One reader posed a question: how can we estimate Russian losses when not every tank is accounted for in the visually confirmed data? This is a war, not a talk show. What about General Staff? Relying solely on figures from the General Staff may raise doubts about their accuracy. Today, we will endeavor to address these concerns using the reliable tool of statistical analysis.
How Statistical Analysis Works: Understanding Population and Sample
Let's delve into the fundamentals of statistical analysis, beginning with the concepts of "population" and "sample." While not overly complex, grasping these concepts is essential for a solid understanding of statistical analysis.
Population: A population is the complete group from which we seek to draw conclusions. In simpler terms, it encompasses the entirety of the subject under consideration. For example, if you have a farm with 100 cows, that entire group of cows is the population.
Sample: On the other hand, a sample is a subset of the population from which we collect data. Using the farm analogy, the sample would be a specific number of cows chosen from the entire population. It's a representative group that allows us to make inferences about the entire population.
Consider this scenario: on your farm, 20% of the cows are black, and the remaining 80% are white. This distribution represents the entire population. Now, if you were to randomly select 30 cows as a sample, you would expect around 20% of them, or roughly 6 cows, to be black. The fundamental principle of statistical analysis asserts that a well-chosen sample should accurately reflect the characteristics of the entire population.
Sample Size Matters: Moreover, the larger the sample size, the more reliable and accurate the conclusions drawn from the sample are likely to be. If you were to increase the sample size to 60 cows, the percentage of black cows in the sample would likely be much closer to the overall 20% observed in the entire population.
The nuances of this process are explored in greater detail in the accompanying video, offering a more comprehensive understanding of population and sample in statistical analysis.
Using Statistical Analysis to Assess Tank Losses
You might be wondering, "What cows? What does this have to do with tanks?" Bear with me; we're drawing an analogy to explain how statistical analysis applies to the assessment of tank losses. Just as a sample of cows can represent the entire population on a farm, in our case, the population is the total number of destroyed tanks reported by General Staff, and the sample is the visually confirmed losses (Oryx).
At the time of analysis, the General Staff reported approximately 5,513 lost tanks, while Oryx indicated 2,481. This sizable sample is essential for representing the broader population. But is it a random sample? Considering the span of active hostilities from Kyiv to the Oleshky district of the Kherson region, involving various units and tank models ranging from the T-90M to the T-55, we can argue that the sample is diverse and encompasses the spectrum of tanks used in the conflict.
Moving forward, our analysis aims to determine the percentage of visually confirmed losses attributed to each tank model. This breakdown will provide us with valuable insights into the distribution of losses among different tank types.
Applying these percentages to the population (General Staff figures), we can extrapolate the number of destroyed tanks based on tank models. This calculation allows us to estimate the distribution of tank losses according to specific models, providing a more detailed breakdown of the overall figures reported by the General Staff.
Certain models needed to be grouped for further analysis. Notably, the Soviet T-80 and T-72 modifications emerged as the frontrunners in terms of losses, with the T-72B3 in service following closely behind. The prevalence of the T-72B3 in the Russian military contributed significantly to its representation in the losses.
It's important to note a gap in the visually confirmed data, represented by unidentified tanks. Despite this gap, the decision was made to retain them in the analysis. This choice serves to temper the General Staff's data, introducing a degree of conservatism and mitigating the risk of overestimating the number of losses.
Losses of Tanks in Service
With these approximate data, we can determine the number of destroyed tanks in service within the Russian Federation. To achieve this, we rely on The Military Balance 2021 guide, which is the primary source available. According to TMB, as of 2021, all Russian units, including foreign bases, operated approximately 3,600 tanks.
Given that certain tank models are grouped in the directory, it was necessary to align our data with this configuration from the outset. As an example, the T-80BV and T-80U are combined into the same category.
The outcome of the analysis reveals that the Russians lost 78% of their tanks in service. This statistic provides a significant insight into the impact of the conflict on the overall tank inventory of the Russian Federation.
Next, we will break down the lost tanks by model. The graph illustrates losses only for those models that were in service before the war. The Russians suffered significant losses across various models, including almost all regular T-80BVM, T-90M, as well as older T-72 and T-80 models. Notably, the losses of the latter surpass the pre-war number of those tanks in service.
Taking the T-80BV as an example, the number of visually confirmed losses even exceeds the pre-war quantity of these tanks in service.
Now, let's shift our focus to the remaining tanks in service. The predominant type is the T-72B3, a model frequently observed in Russian losses during recent assaults in the Donbas region.
The second model is the T-90A, which presents an intriguing aspect in our analysis. Surprisingly, the Russians lost more T-90M tanks (108 units) than T-90A tanks (73 units), despite the T-90A in service count being higher (350 units) than regular T-90M (67 units).
This discrepancy can be attributed to the fact that the initial deliveries of T-90M in 2019-2020 mainly comprised upgraded T-90A units rather than entirely new builds. Russian sources have reported that T-90A tanks were elevated to the T-90M standard during this period.
This raises a critical question: are the Russians utilizing the same T-90A tanks as a "raw material" for the production of T-90M tanks? Is every T-90M produced from the beginning of 2022 genuinely constructed from scratch?
Lost Tanks from Storage Bases
Moving forward, let's identify the lost tanks from storage bases (not modernized). To obtain this figure, we subtract the number of tanks in service from the total count of lost Soviet modernization tanks. This calculation provides us with the following result:
In total, the Russians lost an estimated 1,567 tanks from storage bases. More than half of all losses are T-80s, particularly the actively deployed T-80BV models involved in offensive operations across almost the entire front line. This aligns with the strategic utilization of T-80s, considering that many tank repair plants focus on the modernization of the more prevalent T-72. Consequently, the surplus of T-80s stored in favorable conditions is contributing to their prominent role in current operations while factories enhance their capacity for T-72 modernizations.
It might seem incredulous — you might say. Well, let’s check it. The number of Russian tanks at storage bases has been meticulously calculated by numerous analysts, converging around 5,000 to 6,000 tanks in outdoor storage, with an unspecified additional quantity housed in hangars.
Let's begin by examining the T-64 (constituting 11% of losses), which was utilized by LPR/DPR units. These tanks were withdrawn from storage long before February 24, 2022, and weren't factored into analysts' calculations for 2021, as they had long been present on Donbas training grounds rather than languishing in storage bases.
Now, let's swiftly traverse the expansive tank storage sites, visually estimating the level of deconservation. While we won't conduct a meticulous count of each tank on the base, we'll make a visual comparison.
1113 base - minus ~300 units
Buryatia - minus ~250-300 units
111 base - minus ~150-200 units
734 base - minus ~250 units
And so on and so forth. The stocks are prominently decreasing.
Continuing our analysis, the significant decrease in stockpile raises a pertinent question: Could the Russians genuinely have lost 1,567 tanks from storage? Revisiting the units of the LPR/DPR armed with both T-64 and T-72B, which were also drawn from older reserves, their combined arsenal is estimated to be around 400 tanks.
Regarding satellite images of the storage base, as reported by Covert Cabal analytics approximately 7 months ago, it is indicated that the Russians removed around 1,500 tanks from storage.
With the possibility of up to two thousand tanks stored in hangars, making an accurate count becomes an intricate challenge. In this context, the figure of 1,567 tanks lost from storage bases gains a more plausible perspective.
Ratio of Losses by Origin
No Unspecified Tanks
Utilizing the same methodology for tanks modernized after 2022, it's evident that the share of losses from tanks produced or modernized after 2022 stands at 5% of the total losses. These modernized machines are actively deployed at the front lines and quickly integrated from the factories, leaving no room for the inclusion of unused reserves that could distort the assessment.
On average, the production and modernization of tanks during the two years of the war covered at least 5% of the needs of the Russian army. It's essential to note that not all modernized tanks deployed at the front are immediately destroyed, so 5% represents the lower threshold.
Moreover, 13% of the compensation for losses in the Russian Federation's tank inventory comes from tanks that are either manufactured or undergoing modernization. The remaining 87% come from storage bases without further modernization.
Is the General Staff Lying?
The entire analysis relies on the population — namely, the figures provided by the General Staff. An essential question arises: Can these figures be trusted? To address this concern, a comparative analysis was conducted following the same principles. If monthly population losses (General Staff figures) increase, the visually confirmed losses should also increase. Conversely, if the population decreases, a corresponding decrease in visually confirmed losses is expected.
The "Vishun Military" team undertook this analysis, and the graphs of verified losses and losses according to the General Staff's version exhibited a high degree of correlation with minor deviations. This alignment strongly suggests that the numbers reported by the General Staff are not arbitrary but rather grounded in verifiable data.
Visually Confirmed Losses: Challenges in Obtaining Real Numbers
The process of confirming losses visually faces several challenges that can lead to underestimating the actual numbers. Here are a couple of significant factors:
OSINT-investigators are not automated robots scanning the entire internet daily for information on lost equipment. Some instances of documented equipment losses may exist in personal accounts on platforms like TikTok and Facebook, where OSINT investigators may not physically encounter or extract the data, resulting in an omission.
Not all reconnaissance tasks are carried out with UAVs like Mavik, and even when they are, not all recorded videos are saved or published. Various operational and technical constraints may limit the publication of saved videos, creating a gap in the visual confirmation process.
As an illustrative example, consider the destroyed hangar with equipment in Balaklia. Warspotting analysts identified 24 units of destroyed equipment, knowledge that only became available after a counteroffensive operation and the physical presence of fighters at the site. How many more are there?
The residents of Kyiv Oblast even managed to conceal ammunition from the Tor air defense system and, as stated by law enforcement authorities, the air defense system itself. How many more are there? Especially if we consider trophies.
Statistical analysis is not tarot cards
Statistical analysis is a reliable and systematic tool used across various fields, including research, business, marketing, and sociology. For skeptics questioning its use in assessing losses, it's important to note that statistical analysis is grounded in empirical evidence and mathematical principles.
If you found the article valuable, consider supporting the author through Buy Me A Coffee. Your support encourages the creation of more insightful content for the audience - https://www.buymeacoffee.com/k.natovskiy
