How does this work in practice ?
Q² continuously gives each stock in its universe of approximately 4000 stocks a valuation figure, a score reflecting a stock’s over- or undervaluation. We use this output as a basis for our investment decisions. In- and out-of-sample back testing has shown that the stocks favoured by Q² on average outperform the market, and vice-versa. This last point is important because it helps address some issues relating to survivorship-bias.
Simply following the model blindly however is not very wise. The market processes an enormous tide of information daily, and compels us to do likewise. Actual portfolio selection will consist of understanding what facts (if any) led to a certain candidate stock’s undervaluation, and judging whether the market’s reasoning is sound. The ‘if any’ above is significant, because sometimes stocks do tend to just ‘random-walk’ their way into an undervalued position, although more often than not plenty of reasons can be found to explain the negative attitude of the markets.
The models are run on a daily basis, which means all decisions are constantly monitored. This allows us to avoid most classic pitfalls of quantitative investing, and makes it a lot easier to follow classic risk management guidelines.
The human decision-making porcess is fallible, and emotions (fear, greed) provide very often for sub-optimal decisions. Quantitative management resolves this, not so much by leaving the decisions to a computer, but because of the discipline that it imposes on us.