Our system can invest better, because it is able to take a client's whole financial picture into consideration - not just the portion of their assets we happen to be holding.
Typically, a financial advisor (automated or human) will ascertain a client's personal situation and recommend an appropriate portfolio. There are many interesting dimensions to this decision (liquidity, tax, personal preferences, etc.), but the most important one is risk/reward preferences: to which extend is the client willing to risk losses in order to improve (on average) his or her returns?
Those risk/reward preferences are then typically translated into a target allocation, e.g. "20% foreign stocks, 30% US stocks, 50% bonds". For example, the more risk a client can take, the higher the % of stocks will be in the target allocation (vs. bonds). This is because stock prices vary more than bond prices, but stocks go up more than bonds on average. [Note: this is just one particular way to do this translation. It is the industry standard because it is simple for a human advisor to implement. However, computers do not have this problem. We have specific plans to generalize this soon.]
The target allocation is meant to match a client's entire holdings. This is fine if an advisor manages all of a client's assets. However, this is rarely the case, due to:
We believe our system is the only one that can incorporate external assets in a way that is both:
This is a simple example for purposes of demonstration.
If we simply ignore the external VIG holdings, the portion of assets that we hold might closely track the asset class mix, but the customer's total portfolio will be quite overweight in the high-dividend US stocks asset class, because of the externally held VIG.
The key difference is most apparent on the 'Asset Tracking' tab, which shows how much we are holding above an asset class's target, as a fraction of the total assets (including external).
When we ignore external assets, the portfolio tracking is poor; we are roughly 7% overweight ($75,000 / $1,075,000) in high-dividend US stocks because of the ignored external VIG position. You can see this in the light green/blue line for high-dividend US stocks (labeled 'div' in the chart's legend), which hovers around 7%. As a result, we are underweight in all other asset classes, all of whose lines show below the black horizontal line that denotes 'right on target'.
When we include external assets, the tracking is close to ideal, since we buy less VIG to compensate for the external holdings. Of course, tracking is not perfect, but that is expected, as we do not rebalance every day, and the prices of different ETFs do not move in unison. The lines for the different ETFs are roughly centered on 0; they are not consistenly below 0 like in the 'ignore' case. This means that, on average, each asset class tracks its target.
[Note: If the external VIG position were so large that we could not simply buy less VIG to compensate, we would buy less of other stock-like ETFs. The process of knowing which ETFs can compensate for each other is discovered statistically from out-of-sample training data. This is the subject of another example.]
The improved tracking can also be clearly seen in the Asset % bar', which shows the breakdown of asset classes held as a % of the total holdings. The case where we include external assets shows every asset class at almost exactly the desired target. When we ignore external assets, note how the high-dividend US stocks 'band' is fatter than before, meaning we hold more than we should.
It is not meaningful to suggest a target allocation to a client and then only apply it to a portion of their assets. Our system has the ability to look at a client's whole financial picture and consider portfolio risk holistically.