journal is also reporting that a few high-frequency firms, including Tradebot Systems, closed for business on Thursday. Their systems are designed to pull back when trading gets too volatile.

This is something we've heard before – most notably in a working paper from the Fed.

That paper studied the behaviour of FX trading bots as US payrolls were released — a key datapoint for currency traders. It found that in the first minute or so after the release, computers backed away from trading. Reasons given then were that algorithms simply weren't designed to deal with sharp moves.

Considering that the defense of HFT and algorithimic trading often centres around the provision of liquidity — often in times of severe market stress — events like Thursday's probably won't help endear the trading bots to many.

Speaking of which, whatever happened to the SEC's review of HFT?

And will NYSE Euronext swiftly be revising its stance on the the practice?

[Rumckle]

CD: has a good point.

I was thinking (though it is hard to tell without knowing what prices they were trading), that perhaps it is some kind of insurance against an unexpected boom or drop in the markets (it would probably have to be fairly large though).

[Golden Applesauce]

Honestly, when I read of this my first thought was that other trading programs would be reading those crazy values, incorporating them into averages for their own calculations, and thus make slightly worse decisions.  Maybe just enough to account for a few million dollars a day or more, but not enough to set off alarms, attract attention and warrant an immediate investigation likely to uncover the source of the error.

It's what I'd do if I wanted to give my AI trader an advantage over those from competing organisations.  And if I were a fucking banker.

I suspect this is the reason.  The people who write these programs are very, very smart people, and one of them must have thought along the lines of:
Which cases does my algorithm not handle well? -> Which cases do my competitors' algorithms not handle well? -> How can I cause those cases to happen?

On the other end of malice vs. stupidity, it is possible that at least some of the HF programs didn't properly take into account that the majority of trades are now done by other HF programs and therefore make their assumptions based on what kinds of patterns happen in real human trading.  If that's the case, then 'garbage' input would cause equally garbage output, and the programs would feed bizarre quote patterns into each other.  Somewhere out there is an HF program that really thinks the share price for X is oscillating between $12.24 and $0.01 without hitting any point in between, and making screwy oscillating quotes of its own based on that interpretation.

[Cramulus]

so is this a russian botnet or what?

[Bebek Sincap Ratatosk]

so is this a russian botnet or what?

DAMN RUSKIBOTS!

[Rumckle]

so is this a russian botnet or what?

Have you not been paying attention to the past four years of posts?

Everything is a Russian botnet, including this.

I mean, in soviet Russia, financial markets introduce algorithms into you.

[Triple Zero]

so is this a russian botnet or what?

in case you were being serious: no. at least, the russian botnets that have been discussed here (in this subforum) concern themselves mostly with phishing, spam and DDOSing stuff. I haven't read anything about them being on the stock market. neither do I think it'll be very likely for them to be, since they consist for 99.9999% of hacked personal desktop computers, that can't get onto these high-frequency trading stock market thingies.

as far as I can tell from the article, these are scripts coded on special terminals (computers) linked via a very-low-latency network connection (possibly not even Internet) to the stock markets. and it's all sort of legal.