High Frequency Trading
Dennis Dick discusses “Information Arbitrage” at Bright Trading. Topics include high frequency trading, informed and uninformed order flow, strategy, and tactics for more efficient execution.
AEP and NEE lose more than 50% of their value in one second today, and then rebound entirely. This entire move caused by market structural issues. Here is the tale of the tape:
To understand more about trade-throughs and mini-flash crashes, see this month’s CFA Magazine: Erroneous Combustion
Harris Corp reported preliminary earnings tonight. They cut their 2013 outlook and announced job cuts, which is obviously not good for the stock. HFT news algorithms specialize in trading off these headlines and picking off limit orders that are out to lunch.
Check out the tape:
The news was reported at 16:20:00. Notice the time it took the HFT news algorithm to respond to the news – instantaneously.
It is a very dangerous game to make markets after hours in this HFT world, and the traders that bought this stock between $45 and $46 just learned the hard way, as the stock is now trading at $43.50.
They were sitting ducks.
Some market participant is running a very aggressive news algorithm. At 16:10:10, Procter & Gamble (PG) lowered their 2013 guidance down slightly, citing devaluation of Venezuelan currency. The news algorithm went to work, responding instantaneously to the news. Here is the tape:
(Note the ticker tape reads from the bottom-up)
As you can see from the tape, the news algorithm sold the stock down to a price of $74.10. The stock a few seconds later jumped back up to $76. This algorithm just threw away a huge chunk of change as nearly 10K shares were sold at the bottom.
These trading overshoots have been happening a lot lately, especially in the most recent earnings season. I wonder if the participant that is running this algorithm will eventually stop running it so aggressively as it appears to be burnings some serious cash.
Two HFT algorithms battled it out for 12 minutes today in preferred stock FNFG.PRB. Both of these algorithms wanted to be the best bid. One algo bids $29.45, the other penny jumps and bids $29.46. Then the other penny jumps back and bids $29.47, the other algo follows and bids $29.48. They battle up to a bid of $29.62 at which point the one algo cancels and moves its bid back to 29.45, and then the cycle repeats itself. This battle continues for the next 12 minutes with the bid price changing more than 6000 times. Total shares transacted on the exchange during this period is zero.
Take a look at the video to get a feel for how quickly the quote was changing:
Note: Video shows entire 12 minute period of algo loop that repeats itself every two seconds.
Yesterday, the SEC held a roundtable discussion on decimalization. Some participants suggested a pilot program which would change the minimum quoting increment in the mega-caps to tenths of a penny from the current penny. Could you imagine the quote traffic when algos battle for the top of the book in sub-pennies? At least Nanex will have fun analyzing all the quote pollution.