The Salient Active Quants process (as used in the Salient SA Hedge Fund) aims to generate alpha through the simultaneous exploitation of a number of underlying (i) behavioural errors using (ii) powerful information processing techniques.
(i) Behavioural errors
Examples of behavioural biases include the ‘herding effect’ which results in medium-term over-reaction (the 'momentum' effect) and a corresponding long-term mean reversion (the 'value' effect) as the effects of these emotional extremes are unwound over time. The ‘increase in trading volume’ effect also has similar behavioural origins. The ‘anchoring’ heuristic plays role in a number of short-term effects such as that of short-term price reversal, the forecast revision effect, the broker downgrade effect, the earnings surprise effect (although there is some evidence that this can last into the medium term) inter alia. Ambiguity aversion leads to neglected firm and size effects.
(ii) Information Processing
Our alpha generation capability is based on the simultaneous exploitation of a number of the above empirically identified market imperfections. The multi-factor quantitative model provides exploitable stock-picking signals in the form of a table of all stocks in the relevant universe ranked by expected performance over the next month. Aside from shorter-term reactions to relevant newsflow, the portfolios are typically systematically re-evaluated and, where necessary, rebalanced on a monthly basis.
We complement the above quantitative signals with a brief qualitative overlay or “sensibleness check” that (i) assesses the validity of the inputs into the model (e.g. incorrect data or ‘quality of earnings’ issues) and (ii) considers whether there is relevant information that the model is not aware of (e.g. corporate actions and other recent news-flow). The approach, although quantitative, is certainly not a ‘black-box’. We do override the model in a minority of cases.