Quarterly Results and Stock Prices
Study 2: Quarterly Results and their effect on Stock Prices
Abstract/Summary
The objective of this study is to investigate if the Quarterly Results of companies influence the price of their stocks and to investigate if patterns exist in Analyst Estimate and price change data using the C4.5 algorithm. The study suggests that there seems to be a pattern to a limited extent linking Quarterly Results and subsequent Stock Price Movements. The Stock Price of a company often rises when quarterly earnings beat analyst expectations and often falls when quarterly earnings fall short of Analyst expectations.
Objective
The objective of this study is:
- To investigate if the Quarterly Results of companies influence the price of their stocks and to investigate if patterns exist in Analyst Estimate and price change data using the C4.5 algorithm.
Method
- The historic earnings estimates and actual earnings data of 4 companies (for specific time periods) are tabulated. Stocks and Time periods are chosen so that they reflect bullish(rising stock prices) as well as bearish (falling stock prices)
- The Surprise percentage (Percentage margin by which the actual earnings beat/fell short of expectations are calculated.
- The stock prices are adjusted for Dividends and Stock Splits and tabulated.
- The normalized stock price changes after periods of approximately 3 months and 1 year are tabulated. For example, Apple’s quarterly earnings in June 2007 beat analyst expectations by 28%. Also, these results were better than those of June 2006.The Stock price of Apple rose by 15% within 57 working days. The same formula for Normalized Price Difference is applied.
Normalized Price Difference = (New Price – Old Price )/ Old Price
- Thus, 3 month normalized price changes are tabulated.
- This data is further “discretized” in the following format:
Surprise, Growth, Price Movement
For the previous example, the discretized data would be:
Positive, Yes, Rise
The Price Movement in this case is “Rise” since the price rose after 3 months
- This data is fed into the system which analyzes this data using the C4.5 algorithm to produce decision trees which reflect the patterns that the C4.5 algorithm has found.
- These decision trees are examined and checked for errors.
- The predictive values of quarterly earnings are examined using the table generated from step 5.
The stocks considered over different periods of time are -Apple, Citigroup, Merrill Lynch and Eastman Kodak
Results
Decision Tree Yielded:
Surprise = Yes: Positive (19.0/7.0)
Surprise = No: Negative (6.0/1.0)
Read 25 cases (2 attributes) from golf.data
Decision Tree:
Surprise = Yes: Positive (19.0/7.0)
Surprise = No: Negative (6.0/1.0)
Evaluation on training data (25 items):
Before Pruning After Pruning
—————- —————————
Size Errors Size Errors Estimate
3 8(32.0%) 3 8(32.0%) (45.3%) <<
Tabulation of Results
| Apple | |||||
|
Quarter |
Surprise % |
Beat Last Year’sQ? |
57 WD Effect | 15 WDMax | 15 WDMin |
|
Jun-04 |
13% |
N/A |
0.11 |
0.02 |
-0.1 |
|
Sep-04 |
56% |
N/A |
0.7 |
0.08 |
-0.01 |
|
Dec-04 |
40% |
N/A |
0.29 |
0.07 |
-0.03 |
|
Mar-05 |
42% |
N/A |
-0.18 |
0.01 |
-0.06 |
|
Jun-05 |
19% |
Yes |
0.42 |
0.19 |
-0.01 |
|
Sep-05 |
3% |
Yes |
0.4 |
0.03 |
-0.04 |
|
Dec-05 |
7% |
Yes |
-0.08 |
0.17 |
0 |
|
Mar-06 |
9% |
Yes |
-0.07 |
0.1 |
-0.08 |
|
Jun-06 |
23% |
Yes |
0.26 |
0.02 |
-0.06 |
|
Sep-06 |
22% |
Yes |
0.18 |
0.03 |
-2.00% |
|
Dec-06 |
46% |
Yes |
0.13 |
0.15 |
0.00% |
|
Mar-07 |
36% |
Yes |
0.31 |
0.05 |
0.00 |
|
Jun-07 |
28% |
Yes |
0.15 |
0.16 |
-2.00% |
|
Sep-07 |
17% |
Yes |
0.33 |
0.21 |
0 |
| Citigroup | |||||
|
Mar-04 |
4% |
N/A |
-0.05 |
0.05 |
0 |
|
Jun-04 |
5% |
N/A |
0 |
0 |
-0.05 |
|
Sep-04 |
3% |
N/A |
0.08 |
0.01 |
-0.03 |
|
Dec-04 |
1% |
N/A |
0.04 |
0.04 |
0 |
|
Mar-05 |
2% |
Yes |
0.05 |
0 |
-0.04 |
|
Jun-05 |
-5% |
No |
-0.05 |
0 |
-0.03 |
|
Sep-05 |
-2% |
No |
0.09 |
0.01 |
-0.02 |
|
Dec-05 |
-2% |
No |
-0.03 |
0 |
-0.02 |
|
Mar-06 |
9% |
Yes |
0.03 |
0.02 |
0 |
|
Jun-06 |
-1% |
Yes |
0.04 |
0.02 |
-0.04 |
|
Sep-06 |
3% |
Yes |
0.12 |
0.03 |
0 |
|
Dec-06 |
3% |
Yes |
-0.09 |
0.03 |
-0.02 |
|
Mar-07 |
-7% |
No |
0.07 |
0.03 |
0 |
|
Jun-07 |
10% |
Yes |
-0.07 |
0.01 |
-0.02 |
|
Sep-07 |
7% |
No |
-0.1 |
0.04 |
-0.06 |
| Merrill Lynch | |||||
|
Dec-05 |
16% |
N/A |
0.15 |
0.03 |
-0.01 |
|
Mar-06 |
38% |
N/A |
-0.09 |
0.03 |
-0.01 |
|
Jun-06 |
7% |
N/A |
0.12 |
0.03 |
-0.02 |
|
Sep-06 |
36% |
N/A |
0.17 |
0.07 |
-0.01 |
|
Dec-06 |
26% |
Yes |
-0.12 |
0.06 |
-0.01 |
|
Mar-07 |
15% |
Yes |
0.03 |
0.02 |
-0.04 |
|
Jun-07 |
11% |
Yes |
-0.16 |
0 |
-0.08 |
|
Sep-07 |
-533% |
No |
-0.21 |
0.08 |
-0.01 |
| Eastman Kodak | |||||
|
Mar-06 |
-780.00% |
-0.21 |
0.00 |
-0.07 |
|
|
Jun-06 |
-186.40% |
-0.09 |
0.00 |
-0.1 |
|
|
Sep-06 |
136.80% |
0.20 |
0.06 |
0 |
|
|
Dec-05 |
35.90% |
0.21 |
0.07 |
-0.04 |
|
|
Dec-02 |
-4.40% |
-0.20 |
0.01 |
-0.04 |
|
|
Mar-03 |
0.00% |
0.04 |
0.04 |
-0.03 |
|
|
Jun-03 |
106.90% |
0.00 |
0.00 |
-0.05 |
|
|
Sep-03 |
54.40% |
-0.14 |
0.00 |
-0.25 |
|
|
Dec-03 |
37.30% |
Yes |
0.04 |
0.10 |
0 |
|
Mar-04 |
5.90% |
Yes |
0.03 |
0.03 |
-0.02 |
|
Jun-04 |
49.20% |
Yes |
0.19 |
0.07 |
0 |
|
Sep-04 |
9.70% |
No |
0.02 |
0.09 |
0 |
Terminology
- 57 WD Effect – Normalized Price change after approximately 1 quarter.
- 15 WD Max – Normalized Difference between Maximum Price over a 15 day time period after the declaration of results and the Opening Price on the day of the declaration of results.
- 15 WD Min – Normalized Difference between Minimum Price over a 15 day time period after the declaration of results and the Opening Price on the day of the declaration of results.
Analysis of Results
- The C4.5 algorithm returned a very simple tree with a low depth. However, this tree has a high error rate of 32% suggesting that the algorithm found a pattern to a limited extent within the data.
- Quarterly Earnings seem to have some predictive value with regard to future price movement.
- The Stock Price of a company appears to rise when quarterly earnings beat analyst expectations and gall when quarterly earnings fall short of Analyst expectations. However, there are several exceptions to this rule.
- A strategy that takes advantage of this pattern is:
- i. IF Results beat expectations : BUY / Cover
- ii. IF Results fall short of expectations : SELL/ ShortSell
- It appears that whenever a company’s profit falls short of expectations by over 50%, the stock price plummets in excess of 15% of its value at the time of the release of results.
Conclusions:
The study attempting to find patterns in Quarterly Results and subsequent Stock Price movement data using ID3 and C4.5 algorithms suggests that there seems to be a pattern to a limited extent linking Quarterly Results and subsequent Stock Price Movements. The Stock Price of a company often rises when quarterly earnings beat analyst expectations and often falls when quarterly earnings fall short of Analyst expectations.
References
- Ross Quinlan’s C4.5 library source code was used to generate decision trees
- All financial data was retrieved from Yahoo Finance URL – http://finance.yahoo.com
No comments yet.
Leave a comment!
You must be logged in to post a comment.
