Behavioral Analysis of Amazon.com (AMZN)
Reminder: This analysis has nothing to do with illusions or investor mistakes. It is a study of how well-informed, highly-paid, essential investment professionals protect themselves when required to be at risk in specific securities.
The subject is the common stock of Amazon.com, studied live over the last 5 years, some 1260 market days. We daily monitor the hedging activities of market-makers as they risk capital to provide market liquidity in this stock while it trades an average of 3.4 million shares a day. We also do the same for some 2,000 other equities.
The analysis produces price ranges of sufficient likelihood, such that the market pros are willing to spend part of their trade spreads to counter price changes that would be damaging to their capital.
Experience over several years demonstrates that the ranges can have productive investment guidance over 3-6 subsequent months. Historic records indicate in which issues and under what circumstances that guidance has been strong and persistent.
Here is the daily progression of AMZN forecast price ranges over the past 6 months. The end-of-day price of the stock is the heavy dot in the vertical bar price range forecast from that day's hedging.
These forecasts have an encouraging longer history, when viewed in terms of where the e.o.d. price is located in the forecast range. Our terminology for the measure is Range Index, where the forecast range's low end = zero and its high end = 100.
AMZN came into existence in May, 1997, so there is over 3,000 market days of history, back to the start of the century. Recognizing that markets evolve, competition changes and attitudes towards individual stocks may shift, we limit our guidance from past forecasts to the most recent five years, now starting in early-mid August 2007. That happens to be roughly a time of recent peak market prices in major market indexes.
The current Range Index for AMZN, based on the prices noted in the picture above is 25, which means that 3/4ths of the price change potential seen by market pros is upwards, and 1/4th is down. The size of change forecast in each direction is +9.2% and -3.2%.
There have been 110 prior Range Indexes this favorable (or more) to a buyer of the stock. In the 3 months following each, AMZN's price was higher 74% of the time (by an average of +13.1%) and lower 26% of the time (by -5.1%).
Weighting the gains and losses by their associated odds of occurrence, we get a combined prospect of +8.4%. Our preference is to avoid buying any issue with a combined odds & payoffs below +5%. In a 3-month period +5% is an annual rate of +22%.
If all of the 110 opportunities had been exercised by following a simple, rigorous strategy, the results would have cumulated into an attractive outcome of a 102% annual rate of return on the capital involved.
The strategy involves a buy at the end of the forecast day+1 with a sell target at the top of the forecast range and a time limit of three months (63 market days). Closeout sales are produced at the first end of day price equal to or greater than the target, or at the end of the time limit regardless of any developments in the stock subsequent to the forecast. Each buy has its own target and its own clock.
The specifics here for AMZN are: Less than 9% of the 1261 days qualified as buys; 89% of the 110 buys were profitable; the cumulative gains, including the 12 losing positions, averaged +10.1%; were held an average of 34 market days; and experienced an average maximum drawdown of -6%.
Past results of any such exercise are no guarantee of future outcomes. Elements of this analysis encouraging your attention are:
1. All forecasts were made in real time and each day's forecast has had no subsequent revisions
2. Forecasts have been imputed from rational self-protective judgments made by well-informed, highly motivated, every-day-active market professionals
3. The forecasting model is based on rational behavior and has not been modified in over a decade.
4. Selection of action days to be measured, and their subsequent price change descriptions, are dependent on the current forecast for the stock.
5. The suggested investment strategy and disciplines are standard for comparisons with alternative investments.
6. Key action directions here are forward-looking. The only history involved is to get measurements under prior like expectations, which were made explicit and quantified.
7. Performance measures conform to CFA standards for individual investments, and are indicative of a specific stock, not measures of a portfolio.
These surroundings are currently a strong support for active investment in AMZN in a time-disciplined program such as is described above. This conclusion is reached in competition with other similarly evaluated alternatives, and is made despite the stock's present price/earnings ratio of 280+.