Sub-Par Generic (VWAP, TWAP) Algorithms

We asked head and senior traders at leading asset managers about their experience with generic broker algorithms in the course of a survey for our Market Intelligence Service members.
Sixty-two respondents provided a wealth of feedback.
The results were not good for Broker X, one of our members.
Forty-seven respondents indicated they used Broker X’s generic algorithms and those of 27 other brokers.

They independently rated each broker’s generic algorithms as  “Excellent,” “Very Good,” “Good,” “Moderate,” or “Poor.”

We translated this information into a scale.  The results looked like this:

Head and senior traders thought, on the whole, that the quality of broker-provided generic algorithms was “Good.” The horizontal red line at 1.97 in the figure above shows the industry-wide average rating for generic algorithms.

Some brokers’ generic algorithms better than others.  Some were worse.  Others were “in the middle.  Distinctive clusters emerged from response information.

Users were clearly less impressed Broker X’s generic algo performance. Broker X’s generic algorithms were rated slightly better than “Moderate” with an average rating of 1.38.  In contrast, the industry-wide average of “Good” at 1.97.
To bring out additional color, we compared the distribution responses on Broker X’s generic suite to those of a meaningful comparison group, in this case Tier 1 brokers.  See the figure below.
The distribution provides greater insight into the market’s dissatisfaction with Broker X’s generic algo suite.  80% of pertinent respondents thought Tier 1 broker generic algo suites “Good” or better. Only 50% considered for Broker X’s generic algo in this category.  20% of relevant respondents considered Tier 1 broker generic suites “Moderate” or “Poor” compared with 50% for Broker X’s suite.  One-in-four respondents considered Broker X’s generic suite outright worse than Tier 1 broker competition, rating it “Poor.” This same group did not rate place any Tier 1 generic algorithm suites in that category. 

We relayed our findings to our member, Broker X, who took action.  They had acquired scattered, anecdotal information over the prior two years about client dissatisfaction with their generic algorithm suite.  However, the feedback had been acquired informally though sales team response.  The data had not been collected in a structured, organized manner an therefore had not registered with management as an issue.

Broker X management has focused on correcting the issue.  We have provided them with additional cuts of data and broader pertinent information by user demographic.  We, of course, provided Broker X management with the names of competition in the above two figures.   Work is now in progress to address issues with the algorithm and related development, testing, and user experience.

Leave Comment

Your email address will not be published. Required fields are marked *