Flashbacks of the computer being frozen for an hour or so are stopping me from testing which of these choices to avoid and recommend. Best to keep the production function choices to the cobb-douglas specification it defaults to in each case and keep slider changes small and slow.Abs choices cobb defaults douglas production slider Slow small specification
What counts as problematic in my case is when there are fitting problems or when the selection criteria are negative.
In the latter case we deleted 4 funds across ALL models because they had model fitting issues across SOME models.aic designated funds negative noted problematic retained SBC values wrong
A way to store regression results for multiple models and/or variables that I have come across involves the use of nested list objects. The easiest way to visualise nested list objects is to imagine them as folders in windows explorer, with each subfolder (or item contained therein) being separated by the $-sign.base dataframe dataframe dataframes element fit models multiple nested object results store variable variety
I have grown a bit tired of R and the blind ritual of haphazardly replicating things here and there.
The only issue here worth mentioning is that the hlt.row / hlt.col arguments are swapped.If one wants to highlight rows (cols) then hlt.col (hlt.row) should be specified.blame bothered hlt made moment object quick quirk returns summarise table
The previous post ended with the creation of a matrix that contains an unlisted version of the selection criteria for each fund & model combination.
Interpretation of the selection criteria is straightforward : the higher the value the better the model.aic data differences frames fund ggplot inputs manipulate model object significant
Once we have fitted each of the nine models to the data for each of the 13 funds in the edhec dataset,let’s summarise this information in a new list object by saving across models.
The purpose of this step in the process is to summarise the data contained in the edhec.ret list object above into something that can be conveniently used in ggplot functions.continues data deleted funds identified model post problematic Subsequently temp
In my attempt to replicate some of the methodologies of the Kritzman paper, I stumbled across an article by Hardy (2001) which I thought I might try to tinker with as well. Provide a summary of the selection criteria (used to rank fitted results across models within and across fund returns).column criteria data fund highlight model object problematic provide results row table values
The final task we have set ourselves, and which also happens to be a welcome initiation to the excellent ggplot2 package for me, is concerned with summarising the previously saved plots and tables into a single (and hopefully useful) dashboard.
The Inflation economic regime variable begins in the normal state,with a high probability of remaining in this state (Persistence : 90.1%) and a low probability of transitioning to the event state (Transition : 9.9%).The estimated mean and sigma for this normal state are 0.2559 and 0.275 respectively.fund global higher macro normal regime state states the values variable
Standard-deviation-Scaled difference across state means as per the paper) Cumulative returns and drawdowns when states are known versus unknown Event-state mean vs non-event-state mean for the chosen fund across all economic regimes.
To give a sense of the output that this function returns.chosen chosen fund code data economic fund map multiplied regime simply variable
Fit a two state hidden markov model to economic regime variables (Equity turbulence,Currency Turbulence,Inflation,Economic Growth). For each regime variable, overlay the resulting probability state map on a time series plot of the regime variable and save final graph as an element of a list.aligned dashboard economic element fund lapply map object Overlay pass regime variable regime variables regimes resulting state