Once you've chosen the ingredients and know why the next step is to start pairing them. In asset allocation, the industry and academia have been arguing about how to portfolio investment targets. At first, according to Professor Markowitz's theory, no matter how many assets you have, given a goal, you can always find the optimal solution of a collocation mathematically. According to the mean-variance portfolio theory, we need to predict the future performance of the asset target according to its past returns so as to obtain the input value of the portfolio model.

However, in practical application, it is challenging and unstable to predict the future returns of certain assets through models. So later, based on the mean-variance portfolio theory, a portfolio model such as black letterman, which does not need to be predicted, was derived. But most investors still have to do a series of very complex model calculations.

In the past years, according to the development of science and technology, there have been a large number of robot investors in the market, also known as intelligent investors, such as wealth front and betterment, which basically use Markowitz theory (or its derivative and improved models, such as B-L) to construct the essential asset allocation portfolio, and then select the optimal allocation under the given risk conditions according to the risk preference of customers. Through technological means, these companies provide customers with very convenient solutions so that a considerable number of investors enjoy the treatment that only private customers could have before.

Robot investment definitely brings better customer experience, but from the perspective of the model, the asset allocation model obtained through very complex calculations is not necessarily better than the result of a reasonable simple matching model.

Some experts have tested various very complex asset allocation models and compared them with the average weight model. He found that complex models, whether based on historical or predictive data, did not perform better than average weighted models after risk adjustment; Moreover, in out-of-sample tests, the additional benefits of complex models are often offset by problematic prediction errors.

Therefore, complex models do not necessarily lead to good results.

The historical rewards of the Masters

We look at these nine well-known asset allocation models. Although they are all written by masters, they do not excessively seek the mathematically optimal solution of the ratio, basically select a few assets, and refine the classics. Daomingge said: simple but not straightforward.

So how have these models, based on the wisdom and experience of masters, performed in history, and can they really bring the benefits of asset allocation? The models were fully replicated, and historical backtesting was done with data from 1973 to 2017 (monthly rebalancing, gross income, without considering any fees, market shocks, and taxes).

After the above-detailed analysis, we can actually reliably sum up some truth:

Complex models don't necessarily mean good models. Simplicity, not simplicity, is the truth.

You'd better think twice before you make the decision to invest because you need to

manage your resources well to get profits.

（Writer：Lany）