I've been looking into Ledoit-Wolf shrinkage but I've found the papers concentrate on large numbers of assets that tend to all be highly correlated. Often a universe of large cap stocks.
I'm concerned that portfolio with a mix of stocks and bonds ETFs the bonds may in steady state have a near zero correlation with the stocks and the constant correlation matrix would be a poor prior.
I note that Betterment uses shrinkage in their stock/bond asset allocation but they don't talk about their prior. Does there exist well reasoned work exploring good priors for diverse portfolios?
