Please Note: Blog posts are not selected, edited or screened by Seeking Alpha editors.

Cheetahs, High Frequency Trading & Risk Management

By now most of us have been exposed to electronic trading, market rally & dips and to some degree high frequency trading (“HFT”).  Earlier this year, CFTC Commissioner Bart Chilton, in an Amsterdam speech, referred to HFT as the “cheetahs” of trading.  He went onto explain that although they add liquidity, they can be dangerous.  Clearly, HFT is here to stay and it brings with it some headaches.  Anyone of us who have executed a trade and a few minutes later end up getting stopped (or stomped) out has probably wondered if our “road kill” was the victim of HFT.  Add into the mix the complexity of “flash crash”, endless bandwidth, lots of capital with short term objectives, and it is easy to understand that these “cheetahs” are here to stay. 

 

Not to carry the analogy too far - but in nature, cheetah mortality mostly occurs when they are cubs.  Similarly, HFT mathematical algorithms either prove themselves early or die on the quant’s spreadsheet.  As such it is theoritcally possible that a lot of HFT models have not been vetted for all market conditions.  Even more interesting is that they seem to work in a consistent way.  That isn’t to say that the individual math employed within is not unique, rather that HFT’ers eat the same thing and have the same objective – making money.  We can imagine a computer being served a diet consisting of large bandwidth, excellent helpings of data with robust underlying, and sprinkling of resting bid/ask.  Yes, the math within the machine may be unique but the starting point and the objectives remain the same.  Herein lays the risk. 

 

Before we deduce anything about HFT, I would like to re-visit the importance of thinking differently.  As I have said before, concentration risk is the only way to pass the Joneses – making wealth involves a winning trade. (Unfortunately, the opposite is true.)  The best thing about a portfolio is maintaining the status quo – you and everybody else make money and lose money together.  There is safety in being part of the herd – no one can really blame you – call it job security.  On the other hand, this is also why it is important NOT to read all the news, as reading all the news gives you a portfolio of opinions, clouds the way you think – your mind becomes the efficient market.   Systemic risk is a form of concentration risk that really does not offer any hedge opportunities – the “beta” or “alpha” earned carries with it the pollution of systemic risk.  A market place of buyers and sellers offers a portfolio of thought which may or may not be “systemic”.  Tying back to the Serengeti, a market place full of cheetahs introduces its own risk & consequence.  Understanding the hunt & prowl of the cheetah should help in improving risk management. It could be that herds of investors continue to stampede as the cheetah of HFT trims off what it can.   Perhaps the herd gets smaller, perhaps it figures out how to manage ahead.

Considering all, I think we can understand a few perimeters in the HFT world:

  •  HFT requires data and deep markets.  Generally finite detail, mostly likely “live” data.
  • Data will be analyzed in measurable intervals, bins, seconds, minutes, etc
  • Peripheral data is probably collected.  It can be arbitrage data, beta data, or “sniffers” that size up underlying bid ask.  Peripheral data may include volatility, competitors or hedge & arbitrage targets.
  • Relationships & assumptions are established between target and peripheral information – assumptions probably include convertibility, mean regression, etc.
  • If the HFT is programmed to be long, it will buy more than it sells. (and vice versa)
  • At the on-start and at the smallest time interval, the HFT position equals the moving average of that time interval.  As time proceeds, the average price of the HFT position deviates from the moving average of a specific time interval. 
  • Profit/loss points are required to manage & avoid “gambler’s ruin”.
  • Market gaps or pockets of illiquidity are problematic, hence deep market requirement.

 

On the face value, we do not know if HFT’s approach is from the long side or from the short side.  However, we do know that given “gambler’s ruin”, the position has to be cleared out.  Basically, gambler’s ruin is a statistical study that explains the expression “the market can be irrational longer than you can be solvent”.  In other words, Mr. Market has more money and capital than any HFT - eventually the position has to be exited. 

 

Keeping in mind the money making objective - within a programmed HFT, trading rules would be imposed to help achieve profit and loss objectives. Some of these rules could be as simple as saying that if the HFT average position’s price within a specific interval equals the moving average of another time interval; the position would be held or allowed to grow.  It stands to reason, given a HFT program based off moving averages, that any significant departure in the weights would trigger an action.

 

As a risk manager, the challenge is to manage the “group think” within the HFT machine, within the programming, all of which gives rise to crazy “non-fundamental” behavior (flash crashes, wild price swings, etc).  Operations research, close analysis, and consistent attentiveness are part of the tools deployed to manage concentrated model risk.   Herein lays the opportunity.  Herein lays the risk management.

 

I supposed the good news is that group think (modeling or otherwise) can be understood and anticipated, enabling the wise trader unique profit opportunities.  Patience is a virtue - similar to fishing. 

Thank you for reading my newsletter. Good day and good weekend!