Statistical identification with hidden Markov models of large order splitting strategies in an equity market

Posted: March 30, 2010 in Uncategorized
Tags: , , , ,

I found this jewel of a research paper today: “Statistical identifi cation with hidden Markov modelsof large order splitting strategies in an equity market” and I think that I am in love!  You can download the paper here : 1003.2981v1.

Basically it works the age old question of is there a way of telling if the market is going to mean reverting or directional over some foreseeable trading period.  In the case of this article, you may consider the timeline to be the rest of the trading day after the observation of the opening trades.  The premise is that you develop a model of how institution trading models break large execution sizes into fragments thru some trading algorithm such as Volume Weighted Average Price (VWAP) thru hidden Markov models (HMM) and then once calibrated it would allow the user to detect the future actions before their potential participation happens.

Please read away and ask questions!  I am always up for a great discussion involving trading ideas.

  1. aix says:

    Thanks for posting this. The paper looks very interesting.

    In the markets I am familiar with (US/European equities), it is the norm for the public order book feed to be anonymous. In other words, it is not known which participant submitted which order.

    This makes a direct application of this method a bit difficult.

    The overall approach is interesting though. I would definitely be up for discussing this further if there’s interest…

    • autospreader says:

      Your very welcome; I agree with your statement about an anonymous order book in the domestic and international equities market places. It is my belief that while you won’t know the specific participants, we can group certain participants based upon thresholds for increases/decreases/trades in the overall trading quantity being presented to the marketplace. Furthermore, I believe that with some additional modelling, we could get this to the point of a threshold of those orders whose market impact is an expected amount.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s