why the model portfolio is overweigted to small caps

Once one has chosen which stocks to include in a portfolio, there are many ways to decide how much of each to include relative to the whole (i.e. the weighting of each stock). The three most common methods are as follows:

1. Market Capitalization. Each company has a market value of it's total equity, which is simply the number of its shares multiplied by the price of each share. Most stock indexes are set up such that the portion of the index allocated to each stock corresponds to that company's market cap relative to the total market cap of all the companies in the index. If one wishes to construct a portfolio that will closely mimic the performance of the index (e.g. S&P 500), then the stock allocations in the portfolio will also have to be based on market cap weightings.

For instance, one first adds up the market cap of all the companies in the portfolio (say this total is $100 billion). Then, each individual company's market cap is divided by that total in order to calculate the portion of the portfolio that should be allocated to each individual stock. So if a particular stock's market cap is $5 billion, then 5% of the portfolio would be allocated to that stock ($5 billion / $100 billion = 5%).

Some benefits to this approach are that it's simple to compute and one's portfolio will not significantly under-perform the chosen stock index. A drawback is that the portfolio performance will be most heavily influenced by the performance of the few stocks with the largest market caps. For instance, the top 10 companies in the S&P 500 account for roughly 20% of the total market cap of all the companies in the S&P 500. So, even if a portfolio holds all 500 stocks in the S&P 500, but the portfolio is weighted according to market cap, then 20% of the portfolio's performance will depend on the performance of those 10 largest companies.

2. Equal Weighting. The simplest method - the weight of each stock equals 1 / (# of stocks in the portfolio). So if one's portfolio holds 100 stocks, then each stock is ascribed a 1% weighting in the portfolio. If 200 stocks, then each stock is ascribed 0.5% weighting, etc.

A benefit to this approach is that one's individual stock risk is reduced. No need to worry about waking up one morning to find that the largest company in your portfolio has been falsifying their accounting statements, is declaring bankruptcy, and your portfolio just lost a large part of it's total value. The drawback is that, in comparison to the index, one's portfolio will be more allocated to smaller companies (i.e. 'small caps'). Therefore there will be times when one's portfolio will outperform the index and there will be times when the portfolio will under-perform the index, the latter of which, because we're all evolved with a sensitivity to relative status, will cause one to feel like a failure and question one's own convictions with respect to investing strategy. Aside from the emotional distress (assuming one isn't mentally immunized against it), these feelings might cause permanent under-performance if one capitulates and switches the portfolio to market cap weightings just before the small cap stocks subsequently outperform large-caps because of (think: Wizard of Oz voice) Reversion to the Mean.

Just FYI, historically speaking, in the long-term, small caps have outperformed large caps, but I don't believe this will necessarily always be the case (it's probably just a historical quirk). Better bet is that over long periods of time, large and small caps will realize equal performance.

3. Mean-Variance Optimization. Theoretically, this method should provide for the best performance. Essentially it attempts to weight stocks in the portfolio according to (i) how the price movements of each stock correlate with price movements of the other stocks and (ii) the expected long-term appreciation of each stock, the net result of which should provide for maximum investment returns for any chosen level of stability in the portfolio value (i.e. how much the value of the portfolio bounces around and gives you heartburn). Unfortunately, this method doesn't outperform the simple method of Equal Weighting. Personally, I think that's because the calculated stock weightings according to Mean-Variance Optimization are extremely sensitive to the assumed volatility and expected return of each stock. Since it's impossible to predict the actual returns of each individual stock, it's a matter of garbage in, garbage out.

Side Note: I do think that on average, the volatility of individual stocks tends not to change drastically over time (at least relative to the volatility of stocks in general). So a variation of this mathematically oriented methodology can be useful if one wishes to simply minimize portfolio volatility, or even dial in a certain level of portfolio volatility.

Conclusion. The stocks in the model portfolio were initially weighted by first grouping the stocks into economic sectors. Each sector was allocated somewhat according to Market Capitalization. For instance, the Vanguard Total World Stock Index (ticker: VT, which is one of our benchmarks) has about 14% of its portfolio allocated to companies making consumer goods, so therefore I allocated roughly 14% of the hypothetical money in the model portfolio to consumer goods companies. Now, for various reasons the sector weightings of the model portfolio don't exactly match up to all the sector weightings of VT (mainly because I did not want to include oil companies or banks), but the point is that the sector weightings of the benchmark were indeed a consideration when establishing the sector weightings of the model portfolio.

Lastly, within each sector, the individual stocks were Equal Weighted because (i) I don't want to have a significant allocation to any individual stock and bear the idiosyncratic risk and (ii) I don't want to do the data gathering and mathematics associated with mean-variance optimization when it doesn't work anyway.

You may have noticed I said this is how the model portfolio was initially allocated. Over time, as certain stocks have outperformed others, the weights necessarily drift. Next week, perhaps we'll cover re-balancing and the pros and cons thereof. Then, since re-balancing will provide a nice segue to market-timing, I think we may switch gears from talking about what to buy/sell and begin talking about when to buy/sell, which (especially in the short-term) is a much more important determinant of portfolio performance.