Q. How does Goose differ from performance attribution analysis?
The primary objective of Goose is prediction, whereas performance attribution is a backward-looking descriptive exercise that is silent on the role of skill vs luck. In other words, performance attribution analysis helps investors determine whether the manager’s past performance is something that they desire exposure to going forward, but not if they are likely to get it.
Q. What kind of data does Goose require?
Goose uses information about the mandate (universe and trading constraints), daily returns, daily positions, and market data for the universe. Goose also yields insights without position data, such as when evaluating prospective managers.
Q. What about new managers with little to no historical performance data?
One of the strengths of Goose is its ability to draw statistically significant conclusions in a short time period, and Goose will build a meaningful profile of a new manager as quickly as 3 months.
Q. Is Goose a complete black box like some machine-learning type approaches?
Not at all. Our approach is a scientific translation of how practitioners already think about performance, and we routinely discuss its logical flow with our clients.
Q. How do we know that Goose “works?”
The beauty of Goose is that it is model-independent and produces a set of insightful facts about a manager’s skill that are incontrovertible. Data-driven predictions based on these facts help institutions increase firm-wide returns through improved allocation decisions.
Q. What if Goose says a manager has no skill but we like the performance?
Goose often helps investors realize that the part of performance they like is from the mandate + market and not the manager’s skill / edge, in which case the investor should invest in our conception of a skill-less benchmark (a.k.a. “Montroid”).
Q. Which markets does Goose work for?
As of 2018Q1, Goose is able to work with equities, bonds, futures, and listed options, basically any price-based.