[Tex/LaTex] Making advanced table. Centering and line above tabular environment, 3 columns

tables

I need some help creating the table below.
I only need the full sample and parameters columns, and therefore only three columns. However I have issues with centering, top line above the TABLE 3 and Caption. What i ran so far (now I only use 2 columns, so need to add 1). And yes, I want to fill the entire page..:

\documentclass{article}
\usepackage{booktabs,amsmath,caption,dcolumn}
\newcolumntype{d}[1]{D..{#1}}  
\begin{document}
\begin{table}[h]
\caption{\textbf {\textsc{Parameter estimates of the No-Arbitrage Yield only model}}}
\label{table:Parameter estimates of the No-Arbitrage Yield only model}
\begin{tabular*}{\textwidth}{@{}l@{\extracolsep{\fill}}d{4}d{4}d{4}@{}}%definerer antall kolonner
\toprule[0.5pt]
\toprule[0.5pt]
Parameters & \multicolumn{2}{c}{Full Sample} \\
\midrule[0.5pt]
\textbf{$\delta_0$}\\
& & \text{Factor autoregressive parameters}\\
$\rho_{LL}$& 1.000 & (0.342) \\
$\rho_{SS}$& 0.999 & (0.233)\\
                    & \\
        $\lambda_S^0$& &\\
        $\lambda_L^0$& &\\
        $\lambda_S^1$& 0.0912 &\\
        $\lambda_S^1$&-0.1612 &\\
        \\
        $\sigma_L$   &0.0340 &\\
        $\sigma_L$   &0.0795 &\\
  \hline \hline
\end{tabular*}
 \scriptsize
 Notes:\hfill\parbox[t]{14cm}{The table shows the results from the market model estimations. Each column presents the mean and standard deviation for all the companies' CAR in the designated event window period. The first column, (0, 2), reports $\mu$ and $\sigma$ for the event window period spanning from day 0 to 2 days after the dividend announcement (3 day window). Where day [0] is the dividend announcement day. Columns (2), (3) and (4) widens the event window to 5, 11 and 21 days. The t-values are reported below in parentheses and calculated as $\frac{\mu}{\sigma/\sqrt{n}}$. ***, ** and * denote significance at the 1, 5 and 10\% levels, respectively.}
\end{table}
\end{document}

Table

Best Answer

Here is a start using a combination of float (for styling the layout of the float with a top rule), tabularx (for easy setting of fixed-width tables), booktabs (for awesomeness) and caption (for formatting of the caption style):

enter image description here

\documentclass{article}
\usepackage{float}% http://ctan.org/pkg/float
\makeatletter
\newcommand\fs@topruled{\def\@fs@cfont{\bfseries}\let\@fs@capt\floatc@ruled
  \def\@fs@pre{\hrule height.8pt depth0pt \kern2pt}%
  \def\@fs@post{}%
  \def\@fs@mid{}%
  \let\@fs@iftopcapt\iftrue}
\makeatletter
\floatstyle{topruled}
\restylefloat{table}

\usepackage{tabularx,booktabs,caption}% http://ctan.org/pkg/{tabularx,booktabs,caption}
\captionsetup[table]{
  labelsep = newline,
  textfont = sc, 
  name = TABLE, 
  skip = \medskipamount}
\begin{document}

\begin{table}[h]
  \caption{Parameter estimates of the No-Arbitrage Yield only model}
  \label{tab:parameter_estimates}
  \begin{tabularx}{\linewidth}{@{} X r r @{}}
    \toprule[0.5pt]
    \toprule[0.5pt]
    \small Parameters & \multicolumn{2}{c}{\small Full Sample} \\
    \midrule[0.5pt]
    $\delta_0$ & \multicolumn{2}{c}{6.2030} \\
    & \multicolumn{2}{c}{Factor autoregressive parameters} \\
    $\rho_{LL}$ & $1.000$ & $(0.342)$ \\
    $\rho_{SS}$ & $0.999$ & $(0.233)$ \\
    & \multicolumn{2}{c}{Risk pricing factors} \\
    $\lambda_S^0$ & \multicolumn{2}{c}{$0$} \\
    $\lambda_L^0$ & $-0.0174$ & $(0.041)$ \\
    $\lambda_S^1$ & $ 0.0912$ & $(0.089)$ \\
    $\lambda_S^1$ & $-0.1612$ & $(0.034)$ \\
    & \multicolumn{2}{c}{Factor shock volatility parameters} \\
    $\sigma_L$ & $0.0340$ & $(0.019)$ \\
    $\sigma_L$ & $0.0795$ & $(0.010)$ \\
    \bottomrule
  \end{tabularx}

  \medskip

  \parbox{\linewidth}{\scriptsize%
  \textsc{Note}:
  The table shows the results from the market model estimations. Each column presents the mean and standard deviation for 
  all the companies' CAR in the designated event window period. The first column, (0, 2), reports~$\mu$ and~$\sigma$ for 
  the event window period spanning from day~0 to~2 days after the dividend announcement (3~day window). Where day~[0] is 
  the dividend announcement day. Columns~(2),~(3) and~(4) widens the event window to~5,~11 and~21 days. The $t$-values are 
  reported below in parentheses and calculated as $\mu / (\sigma / \sqrt{n})$. ***, ** and * denote significance at 
  the~1,~5 and~10\% levels, respectively.}
\end{table}
\end{document}
Related Question