Using Markov Chains to Help Extract Profits From DPZ, AKAM and DOCU

While the use of Markov chains — a statistical framework to decipher the probability of one event transitioning to another — in finance is not a novel concept, it’s not deployed effectively. In my research, I have come across two papers analyzing Markovian principles in the stock market: “Stock market analysis with a Markovian approach” by the KTH Royal Institute of Technology and “Forecasting Stock Prices using Markov Chains: Evidence from the Iraqi Stock Exchange” by the University of Sumer.

Conceptually, both papers attempt to decipher the utility of Markov chains to predict future market trajectories, which should yield compelling results. After all, the concept originated from Russian mathematician Andrey Markov, one of the most brilliant scientific minds and thought leaders. Unfortunately, the researchers from the aforementioned academic institutions extracted only negligible to marginal performance metrics relative to a coin toss — so, what the heck is going on here?

Fundamentally, the problem centers on the researchers’ deployment of a “literal” Markov chain — one time unit in the past to determine one time unit in the future. To be fair, KTH ran a study featuring two time units in the past but the same problem applies — the analysis would only capture an isolated price action without consideration of the underlying context or sentiment regime.

In short, the academic papers’ input is Gaussian in nature; therefore, we shouldn’t be surprised if the output is also Gaussian. In order to generate a true Markovian framework, the input must also be Markovian.

To achieve a proper framework, it’s vital to apply the spirit of the law rather than the letter of the law. My solution is to discretize the last 10 weeks of price action and segregate the profiles into distinct, discrete behavioral states. This way, we’re not just capturing isolated price action but sustained behaviors — behaviors that can better predict outcomes based on underlying situational dynamics.

Using modified Markov chains optimized for the stock market, below are three statistically compelling ideas to consider this week.

While shares of Domino’s Pizza (DPZ) are up nearly 8% so far this year, they’re down nearly 3% in the trailing month. In the past two months, the price action of DPZ stock can be converted as a “3-7-D” sequence: three up weeks, seven down weeks, with a negative trajectory across the 10-week period.

Naturally, this conversion process compresses DPZ’s price dynamism into a simple binary code. The benefit, though, is that the price action can be distinguished as belonging to one of several distinct, contingent demand profiles. Subsequently, these profiles serve as the backbone of past analogs, from which probabilistic analyses can be extracted.

Regarding DPZ stock, whenever the 3-7-D sequence flashes, the price action for the following week (which corresponds with the business week beginning July 7) results in upside 61.54% of the time, with a median return of 2.93%. Should the bulls maintain control of the market for a second week, investors may anticipate an additional 1.69% of performance.

Using data from Barchart Premier, we can mathematically determine intriguing options strategies based on risk-reward ratios. In my opinion, the potential reversal signal of the 3-7-D sequence shines a spotlight on the 460/470 bull call spread expiring July 18.

Interested speculators can learn more about the capped-risk, capped-reward structure of bull spreads here.

Plenty of finance gurus hawk the trite adage “buy low, sell high.” Yeah, well, is anyone going to explain when to buy low? That’s the beauty of using Markov chains — when applied appropriately, they can provide an empirical guideline to augment your decision-making process.

Let’s use Akamai Technologies (AKAM) as an example. Since the start of the year, AKAM stock has dropped nearly 17%. With a Markovian framework, I don’t really care why it fell; only that it did and specifically how it did. By observing the past analogs of market behaviors, we can determine the probability of how the security may react in the future.

In the past 10 weeks, AKAM stock printed a 4-6-D sequence. Since January 2019, this sequence has materialized 34 times. Further, in 61.76% of cases, the following week’s price action results in upside, with a median return of 2.65%. If the bulls maintain control of the market for a second week, there may be an additional 0.89% to 1% of performance tacked on.

For a wildly aggressive but still rational trade, speculators may consider the 81/82 bull spread expiring July 18.

Another high-risk, high-reward idea is DocuSign (DOCU), a global provider of cloud-based software. As you can tell if you pull up its chart, DOCU stock isn’t having a great time this year, down more than 12% since the January opener. Again, I don’t really care why the stock fell but the way that it did.

Honestly, providing an opinion of DocuSign would be like summarizing yesterday’s newspaper. By the time you actually read the story, the narrative could be two days old. I aim to provide an empirically grounded Markovian analysis, not post-hoc rationalizations that dominate the financial publication ecosystem.

Getting back to DOCU stock, the security printed a 6-4-D sequence, a relatively rare pattern. Since January 2019, the sequence has materialized 17 times. In 58.82% of cases, the following week’s price action results in upside, with a median return of 3.57%. Should the bulls maintain control of the market for a three-week period, traders may anticipate another 2.24% of tacked-on performance.

If you’re feeling bold and want to throw into double coverage for a chance of a big score, you may consider the 81/83 bull spread expiring July 25.

On the date of publication, Josh Enomoto did not have (either directly or indirectly) positions in any of the securities mentioned in this article. All information and data in this article is solely for informational purposes. This article was originally published on Barchart.com