TUPRAS, or the Turkish Petroleum Refineries Corporation, is Turkey’s largest oil refining company. It was established in 1954 and is headquartered in Kocaeli, Turkey. TUPRAS operates four refineries in Kocaeli, Izmir, Batman, and Kirikkale, with a total capacity of 28.1 million tons annually. I sold my TUPRAS shares two days ago. My answer to those who ask why is clear: the price has reached the point I wanted. In my article titled Election Volatility Reveals Itself, I stated that if the price comes between 630 TL – 650 TL, I will sell my remaining Tüpraş shares. I sold my last 150 shares at approximately 632 TL (33$). Thus, I took a break from my Borsa Istanbul adventure for a while. For now, I will follow political and economic developments. Of course, I will also continue to look for new opportunities.
TUPRAS: High returns in dollar terms
In November 2022, I published my article titled A Note on the Return Potential of TUPRAS. In that article, I tried to estimate the remaining return potential of the TUPRAS shares I held using the regression analysis I always use. I put a note at the end of the article for those curious about the method. They can look there. Anyway, let’s get back to the subject.
Below you see a graph of the analysis I did on that date. The dark red line is the regression line we estimated, the wavy dark blue line is the observations, and the dashed lines are the 1, 2, and 3 standard deviations. The dashed line on the vertical axis shows the dollar value of the stock at the time of publication of this article. We can say that its value on that date was approximately 23.5 dollars. Based on this figure, I commented that TUPRAS had converged close to its long-term trend value.
TUPRAS farewell scenarios
In my November 2022 article, I examined the farewell scenarios for TUPRAS in a sense. Within five months, the second scenario almost came to life. What was this scenario? In the next two years, Tüpraş would converge to USD 36.3, two standard deviations above its long-term trend in USD terms. While writing this article, TUPRAS saw 645 TL / 18.93 $/TL = 34.07 USD. In my regression analysis five months ago, two standard deviations above its expected value were $34.4. In other words, it reached approximately this value in 5 months! This price level represents the second-highest point in the data set. Can it go higher? Yes, it can go higher. It is impossible to know the future.
Expected return calculation for new investors
How much return can new investors entering at this point expect? Let’s do the same analysis using today’s current price. Let’s bet that the price will see three times its long-term trend value ($39.5) by the election. It has seen this point only once in history. Suppose the observations in the data have a normal distribution. In that case, the probability that the price is within one standard deviation below its trend value is about 68%. There is a 95% probability that the price is within two standard deviations below the trend and a 99.7% probability that the price is within three standard deviations above the trend. In addition, the normal distribution is symmetric around its mean.
In this framework, we can mathematically express our expected profit from a bet by buying as E[Profit] = price upside potential2.5%+ price downside potential97.5%. Let’s assume that the firm will not go bankrupt and will, at most, fall to its long-term trend value ($24.4). In this case, our expected profit is (39.5 – 34.07) * 2.5% + (24.4 – 34.07) * 97.5% = -$9.3! In this scenario, we have a negative expected return, not a positive one.
How much return did I get?
It was my most profitable investment in the last four years. My average cost was around $11 – $12. Accordingly, in USD terms, I have almost tripled my return in the previous four years. At the cost of $12 and a selling price of $33, that’s an average annual rate of return of 28.8%. That’s a great rate for me.
The current asset allocation of the portfolio
After saying goodbye to TUPRAS, there was a radical change in my asset allocation. As you can see from the chart below, the proportion of cash in my financial independence portfolio rose to 27.8%. The S&P 500 ranks second with 26.3%. US long-term treasury bonds are in third place with 21.6%. The share of Turkish Treasury Eurobonds fell to fourth place with 21.1%.
In conclusion
I said goodbye to my old favorite company TUPRAS. In the long run, this may not be a permanent separation. Turkey’s economy is on a long-term up-and-down course. When new crises come, stock prices may reach lovely points again. That’s when serious purchases can bring big profits. Of course, there is also this: The recent upward wave in Borsa Istanbul did not happen in a free market environment. There is an intense public intervention.
Moreover, the election process and what will happen after the elections are still behind a thick smokescreen. This smokescreen may lift in the coming days. In general, it is unclear what will happen tomorrow in Turkish politics. You know the opposition’s back-and-forth in the candidate selection process. As a conservative small investor, I have reduced my country’s risk to 21.1%. I may act entirely differently if the uncertainty decreases and Turkey has a new story.
What do you think about Borsa Istanbul and the election process? Should I be more cautious? Goodbye, see you in the following article.
Note: method
Let me briefly inform my new readers about my methodology and data sources. I obtain monthly price data from Yahoo Finance. Yahoo Finance adjusts the price data for stock splits and dividend distributions. I download monthly $/TL exchange rate data from the Central Bank of the Republic of Turkey’s Electronic Data Distribution System database. I convert TL-based price data into dollars.
To adjust current prices for inflation, I obtain data on the US consumer price index from the Federal Reserve Bank of St. Louis Economic Research database. I then adjust current prices for inflation and pull them to the most recent month’s data. Using a statistical program, I run a linear regression of the price data against time and obtain the trend and standard errors. Lastly, I graph the data.