Too Dumb to Fail?
logiccolony.com
logiccolony.com
I have less sympathy for this exploited bot, though, since you go in knowing the stock market can be a ruthlessly zero-sum kind of place.
That's my brain talking, of course. My gut reaction is to say, "stupid is as stupid does", and blame the person who wrote a buggy piece of software.
I disagree. The classic, respectable "value investor" takes advantage of other speculators mistakes by finding stocks who's value is irrationally low or high and betting against them. What's the moral difference?
If automated trading systems can't handle someone playing back at them without causing a "flash crash" then that's an argument against letting automated systems trade (at least under current regs) not an argument against the opponents who outsmarted one.
Outwitting other people gambling on the stockmarket is explicitly legal, expected even. Thieving from unguarded targets is not.
If you make a "clearly erroneous trade" (e.g., 500 shares of GOOG at price $100 instead of 100 shares at $500) you can have the trade broken. You can say "oops, undo", as long as you do it within 30 minutes.
Market manipulation, specifically deliberately placing orders/trades/news to confuse other participants, is also illegal. It's illegal if you do it by pump up a stock on yahoo message boards, and it's illegal if you do it in a boiler room, and it's illegal if you do it to a bot.
Maybe the rules should be changed, but this is what they are at the moment (at least for US equities [1]). Everyone, besides a few bad actors (Trillium, these norweigans) is playing by these rules.
[1] I know Japan, for example, does not have the "clearly erroneous trade" rule.
However there are fat-finger protection barriers. TSE, for example, will limit prices based on previous day's close.
http://www.tse.or.jp/english/rules/regulations/rulesconcerni...
If it's a gateway to business ownership then it's not zero-sum.
If it's purely game to out-arbiter other participants then it is zero-sum.
If I _know_ that a news item about a major law suite against a company
will bring down it stock (predict behavior of others) - trading is valid.
If I create a false news item in an attempt to bring down the stock - jail.
They didn't analyze current affairs to guess how the trading algorithm will respond, rather they used it to manipulate the stock price.The key phrase here is "influence the price". Taking the quote at face value, they deliberately tried to manipulate the market.
Also, there seems to be a knee-jerk reaction to HFT in the comments to this story posted anywhere. If you just run an algo, it doesn't automatically mean you manipulate the market.
The Zuckerburg in the movie thought that he deserved credit for pointing out security vulnerabilities in the Harvard network.
I do not believe the actions of these traders (or Zuckerburg) were correct. That being said, I still think their actions have value.
One way we learn is through our mistakes. There are many proprietary electronic trading platforms. Who checks them for bugs? Who checks them for accuracy or possible manipulation? I do not think these Norwegian programmers should be gaming a trading robot. At the same time, the SEC and other market regulators need to uphold their end of the bargain. If they would like to see a fair market, I think they should check trading robots for possible manipulation.
How are high speed trading systems not manipulating the market by making faster than human trades/micro-transactions?
how is it market manipulation if someone automates a patterned behaviour of trading and someone else exploits that behaviour to their advantage at the expense of 1 and only 1 party, the one overseeing the automated transactions.
if only the one HFT machine is affected, or exploited, how is it that exploiting a patterned behaviour isn't considered to be a mistake, or a bad decision, on the part of the HFT algorithm but instead considered a manipulation of stock?
if the exploit was subtle enough that no human trader could distinguish it happening, or if a human trader would never fall for the exploit (as a matter of common sense)then could the exploit be attributed to a bad algorithm and by extension a mistake by the trader who owns the HFT machine?
He apparently also documented his findings and conclusions on a trading forum so there might have been other people out there doing the same thing that haven't been caught.
I can see there is some distinction if you make trades purely for the purpose of invoking a reaction from other traders, but an intelligent AI system will discover this effect and do the same thing, albeit without any explicit human intention behind.
description of what they did from another forum:
- TMB is showing 5k bid @ 9.7 / 5k ask @ 10 - Norwegian viking takes out the ask
- TMB increases the bid to 9.8 and the ask to 10.1 - Viking is a violent heathen so he continues to hit the ask, though with less size
- Iterate this for a couple of steps
- Suddenly the viking feels full and decides to drop his inventory at TMB's bid, and then he repeats the same procedure but by shorting in several steps instead
Looks like incredibly dumb system to me. Albeit that was 2007 when the HFT arms race has just begun.