I'm not sure how these arrays are shaped or what the contents are, but I'm guessing you can do something like this:
mean(vertcat(colData{:}))
to collect the multiple outputs of coldata{:} into a single vector.
If that's the case, it would be more efficient to avoid replicating the operation:
stock_priceStats = table();
stock_priceStatsVars = stockprice_data.Properties.VariableNames;
for i = 1:width(stockprice_data)
colData = stockprice_data.(i);
colData = vertcat(colData{:});
stock_priceStats.(i) = mean(colData);
stock_priceStats.(i) = std(colData);
stock_priceStats.(i) = min(colData);
stock_priceStats.(i) = quantile(colData, .25);
stock_priceStats.(i) = median(colData);
stock_priceStats.(i) = quantile(colData, .75);
stock_priceStats.(i) = max(colData);