Week 35: Final Phase 2 Breakout Stock Forecast And Test Results

|
Includes: BRS, CAR, CECO, CSII, CZZ, FLDM, HLX, KPTI, OMER, SALT, SVU, TUES
by: JD Henning

Summary

Final Test Results of Phase 2 Stock Selections from Week 25 to Week 34.

Eight new breakout stocks for Week 35 with better than 10% short-term upside potential.

Four new breakdown stocks for Week 35 with substantial downside potential.

Week 10 of the 10 week MDA stock selection test period.

Hope to see you again in about a month with more improvements in the stock selection modeling!

Thank you!

In this the final forecast article for Phase 2 testing, I want to thank all of you who joined me on this informal 10 week study of the breakout and breakdown stock forecasts based on my Multiple Discriminant Analysis (MDA) research. With so much financial material available for you to read and little time in your busy schedules, I am really thrilled that so many intelligent students of the market have followed this study over the past two months. At the writing of this tenth and final forecast article, I have received over 56,000 page views since my testing began and my subscribed readership has more than doubled.

This is for me a very sizeable following that I think reflects your intuitive observation that something meaningful appears to be occurring with this unique methodology. In fact, the results are quite significant as shown below. While these market selection systems all have their imperfections, I want to show you in statistical terms what this small sample study has delivered and give you some additional insights that may contribute to your ongoing success. If it has been profitable for you I would especially enjoy hearing your best personal results over the last 2 months.


Statistical Results

The following Chi-Square tests were applied to the sample data over the past nine weeks from Week 25 through Week 34. Two of the test weeks had only four trading days, the rest had five trading days for a total of 43 test days. Week 27 was deliberately skipped during the study as it had only 3 trading days. For daily long positions this resulted in 344 event opportunities (8 stocks x 43 days). For weekly long positions this provided 72 event opportunities (8 stocks x 9 weeks) although two of these weeks had only 4 trading days. Application of a non-parametric test like Chi-Square to compare two populations is most appropriate with this small sample of heteroscedastic stock market data that is not normally distributed. The results are as follows:

Test 1: Daily MDA Selection vs. Daily Market Average (> +10%) The Daily MDA method produced >10% daily stock gains at frequencies that exceeded expected daily market frequencies for >10% gains at a highly statistically significant level of p < 0.01. This rate of occurrence was approximately 4x higher than the daily expected number of >10% stock gains in the market population.

Test 2: Daily MDA Selection vs. Daily Market Average (> +5%) The Daily MDA method produced >5% daily stock gains at frequencies that exceeded expected daily market frequencies for >5% gains at a highly statistically significant level of p < 0.01. This rate of occurrence was approximately 2.5x higher than the daily expected number of >5% stock gains in the market population.

Test 3: Weekly MDA Selection vs. Weekly Market Average (> +10%) The Weekly MDA method produced >10% weekly stock gains at frequencies that exceeded expected weekly market frequencies for >10% gains at a highly statistically significant level of p < 0.01. This rate of occurrence was approximately 3.5x higher than the weekly expected number of >10% stock gains in the market population.

Test 4: Weekly MDA Selection vs. Weekly Market Average (> +5%) The Weekly MDA method for >5% weekly stock gains did not produce frequencies of >5% weekly stock gains at frequencies that were statistically significantly different from the broader market. However, during this test period the rate of occurrence was approximately 1.6x higher than the weekly expected number of >5% stock gains in the market population.

This outcome may indicate some advantages related to the optimum length of holding periods. More research is being done on this front and some related considerations are addressed in the interpretive section of this article.

Test 5: Daily MDA Selection vs. Daily Market Average (< -10%) The Daily MDA shorting method produced <-10% daily stock returns at frequencies that exceeded expected daily market frequencies for <-10% gains at a statistically significant level of p < 0.10. This rate of occurrence was approximately 5.2x higher than the daily expected number of <-10% stock returns in the market population. However, the sample size of 1 was significantly below the minimum value for Chi-Square testing to generate reliable results.

Test 6: Daily MDA Selection vs. Daily Market Average (< -5%) The Daily MDA shorting method produced <-5% daily stock returns at frequencies that exceeded expected daily market frequencies for <-5% gains at a statistically significant level of p < 0.10. This rate of occurrence was approximately 2.2x higher than the daily expected number of <-5% stock returns in the market population.

Test 7: Weekly MDA Selection vs. Weekly Market Average (< -10%) The Weekly MDA method produced < -10% weekly stock gains at frequencies that exceeded expected weekly market frequencies for < -10% gains at a highly statistically significant level of p < 0.01. This rate of occurrence was approximately 8.5x higher than the weekly expected number of < -10% stock gains in the market population.

Test 8: Weekly MDA Selection vs. Weekly Market Average (< -5%) The Weekly MDA shorting method produced <-5% daily stock returns at frequencies that did not exceeded expected weekly market frequencies for <-5% gains at a statistically significant level. However, the sample size of 1 was significantly below the minimum value for Chi-Square testing to generate reliable results. This rate of occurrence was approximately 2.1x higher than the weekly expected number of <-5% stock returns in the market population.

Interpreting the Results

These breakout selections over the past 9 weeks using my MDA selection process are focused on a very small population of stocks in the market capable of +/- 10% daily and weekly moves. Out of the seven different momentum segments tested in my research, these articles corresponding to segments 2 and 6 in the published study are the most volatile and potentially rewarding.


So as I run the selection screens I try to limit selections to stocks with share prices greater than $2/share and preferably greater than $5/share. I understand that this higher share price range is more attractive to most investors and avoids penny-stock speculation and higher risk and volatility.

However, these parameters also greatly reduce the available sample of highly volatile stocks capable of producing greater than 10% short term returns. For example, on any given day, say like Friday, only 38 stocks out of more than 4,000 stocks (non-OTC, ex-funds) gained more than 10% in a day. That is a segment of less than 1% of available stocks that my discriminant methodology is trying to both effectively identify and time. When this screen is limited by a $2/share threshold, then the number of 10%+ gainers is reduced by 44%. At an arbitrary $5/share cut-off fully 68% of Friday's 10%+ gainers are no longer available for selection. This is the challenge in this segment.

Additionally, a very high proportion of stocks that ever produce 10%+ daily or weekly returns do so because of idiosyncratic events that are not distinguishable or predictable among the more than 40 key variables I have analyzed in depth. For example, if you survey the top gainers on any given day a very high proportion had a sudden takeover offer, merger, surprise earnings, or significant leadership change announcement that could not possibly be anticipated by standard financial variables in quarterly earnings reports. It is for this reason that I have found significant value in a number of technical variables that add to the selection effectiveness in real time. It is also why having selected stocks in the past week and in prior weeks that repeatedly reached the top spots on the NYSE and NASDAQ largest gainer board is quite a statistically significant achievement.


Future Plans

My most immediate future plans are to take a break starting this week and enjoy some travel time with my family. As I complete this article here in Houston, I am keeping a close eye on Hurricane Harvey as it dumps record rainfall on the area. We are quite fortunate to be a safe distance from the hardest hit areas around Corpus Christi and I pray for those directly in the path who are experiencing the worst of nature.

So while I have included a last selection of 8 long positions and 4 short positions for this 10th week of the study, I do not plan to generate daily performance updates during the final week. I hope these selections will be of benefit to you and generate results consistent or better than the documented past results.

I plan to return later next month with continued improvements and a new approach toward offering MDA selection stocks for your consideration. Also, SeekingAlpha has invited me to pitch them a subscription based service for their Marketplace platform where I plan to provide an even larger sample of these breakout and breakdown stock selections with additional high-value services for subscribers. I hope many of you will join me there and invite others that you think may find value in this methodology or my other studies of financial anomalies. Your ideas for delivering exactly what you most want to see are greatly appreciated and I sincerely welcome all your direct communications with thoughts for increased value for readers.

In the meantime, there is significant ongoing work toward improved modeling that will increase my current analytical database from a 1 year and 39,000 data point system to well over 1.7 million data points over 26 years using a higher quality stock market data source. I certainly hope this will also translate to improved results in forward stock selection articles.

Again, thank you all very much for joining me in this study and I sincerely hope you obtain profits and value in these informal research articles.

Benchmark Indexes End of Week 34

Selections for Week 35 are as follows:

Positive Acceleration Momentum Stocks

These stocks have strong characteristics of positive momentum into acceleration for short-term gains. The typical momentum duration horizon from this analysis appears to be one to three weeks. Based on past analysis, it is not uncommon for stocks to sustain the momentum criteria for longer periods. Some stocks from prior week selections remain strong on the parameters screened for high momentum. The new selections of positive momentum stocks for this week include: Bristow Group (BRS), Career Education Corporation (CECO), Cosan Limited (CZZ), Fluidigm Corporation (FLDM), Helix Energy Solutions Group (HLX), Karyopharm Therapeutics (KPTI), Scorpio Bulkers (SALT), Tuesday Morning Corporation (TUES):

Negative Acceleration Momentum Stocks

These stocks have strong characteristics of momentum decline into acceleration for shorting opportunities. The typical momentum duration horizon from this analysis appears to be one to three weeks. Based on past analysis, it is not uncommon for stocks to sustain the momentum criteria for longer periods. The new selections for negative momentum stocks for this week include: Avis Budget Group (CAR), Cardiovascular Systems (CSII), Omeros Corporation (OMER), SUPERVALU Inc. (SVU):

If you find any of this research remotely useful or extremely beneficial, please consider clicking on the Follow button at the top of the page. Thanks!

As always, your comments and direct communications are sincerely appreciated. A list of prior week selections and performance articles can be viewed here.

If you are interested in long-term value stocks, I recommend following the performance of my ongoing study on the Piotroski Value anomaly model. You may also find my monthly performance studies of the Russell 3000 anomaly and the monthly Forensic Value analysis using three forensic models from academic research of interest also. Thank you for your time!

My personal preference is to invest in long positions and not to identify companies for shorting. However, in the interest of documenting both positive/negative momentum breakouts using my algorithms, a small sample of negative accelerators is included. As always, I hope you capture the most beneficial of these breakout (breakdown) candidates and have a profitable week of trading!

Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Editor's Note: This article covers one or more stocks trading at less than $1 per share and/or with less than a $100 million market cap. Please be aware of the risks associated with these stocks.