Tuesday, September 04, 2007

How Large Traders Disguise Their Presence in the Stock Market

I received a note from BZB Trader after he had posted interesting information to his blog. He had collected data to better understand order flow and had observed two things:

1) Retail traders account for a small proportion of total volume;

2) Large traders are heavily using order execution software to break their large orders into small pieces.

I performed a very simple exercise and examined the first hour of trading in a popular stock, AAPL. There were almost 57,000 trades in the first hour alone.

By my calculation of how the trades were reported, roughly two-thirds of these trades were broken down into small pieces for execution. Over three-quarters of all first-hour trades were 100 shares. Of the small, 100-share trades, I estimate that about two-thirds were part of larger trades that were executed in pieces by specialized software.

In other words, if you were to simply look at trades and trade volume, you'd conclude that small traders were dominating the marketplace. The reality is, however, that large traders continue to hold sway, but have succeeded in disguising their presence.

For the trader interested in determining who is in the market and what they're doing, it is either necessary to re-aggregate the trade data by a conceptual reverse-engineering--a difficult, computationally-intensive endeavor not feasible for most traders--or it becomes necessary to make inferences on the basis of larger time units (1 minute data, etc.) that naturally aggregate the trades.

More on this latter strategy soon to come. Meanwhile, consider how the disaggregation of large trades increases trade volume on the stock exchanges, increases the bandwidth necessary to process market information, and potentially distorts such measures as NYSE TICK, Market Delta, and money flow.

RELEVANT POSTS:

How to Track the Stock Market's Large Traders

What Every Short-Term Trader Should Know
.

14 comments:

profste said...

hi,

to recognize large traders activity i found relatively useful practicing in tape reading.

after hours spent in front of the tape (i.e. the Time and Sales flow without looking level II that i suppose is the real "disguising device") i am now able to often recognize when automatized trading kicks in and the machines take over the most part of the action (especially in eurex and cme eminis).

it's a completely discretional method that i could not really explain, and i understand it's difficult to apply for those who can't trade full time.

correct me if i'm wrong, but i guess that tape reading is a mental process that tools like market delta and similar try to replicate in formal ways for an ex-post analisys or for part time traders that cannot accumulate all the screen time needed for the tape to begin to make sense.

i look forward for your additional post on the subject, and as always thanks for the knowledge you share,

regards, stefano

F-Trader said...

What resources would you recommend to learn how to "re-aggregate the trade data by a conceptual reverse-engineering"?

robin said...

Hello Brett,
I have noticed times when the markets stay within a range for long periods during the day. Many have labeled this as accumulation or distribution. Is this caused from the large traders buying or selling in smaller chunks so they do not drive the markets and take money out of their own pockets? Perhaps there is a way to disect this phase to anticipate a breakout or breakdown from this area? Thanks.

SerpentMage said...

Please don't take this as an offence. But tell me something that I didn't know.

My broker has something called the iceberg order. An iceberg order is when you want to disguise your trade to much smaller (like an iceberg). They recommend this trade when you are moving large numbers.

That's why I have rejected looking at volume for quite a while now. The actual numbers are posted AFTER the fact and thus completely useless to you AFTER the fact.

AgeKay said...

Hi Brett,

It's great you finally talk about this topic.
I have wondered for some time now why you place so much importance on tracking the "big players" by filtering small contracts since most volume is generated by large traders anyway and most of the little volume left is probably also large traders that just try to hide that they are large. And who cares anyway if a move is caused by large or small traders? Is a price change not valid just because small traders were responsible for it? Why are large traders supposed to be smarter than smaller traders? Since there is a winner and loser for each traded contract and large traders make up most of the volume, about 50% of the large traders are losers. 90% of small traders might be losers in this game but considering how little volume they are responsible for, you can conclude that almost half of all large traders are also losing money on average.

Please don't take this as criticism, I love your blog and I have learned a lot from reading it, but this is something that didn't make sense to me. Please enlighten me. Thanks!

Brett Steenbarger, Ph.D. said...

Hi Stefano,

Yes, I think you're right: many of the tools used by traders are attempts to formalize tape reading. I do think it's possible for short term traders to get a feel for order flow and *who* is participating in markets. Thanks for your observations--

Brett

Brett Steenbarger, Ph.D. said...

Hi F-Trader,

I don't think most traders have the computing power to examine stocks trade by trade to reverse engineer the disaggregation process. That's why I'll be suggesting an alternative.

Brett

Brett Steenbarger, Ph.D. said...

Hi Robin,

Good question; thanks. I do think volume data is helpful in gauging whether demand or supply is rising as you get toward the extremes of ranges, but I'm not sure that the disaggregation of trades contributes to range bound trade. The same volume is being transacted; it's just broken into more pieces.

Brett

Brett Steenbarger, Ph.D. said...

Hi SerpentMage,

What I found surprising from my analysis was the extent to which volume is dominated by such disaggregated trade. It's not clear to me, however, that this makes volume data useless. Once you aggregate the volume, you can make inferences about expansion or contraction of buying or selling interest. I'll be illustrating that in an upcoming post--

Brett

Brett Steenbarger, Ph.D. said...

Hi AgeKay,

Where I find volume helpful is in gauging whether demand or supply is expanding as we approach critical price levels (support, resistance, range extremes, etc). Because large institutions are capable of moving markets, I find their behavior at those critical price levels to be of great interest. They do their best to disguise their activity, however!

Brett

dk said...

Nice post. I went to a dinner party a few weeks ago and was seated next to a woman who writes the type of "disguising" software you mentioned. Her clients are a group of trading houses here in LA.

Their traders are evaluated on whether their trades move the price up or down when executed (trading inside the bid/ask). Her software breaks the trades into chunks (usually 100 shares), then places them into the order flow according to many dynamic variables determined by the trader. The software then tracks the efficacy of the "no price move" mandate.

Hiding trading activity is quite the rage, and gobbling or dumping shares without moving price is the coin of the realm. Bonuses are based in part on the results of her tracking software....best as always, dk

Brett Steenbarger, Ph.D. said...

Hi DK,

Thanks so much for that insight; it very much fits with what I'm observing in the data.

Brett

frankw said...

Brett,

I found this and related discussions fascinating. I was wondering how you accounts for trends, e.g. using daily charts, trends typically run for days up to weeks. Does this mean that there are other factors/forces at work? Using the last week as an example, a lot of inventory was moved and a trend was definitly created. Without FED intervention, it may last for weeks or continue down for a few more days then move up, possibly very rapidly. Using a few examples, 9/11 and 10/87 were panics with quite rapid downturns and varying upturns, while 2000 was a slow down with a slow to modest recovery. Bubbles are probably unique in their own right.

We definitly have trending and non-trending markets. When there is a trend and it lasts for a period of time, how big is the player or how many of them have to aggrigate to cause the moves we see? From what I observe it woulod appear that when a lot of big or big and small players have to move quickly there are big down moves, and when there is time, they can take days/weeks. If a large hedgefund or mutual fund had to move 10,000,000 shares in XYZ, using disintegration software, how long might this take? I would expect weeks, so the smaller orders could be interspersed so as to not disrupt things very much.

How much would you guess the big funds settling up the quarter/year by 10/31 have been affecting the last week/weeks?

Brett Steenbarger, Ph.D. said...

Hi Frank,

Good question. Institutions vary greatly by their time frames: there are hedge funds that trade very short time frames, and there are mutual funds that invest over long time horizons. Ultimately it's liquidity that determines trends, and it's the participation of longer-timeframe traders and investors that sustain a trend.

Brett