Thursday, December 8, 2016

The indexing effect on bot trades

What many of us did not realize at the moment was the probability that the buy programs may have been triggered by events in another sector. Just before the buy programs kicked in, the rally in the Dow Transports was shifting up a gear as the index was on the verge of punching through to a new record high. If the Transports made a new record high, it would confirm the record high in the Industrials, thus giving a Dow Theory buy signal.As that realization spread, the algorithms kicked in with buy program after buy program and the race was on.The series of buy programs moved the indices up in a nearly a vertical manner until about 2:00 when the buying slowed a bit and stocks consolidated near the highs over the final two hours.The program buying was broad in addition to its intensity. They produced a whopping 422 new 52 week highs. Advances swamped declines better than 3 to 1. Volume was okay but a bit light for such a huge rally.
There is an aggregation delay in spreading information because the theories are index based.  So this sector by sector set of prices adjust, as aggregate variables in a theory.  The result is a networking bottleneck,  many trades are rrquired to set the indices such that they read any theory.

Trading pits change this process.

In the middle of a shift like this, the pits would just spawn and send the bots off to trader the leading stocks by volume.  The indexing effect would be short circuited, and indices would adjust after the new pits had relieved the congestion, and unspawned.

No comments: