High Frequency Trading: Winners and Losers
Board: Macro Economics
This Sunday, 5/6/12, will be the 2nd anniversary of the “Flash Crash.” The Flash Crash will be remembered for seeing the Dow 30 lose about 1,000 points in a matter of minutes, then recovering in a matter of minutes. The round trip down then back up took about 20 minutes. The day was also interesting because many stocks traded at the lower “stub quote” price of 1 cent per share. A few issues traded at the upper stub quote of “99,999.99” per share. Wikipedia has a nicely written history of the day, plus different theories and explanation for what occurred.  The official explanation written jointly by the SEC and Commodities Futures Trading Commission largely blamed a large trade by mutual fund company Waddell-Reed.  I took strong issue with their explanation of the crash. 
A few days after the Flash Crash, I posted a “High Frequency Trading Tutorial” which was the best knowledge I had at the time.  Rimpynths replied: “This post is filled with so much misinformation; I don't even know where to begin.” Clearly he thought I got a lot of things wrong.
As I reflect back on that post, one point still stands out. There is almost a total absence of hard facts about what HFT is and does. I portrayed HFT one way, Rimpy portrayed it another way. Neither one of us could produce any kind of hard documentation at the time that conclusively proved the case. The same situation exists today. This is very different from a lot of topics that were/are discussed on METAR. Interest rates, stock prices, exchange rates, GDP etc are hard facts that everyone can agree on. To my knowledge, NOBODY has or can produce similar hard facts on HFT. I think the main reason for this is that everyone that is making money using HFT does not want to come forward and tell the world exactly how they do it. Plus if they are doing anything marginally legal, they sure do not want to broadcast that. With that disclaimer, this post contains what I think are facts surrounding HFT. At the end, I will include some 100% speculations on what I think HFT’s might be doing. I will attempt to keep them separate. If Rimpy is still around, he very well might take issue with my version of the facts. Rimpy and I probably agree that HFT’s are blamed for more problems than is fair. I do think the public has an unduly negative view of HFT’s in some regards.
Before we get into HFT’s, it is important to briefly review “old” stock trading BEFORE all of the electronic trading started. This is a KEY piece of the puzzle that explains a lot of what HFT is/does today. If you do not understand how trading used to be done, you will not appreciate how trading works today.
Stock Trading- “The way we were”
In the “old” days, nearly all US stock trading was done on the New York Stock Exchange. This was before NASDAQ opened for business in 1971. Each stock on the NYSE had a “specialist” in charge of insuring orderly trading. The specialist had a physical location called “the trading post.” If you wanted to buy or sell that issue, you went to the trading post. The specialist would match up buyers and sellers. In those days, stocks traded in fractions of dollars, instead of decimals. It was common to have the bid-ask spread be “1/8th- 12.5 cents or “1/4” 25 cents per share. For example you might buy IBM for 10.00 per share and sell it for 9.75 per share. Aside from the very high commissions, this bid ask spread made short term trading very difficult for the small investor. If you lost 25 or 50 cents every round trip, it would rapidly add up.
As part of the specialist’s role, he also bought and sold for his own account. Nominally, he would step in to be the buyer or seller of last resort. Specialist trades were on the order of 5% to 10% of the trading volume. It was an open secret that occasionally specialists traded opportunistically even when they were NOT the last resort. If the specialist bought at 9.75 and sold at 10.00, he could make a nice profit. The specialist could also short a stock if he wanted to.
I have heard specialists from that era comment that you “really had to mess up to lose money.” Typically they would lose money one or two days PER YEAR. If they lost money on more days that two, they were doing something very wrong.
“Front running” is where you know FUND XYZ is about to buy a large number of say IBM shares. If you recognize that situation, you would attempt to buy IBM before the fund, then resell your shares for a quick profit. This type of front running was and is 100% legal. In the old days, “floor traders” working for other firms would attempt to front run anybody and everybody. Specialists certainly had the ability to front run orders also, despite the fact that it was against the rules. Off the record, specialists would give you a wink and a nod if you asked them if they ever did this. They only had to do it every once in a while to make a nice profit. Had they done it on every single trade, they would have gotten in trouble.
Bottom line of “old” trading was that floor traders and specialists were pulling off some percentage of your profits at every opportunity.
Stock Trading- “The way it is”
Other than the incredible improvements in technology, two large changes have occurred in stock trading.
Conversion from fractions to decimals occurred in 2001.  All of a sudden, stocks were trading in 1 cent increments. This as a large part of why/how the bid-ask spread for stocks shrank. Before decimalization, the smallest allowed increment was 1/16th or 6.25 cents although most spreads were 1/8th or larger.
Regulation National Market System (aka Reg NMS)  which became effective in 2007. Between the “old” days and the “new days”, you could trade stocks on many different electronic systems aka ECN’s, SRO’s, exchanges. In round numbers there are about 50 different electronic exchanges/ECN’s where a stock can trade today. One of the main points of Reg NMS was to require all of the exchanges to have the same bid and ask prices. This is known as the National Best Bid and Offer aka NBBO.
Before Reg NMS became the law, it was common to have different bid/ask prices on different exchanges. For example, the IBM bid/ask on the NYSE would be 9.99/10.00. On say the Pacific Stock Exchange, it might be 10.05/10.06. You can see what happened. A company with fast technology could buy shares on the cheaper exchange and sell shares on the expensive exchange. In theory, Reg NMS ended this practice. The goal of reg NMS was to insure that investors got the best pricing, regardless of which exchange executed the trade.
Winners and losers
1) Winner- High Frequency Trading that is faster than everyone else for arbitrage. In the real world, it takes some amount of time for all 50 exchanges to match bid and ask prices. If you had the fastest system, you could constantly monitor prices on all of the different exchanges. When they got out of sync, which would occur every single time the NBBO changed, you could legally front run the change. This is one reason why having the latest, greatest, fastest technology is so important.
2) Winner- High Frequency Trading, automated front running detection. In the old days, you had humans attempting to front run large buyers and sellers. These days, you have sophisticated algorithms running on “co-located” computers attempting the do the same thing. Long story short, sell side programs attempt to hide/disguise their trading patterns so that nobody can tell they are trying to sell a large quantity. Buy side programs are constantly studying each and every trade, attempting to ascertain when the sellers are trying to move a large position. If your HFT algorithm is better than the next guys, you can pick up some easy money.
3) Winner- High Frequency Trading- news scanners. There are a number of sophisticated programs that are constantly scanning news sources and tweets looking for breaking news. For example, if you knew that Warren Buffet was going to buy XYZ, a few seconds before everyone else did, you could front run that stock. Once again, there are a lot of PHD’s working on algorithms to detect this and act on it. Think about the IBM Jeopardy computer guessing on stock news.
4) Winner- Average long term investor- Without a doubt, the bid ask spreads have come down, in addition to commissions. IF you are a long term investor, you win, but at the same time, over the long term this should not amount to much. So you save 12.5 cents per share, but hold the shares for 10 years. Big deal.
5) Winner- Day traders. Like it or not, trading in one cent increments, plus HFT makes this entire industry possible. It is clear cut that day trading is a losing proposition overall with a reported 90% of day traders losing money. For that lucky 10%, HFT makes the trading possible.
6) Winner- High Frequency Trading- “liquidity rebates.” Long story short, the exchanges pay the HFT’s for “providing liquidity.” The rules have many moving parts, but for example the NYSE pays 0.3 cents per share for providing liquidity. This is ONE of the primary ways that HFT’s make money. Obviously, the more shares they can trade while providing liquidity, the more they can make. To paraphrase Everett Dirksen, .3 cents here and .3 cents there and pretty soon you are talking real money. If you are interested, you might read this NYSE ARCA link 
7) Winner-Average investor- liquidity. Probably 99.9% of the time on 99% of the issues, an investor will always find a willing buyer and/or a willing seller at a reasonable bid-ask spread. This is the definition of a highly liquid market
8) Loser- Average investor in illiquid market. For the .1% of the time, like in the Flash Crash, the HFT programs are TURNED OFF. There are NO willing buyers and/or sellers at any price. In theory, this would not occur if the specialist system was still in place. In the October 1987 crash, NYSE issues remained fairly liquid. NASDAQ issues which are more similar to the HFT systems of today were a disaster. There were no buyers or sellers of last resort on many NASDAQ issues then.
9) Loser- Small investor psychology- Without a doubt, many small investors were terrified that the flash crash was even possible. They could NOT rationalize how the Dow could drop that far, that fast. Some percentage of investors have given up on equities due to fear.
10) Largest LOSER- NYSE specialists- The public does NOT understand this point at all. The last figures I saw, showed that NYSE specialists used to skim about $4 billion per year off of stock trades. Specialists have gone the way of the dinosaurs. They are now called DMM’s- Designated Market Makers. Their role in stock trading is dramatically reduced. Essentially all of the electronic stock trading is directly at their expense. My guess is that their $4 billion in skimming’s has directly gone to the HFT guys.
11) 100% Speculation Winner- HFT front running. I will use my favorite whipping boy Goldman Sachs as an example. Goldman’s computers know about every customer order that is placed. Technically it would be trivial to program Goldman’s’ HFT computers to front run customer orders. Recall that Goldman is simultaneously trading for their own account, in addition to handling customer trades. I and many others suspect the Chinese wall between proprietary trading and customer trading has a few holes in it. If Goldman skimmed a penny per share per trade it would add up. Goldman had several quarters in 2011 when their proprietary trading was profitable EVERY SINGLE DAY. I don’t care how many MENSA members you employee, I do NOT think this is possible unless you are front running or using some other nefarious trick. 100% profitable days is not exactly a Gaussian distribution.
12) 100% Speculation Winner- HFT traders that cause different exchanges to become out of sync. For example, if you could force the price in Chicago to be different from the price in New York, you could quickly make a lot of money. The small garage shop NANEX has laid out a convincing case that this occurred on the Flash Crash. Some of the exchanges got out of sync, causing all Hades to break loose. I and others suspect that some HFT’s attempt to cause this situation to occur on purpose. The theory is called “quote stuffing.” NANEX has documented how some HFT customers request 25,000 quotes PER SECOND!
BOTTOM LINE is that the individual investors are NOT harmed as much as commonly thought by High Frequency Trading. On the other hand, the NYSE specialists are now selling apples on the street corners for a living.
BTW, my apology for the length but this is the condensed version. I left out a lot of details for the sake of brevity.
 Wikipedia article on the 5/6/10 Flash Crash
 SEC- CFTC Report on the Flash Crash
 Yodaorange response to SEC-CFTC report “SEC lays an egg on Flash Crash”
 Yodaorange post after the Flash Crash “High Frequency Trading Tutorial:
 SEC Final decimalization rules
[6} SEC regulation NMS final rule
 NYSE- ARCA fees