1. What is the Shortfall Model?
The Shortfall Model generates pre-trade estimates of the likely impact of a proposed trade on the price of an asset. Consider, for example, a proposed trade to buy 50,000 shares XYZ where the pre-trade price of XYZ is $25.00. The Shortfall Model estimates the likely impact of the proposed trade to be 2 cents per share (8 basis points). Therefore, the expected price the buyer will pay is $25.02. Currently, the Shortfall Model
covers the S&P 1500, Russell 3000 and ETFs.
2. Why use the Shortfall Model in portfolio rebalancing?
Using the Shortfall Model in a portfolio rebalancing optimization incorporates the estimated cost of trading into the proposed portfolio. The Shortfall Model reflects the fact that trading more of an asset drives the average execution price against the investor. The more quantity traded, the more the execution price moves. Utilizing the Shortfall Model in the objective of the portfolio rebalancing problem takes into account the tradeoff between the investor’s objective, such as expected return, and the degradation in the objective because of trading impact.
3. How is the Goldman Sachs Shortfall Model different?
The Shortfall Model is based on Goldman Sachs clients’ past trade executions and, therefore, reflects how much it actually cost to execute similar trades in the past. Other models, derived from tick data, only provide a hypothetical cost and a theoretical model of trading.
4. How does Goldman Sachs define the shortfall trading cost?
For buy orders, Goldman Sachs defines shortfall as the execution price minus the prevailing midquote when the trader received the order (strike price) as a percent of the strike price, in basis points. For sell orders, shortfall is the strike price minus execution price as a percent of the strike price. Commissions are excluded.
5. What is the structure of the Shortfall Model?
The model consists of non-linear regressions of the actual shortfall of past client trade executions on seven factors:
1. Executed trade size ($ value).
2. The trade’s execution horizon.
3. Market capitalization of the stock (large-cap >$7.5 bn, mid-cap, small-cap <$1 bn).
4. Listing venue of the stock (NYSE or NASDAQ).
5. Average bid-ask quoted spread of the stock (bps) over execution horizon.
6. Average trading volume ($) over execution horizon.
7. Average price volatility of the stock (intra-day price range as % of average price)
6. On what data is the Shortfall Model based?
Goldman Sachs estimates the Shortfall Model regressions on a large sample of actual Goldman Sachs
all-day and intra-day market orders executed over the sample period. Goldman Sachs re-estimates the model monthly with rolling nine-month data. Asset level data is updated daily. The model’s shortfall
estimates reflect how much it cost to execute similar trades in the past, as opposed to providing a
hypothetical cost derived using tick data.
7. How does the model generate the shortfall estimates?
To generate the shortfall estimates, the Shortfall Model uses the following inputs: proposed trade size and execution horizon, and the average bid-ask spreads, trading volume and volatility over the prior 21 trading days. As the average spread, volatility and volume change over time the shortfall estimates reflect these changes.
8. What is the assumed execution horizon of the shortfall estimate?
The Shortfall Model in Axioma Portfolio™ software assumes a one-day execution horizon. In other words, the model assumes that trade execution begins at 9:30 AM and completes by 4:00 PM.
9. Does large order size affect the model’s accuracy?
Goldman Sachs’ data has relatively few intra-day orders above 25% of ADV. For order sizes above 25%,
the estimates are increasingly less reliable and will most likely underestimate the expected cost. We
avoid using the model estimates for order sizes above 50% of ADV.
10. How do clients access the Shortfall Model?
Users of Axioma Portfolio who choose to subscribe can gain access to the Goldman Sachs Shortfall Model through the Axioma Portfolio GUI or API.
