5 ms·

From what I've read, one key issue was that some exchanges stopped trading or updating certain prices for a few minutes, which led to other exchanges making tra

From what I've read, one key issue was that some exchanges stopped trading or updating certain prices for a few minutes, which led to other exchanges making trades in isolation instead of tying them into the larger market. With just a few more "sell at any price" than "buy at a reasonable price" orders for a given stock on a given exchange, and with no tie-in to other buy orders from other places, the "buy at a ridiculously low price" orders that many traders leave in place just to take advantage of algorithmic glitches of this nature were triggered. It's like being at a car auction with a robot auctioneer and putting in a one-penny starting bid, and then the robot's sensors malfunction so he can't see any more bids. He'll declare you the auction winner, but that result doesn't reflect the car's actual value and will almost definitely be canceled.

rdtsc 17y ago
> result doesn't reflect the car's actual value and will almost definitely be canceled. Unless the value of the car is defined as whatever someone on NYSE or NASDAQ is willing to pay for it. As the result of this there were some losers and some winners. Losers will want the trades reversed, winners won't. Reversal of trades could start a dangerous trend. Who gets to decide which trades get reversed? I wonder if many HFT firms will start to incorporate such kind of behavior into their trading model, so we'll see more of these "accidents" in the future.
- lotharbot 17y ago
  The way the stock exchanges are set up, they can cancel trades that are considered "erroneous". It's in the contract, and it has been done before. Either party can trigger a mediation attempt [1], but in this case the exchange itself determined that they made a mistake in generating trades that shouldn't have actually happened and was only triggered by a glitch in the system, and made a blanket declaration. Since any trade on their system is actually just a contract to exchange the stock in the next 3 days, they don't have to take back any stock, they just have to declare the contract as invalid. Last I heard, the Nasdaq was going to cancel any trades that were during the glitchy time window and were more than 60% away from a baseline price (an amount people were "willing to pay for it" when the system wasn't glitching). I don't know how they came up with the number, nor how deeply the SEC and other agencies were involved in setting it. There will still be plenty of winners and losers coming out of this, just not the ones who thought they bought or sold Accenture at a penny. I'm not saying this is a "good" fix, but it is well within the Nasdaq's authority. [1] http://news.ycombinator.com/item?id=1325574 http://news.ycombinator.com/item?id=1325574 - notes that broken trades are fairly common, and that they're usually quickly mediated
- eru 17y ago
  > [...], so we'll see more of these "accidents" in the future. No, we'll see less of these trades. Because the next guy will set up "buy everything at $ 0.02", so that the $ 0.01 guys don't get everything. Of course they anticipate each other, and will bid each other up to sensible prices.
sireat 17y ago
The question is where do we draw the line? Ok, we disallow 0.01% of price,penny bids. However, should 10% of price bid orders be left standing, how about 20%, 50%? Certainly 75% of price bids should be honored?

rdtsc 17y ago

> result doesn't reflect the car's actual value and will almost definitely be canceled. Unless the value of the car is defined as whatever someone on NYSE or NASDAQ is willing to pay for it. As the result of this there were some losers and some winners. Losers will want the trades reversed, winners won't. Reversal of trades could start a dangerous trend. Who gets to decide which trades get reversed? I wonder if many HFT firms will start to incorporate such kind of behavior into their trading model, so we'll see more of these "accidents" in the future.

lotharbot 17y ago
The way the stock exchanges are set up, they can cancel trades that are considered "erroneous". It's in the contract, and it has been done before. Either party can trigger a mediation attempt [1], but in this case the exchange itself determined that they made a mistake in generating trades that shouldn't have actually happened and was only triggered by a glitch in the system, and made a blanket declaration. Since any trade on their system is actually just a contract to exchange the stock in the next 3 days, they don't have to take back any stock, they just have to declare the contract as invalid. Last I heard, the Nasdaq was going to cancel any trades that were during the glitchy time window and were more than 60% away from a baseline price (an amount people were "willing to pay for it" when the system wasn't glitching). I don't know how they came up with the number, nor how deeply the SEC and other agencies were involved in setting it. There will still be plenty of winners and losers coming out of this, just not the ones who thought they bought or sold Accenture at a penny. I'm not saying this is a "good" fix, but it is well within the Nasdaq's authority. [1] http://news.ycombinator.com/item?id=1325574 http://news.ycombinator.com/item?id=1325574 - notes that broken trades are fairly common, and that they're usually quickly mediated
eru 17y ago
> [...], so we'll see more of these "accidents" in the future. No, we'll see less of these trades. Because the next guy will set up "buy everything at $ 0.02", so that the $ 0.01 guys don't get everything. Of course they anticipate each other, and will bid each other up to sensible prices.

sireat 17y ago

The question is where do we draw the line? Ok, we disallow 0.01% of price,penny bids. However, should 10% of price bid orders be left standing, how about 20%, 50%? Certainly 75% of price bids should be honored?