Q1 - Discuss Numeric's investment strategy? What is its rationale for Numeric? Sol 1 – Numeric’s Numeric’s investment strategies were:
1. Long – Long – only only strategy 2. Long / Short strategy with equitization 3. Long / Short strategy without equitization Numeric managed tax-exempt accounts for pension funds and endowments and funds of funds taxable accounts. Firm managed different funds across seven product categories. Five of the product categories involved long-only strategies whereas rest two categories involved long/short strategies. Both of the categories were offered with and without equitization.
Q2 - What is the difference between long only and long/short strategies? Do they involve the same level of risk?
– Long / short strategy was considered more advantageous in comparison to long-only Sol2 – Long strategy. Long-only strategy could be profitable only if share prices of owned securities increase whereas long / short strategies allowed one to exploit negative as well as positive information. Long-only strategies forced to mirror the sector exposure of the benchmark and might result in owning the stock in a sector having no compelling purchase and thus it tended to bundle diversification along with stock selection whereas long/short strategies could be managed more easily for pure stock selection and offered investor much higher alpha per dollar of assets invested. No they involved different risks. Long only strategies was exposed to two different sector risks which were bundled for diversification and it might not result in eliminating the sector specific risk completely whereas long / short strategies could largely eliminate all except stock specific risks which was also there in case of long-only l ong-only strategy.
Q3 - What is the difference between momentum strategy and value strategy? How do they both work?
– Momentum Strategy Sol 3 – Momentum Momentum strategy was based on the belief that security analysts tended to under-react to th e unexpected changes in corporate earnings, and that investors, by pegging stock prices to
consensus estimates also tended to under-react. Thus if Numeric could forecast revisions in analysts’ earnings estimates it would be able to forecast changes in stock prices. Strategy was to get in early and ride the stock either up and down before its earnings revisions ran out of momentum. Numeric used following approaches to forecast future earnings estimate revisions. 1. Past changes in analysts’ estimates Numeric captured this effect in its Estrend (Earnings estimate revisions trend) model by calculating the recent rate of changes of analysts’ earnings per share estimates for a given stock
2. Earnings surprises It was based upon company announcement of quarterly earnings that were significantly different from the consensus of analysts’ expectations. It was considered significant if the difference between the actual and consensus forecasts of earning per share was large relative to the dispersion in analysts’ forecasts. It was measured in Earning Surprise score. For a given stock Estrend score and Earning Surprise score were combined in a weighted average to form a stock’s Momentum score. Value Strategy Value strategy was based on the premises that stock prices were more volatile than the underlying company fundamentals. It meant that prices would exhibit mean reverting behaviour which could be forecasted based on company fundamentals. Fair value model was estimated nightly based on closing day’s prices. It used cross sectional regression analysis to determine the statistical relationship between certain characteristics of the company and its stock price. Regression equation was used to calculate a fitted price or fair value for each company on the basis of its characteristics. A stock was considered cheap or expensive if its actual price was below or above its fitted price, respectively. Combination of Both Momentum and Value models complemented each other. Thus, incorporating both models led to more consistent performance. Numeric developed Skill adjustment model. Skill of a model was defined as the correlation between a stock’s model score and its subsequent
return. It calibrated the effectiveness of momentum and value scores in predicting future price movements.
Q4 - Analyze the transactions of April 29, 1997. How and why are these representatives of Numeric's general investment strategy? Sol 4 – Different transactions of April 29, 1997 were in following securities:
1. PepsiCo Numeric was already short in PepsiCo on the basis of negative momentum score. Subsequent to positive announcement, quality of PepsiCo’s announced earnings were checked and momentum and value model scores were recalculated. Since both momentum and value score changed to positive as well as information was clean many shares were decided to be covered. As expected subsequently share price of Pepsi increased
2. Stratus Computer Earning surprise by Stratus led to increase in its revised earning estimate by Numeric analyst in comparison to consensus estimate. It resulted in increased momentum as well as value score. Subsequently the stock price was expected to increase
3. Micro Warehouse Numeric was short on its expectation of disappointment from its first quarter earnings. But similar to PepsiCo case earning announcement was above consensus estimate. Again news was confirmed and Numeric covered its position in this share while the stock rose.
4. Peoplesoft Although the fair value score of this share was negative but earning surprise score was very high leading to high momentum score. So Numeric decided to take advantage of its positive momentum and added these shares once their prices came down after initial surge. Subsequently the stock rallied providing benefit t o Numeric.
5. Compuware
This company was unaffected by recent decline in technology stocks so Numeric was having long position in the same. But due to low liquidity its momentum score was low and analyst decided to reduce its holding. But subsequently after the earnings announcement momentum score increased and it was decided to increase its holding
6. Nabisco Holdings Numeric stock’s value score was low and Numeric was short in this. Stock had been weak and falling. Due to continuous fall it looked less overpriced and its value score increased. So analyst decided to reduce its short position to take advantage in other overpriced stocks. Then cover order was executed and subsequently its stock price increased.
Q5 - How does the investment process at Numeric differ from that of more traditional fund managers? Sol 5 – Traditional fund managers usually can be categorized as active or passive, follow
either growth investing strategy or value investing strategy, have preference towards a particular sector/sectors or small cap/ mid cap/large cap stocks, diversified or financial instrument specific. Moreover investment process followed by him would be specific to his investment strategy and analysis. One thing that can be common to any fund manager irrespective of above categorization is the fact that the fund manager would rely heavily on his own analysis of firm and stocks and would base his investment decision on the findings. He may or may not revise his decision on the basis of estimates/revisions of other analysts. Also, generally traditional investors are either bullish or bearish on a sector. Hence, generally, they would either go long on a sector or go short on sector but not both long and short simultaneously for the same sector. On the other hand, Numeric Investor, while still followed momentum model (Estrend Score and Earnings Surprise Score) and value model, but these models relied heavily on Analysts’ estimates and changes & frequency of revision of their estimates. They had researched each analyst’s past predictions, frequency of change, degree of change in his prediction over time and its effect on price of the stock. Basis this, they had formulated models to quantify these and its effect and hence calculate a score which would form a basis for their investment decision. Each morning, portfolio managers at Numeric Investors would spend a lot of time
studying any new earning announcements and pre announcements by companies, and also new revisions by Wall Street analysts of their earnings estimates. The portfolio managers paid the most attention to analysts they felt were particularly trustworthy, and to analysts whose estimates tended to be especially controversial. For stocks where momentum had significant skill, the portfolio managers would re run the Estrend and Earnings Surprise Model with the new information. In the case of stock with significant value skill, but relatively little value skill, the process involved far less human effort. Here, the principal criterion for a trade was the stock’s fair value score, which was mostly driven by its price and much less so by news. Also, Numeric Investors specialization was long/short investing which involved holding a portfolio of long positions and short positions sometimes within a same sector simultaneously.
