Jul 26, 2012
Looking into the PM’s mind
Mark Kuzminskas, director of equity trading at Robeco Investment Management, discusses his aspiration of applying real-time analytics to the natural trading instincts of the portfolio manager.
The
TRADE USA: You’re coming to the end of a two-year project with S.J. Levinson on the use of their optimiser product, START [Strategic &
Tactical Analytic Research & Trading]. What were you setting out to achieve?
Kuzminskas: Five
years ago, we began moving from a high-touch capital-intensive trading style to
a more electronic trading format. We were being inundated by brokers wanting us
to take a look at their algorithms and implement their technology on our
desktop. At the time, I was most interested in understanding how various broker
algorithms functioned and performed under different market circumstances.
I began working
with a third-party technology vendor in Boston, with the idea that we would
develop a real-time dashboard to optimise our trading based on the analysis of
how these different broker algorithms functioned under various conditions. I
quickly realised that was a tremendous task given the amount of data that would
need to be analysed. Furthermore, when I started thinking about the way I was
using broker algorithms, I was moving in the direction of simplicity whereas
brokers seemed to be adding complexity to their algorithms. My view shifted
from a need to better understand off-the-shelf algos to a desire to design our
own where we controlled the routing logic. Unfortunately, due to business
considerations, we needed to abandon that project.
Subsequently, our
project with S.J. Levinson began as a generic transaction cost analysis exercise
which then migrated to trade optimisation when I realised they had already
developed the platform I had been seeking. I don’t have any firm statistics at
this point, but if the results are half as good as the back-testing indicates,
then it will be worthwhile.
What we are
trying to do is marry historical portfolio manager profiles with real-time
trading analytics in a way that optimises the transactions on behalf of each individual
portfolio manager.
The
TRADE USA: Are there any particular types of trades that you
expect to show the greatest improvement as a result of the use of the optimiser?
Kuzminskas: I
think the optimiser will improve performance on those trades where we have won
by a narrow margin or lost by a wide margin. The human psyche being what it is,
oftentimes we’ve cut the winners and let the losers run. In other words, we’ve
been a little bit too anxious to harvest positive results and not had enough
urgency on orders that are getting away. What the optimiser has shown us is that
it’s about letting the winners run and cutting the losers.
Now when I say
cutting the losers, it doesn’t mean the PM has made a bad investment decision;
it just means that the stock is getting away from us.
When you think
about trading in its simplest form, it is often a question of whether to step
on the gas pedal or hit the brakes and back off. It’s all about timing. The optimiser
helps us determine how fast we should go based on the PM’s historical decision-making
relative to the market characteristics at the time we are implementing the
trade decisions.
START helps us
marry that historical data to real time analytics. It can look at current
market conditions and determine whether we can afford to be patient and ease
off on the gas pedal or whether we need to be stepping more firmly on the
pedal.
The
TRADE USA: What inputs do you use to distinguish one investment
style from another?
Kuzminskas: The
PMs in our organisation generally have a value philosophy; that’s the
foundation of our investment thesis. However, each PM has a slightly different
DNA in the form of their trade timing.
You chart this out by looking at the data from their prior orders – how
the stock was trading prior to the PM placing the order, then during the
execution phase, and subsequently, post trade.
The
TRADE USA: What greater control over routing does the optimiser
give you and how important is that?
Kuzminskas: The
routing is very important. On a given day, we select which broker’s pipes we’re
going to use, but then we route using directed DMA. The optimiser will indicate
various aspects to each child order such as: size, exchange, limit, and lit or
dark venue. It continuously reassesses how to proceed based on the technical
data that we’re witnessing and the progress that we’re making on the order.
At the end of the
day, S.J. Levinson crunches all the information to be garnered from the day’s
trading, processes it, and updates the model for the next day, so it’s learning
all the time. That’s one of the unique aspects to this project as opposed to some
of the off-the-shelf type solutions. It’s customised in such a way that we are
learning on a daily basis.
The
TRADE USA: Given the lower volume, lower commission environment,
is that a factor in wanting to make execution processes as efficient as
possible?
Kuzminskas: A
reduction in commission is a benefit, but it’s not really a factor in this
instance. We already tackled that project years ago where we unbundled our
trading and research commissions and adopted the use of commission sharing arrangements and the like. That
exercise slashed our commission costs dramatically. Now we’re getting down to
the point where we’re rounding off the edges, as I like to think of it.
The driver is really
to match the implementation of investment process to what the portfolio manager
has in mind.
Best execution is
about getting the best price given the liquidity characteristic of the stock,
the market conditions, any news catalyst, and most importantly - the portfolio
manager’s objectives for the investment idea. The portfolio manager’s reasons for buying or selling the name at a given point in time shape our execution
strategy, so when you stop and think about it, the person best suited to
execute an order is the portfolio manager.
As such, we have spent the last decade trying to eliminate all the
traditional middlemen in the process and this effort accounts for the biggest
performance enhancement related to trade implementation.
What’s
fascinating to me about the optimiser is that it gets us closer to the reality
of the PM doing it all – deciding what to buy or sell and how to trade it. It
offers a way to fully engage with the portfolio manager’s objectives and think
about order flow in a new light.
Reporting by Richard Schwartz