… Returns are assumed to be arithmetic: r t , i = (p t , i - p t -1, i )/p t -1, i = p t , i /p t -1, i - 1. The
portfolio return at time t is thus … Let d i,j,t denote the loss differential between two models by …

… Exploiting the same basic idea in a different way, Raffinot [2017a] proposes a hierarchical …
The ”Hierarchical 1/N” proposed in Raffinot [2017a] finds a diversified weighting by distributing …

… c t r t ′ r t 1 + δ − 1 r t ′ r t r t (10) where we have used the Sherman–Morrison identity in
the second equality. In each line, c t is the constant that scales the weights at time t. Clearly, in …

… the portfolio (shifting allocation weights) based on the stage of the growth cycle (Raffinot…
The best split s, at node t maximizes this splitting criterion ∆i(s|t), based on the Gini criterion in …

… We can’t take into account the flash estimate, because this estimation is too recent. Over
2000–2002, it turns out that the three series are relatively close. The ‘first estimate’ possesses a …

Interest rates are unreliable indicators of appropriate monetary policy; low nominal rates do
not indicate easy money. This paper attempts to assess the stance of monetary policy without …

… product GDP is used to develop a monthly indicator of GDP 5In unreported results, samples
through month T − 18 and T − 24 have been tested and lead to almost the same results. …

We propose a linearization of rule-based algorithms that reveals the most important interactions
between characteristics and macroeconomic variables when explaining future stock …

… To refine the chronology, the monthly GDP introduced by Raffinot (2007)2 is exploited in this
article. For instance, the monthly GDP allows to select the month with the lowest level within …

We propose a linearization of rule-based algorithms that reveals the most important interactions
between characteristics and macroeconomic variables when explaining future stock …