Collecting data from smartwatches worn by study participants enables us to investigate their emotional arousal to the 2025 cup final by analysing the physiological responses — specifically stress levels and heart rates — during the match and comparing these with patterns observed on regular days. In this section, we (1) present summary statistics on participants’ stress levels and heart rates, (2) consider patterns of vital parameters on the day of the cup final in comparison to the entire observation period, and (3) take a closer look at the course of the cup final itself. The discussion of our findings is left to the subsequent section.
Summary statistics
Table 1 presents summary statistics of the individual stress levels and heart rates. One observation represents the corresponding value of a single participant at a specific time point. We divided the data into the day of the cup final and regular days. Stress levels on both the cup final day and on regular days covered the complete scale (0 to 100). However, all remaining descriptive statistics for stress levels were higher on the day of the cup final, with the average stress level (44.2) being increased by approximately 41% compared with regular days (31.3). Similarly, heart rate statistics were generally elevated on the day of the cup final, with the exception of the maximum value. Moreover, we have calculated the 95% confidence interval for the mean stress level across the 78 regular days. This interval (28.1 to 35.6) did not include the average stress level on the day of the cup final, indicating a statistically significant increase. Similarly, we calculated the 95% confidence interval for the average heart rates on regular days. As with stress levels, this interval did not cover the cup final average heart rate (78.7). Therefore, the average heart rate on the day of the cup final was significantly higher than on regular days.
Figure 1 illustrates the distribution of the average stress levels (left panel) and average heart rates (right panel) for each participant across time points using box plots comparing the day of the cup final to regular days. We find that both the average heart rate and average stress level were elevated on the day of the cup final. While the maximum average stress level on regular days was approximately 50, values rose to as high as 90 for single individuals on the day of the cup final. Notably, the variation in stress levels among study participants was much greater on the day of the cup final. For the relation between heart rates and stress levels on the day of the cup final and regular days, see Figure S1 in the Supplementary Material.
Boxplot of the average stress level (scale from 0 to 100, left) and average heart rate (bpm, right) per individual participant across time points for the day of the cup final (May 24, 2025; orange) and regular days within the observation period (May 14 to July 31, 2025; blue).
Vital parameters over the course of the matchday in light of regular days
Average stress levels (scale from 0 to 100) across participants and across regular days between May 14 and July 31, 2025 (left panel) and average stress levels (scale from 0 to 100) across participants over the course of the day from May 19 to May 25, 2025 (right panel). The black box highlights the cup final on May 24.
In this section, we compare vital parameters on the cup final matchday with those on regular days within the observation period, thereby providing a basis for understanding how matchdays alter typical physiological dynamics. Figure 2 illustrates the average stress levels across participants and across regular days throughout the study’s observation period (left panel) and average stress levels across participants during the week of the cup final (right panel). Darker colours indicate higher average stress levels. As we observe variation in participants’ stress level patterns depending on the day of the week, we averaged stress levels in the left panel not only across all participants but also across the same weekdays of different weeks. Consequently, the values for regular days represent the average stress levels across all participants over the 12 weeks of data collection and across the respective day of the week. Averages per time point were calculated based on the number of applicable measurements, which varied slightly.
Recurring patterns were evident across weekdays, with stress levels generally lowest at night. Additionally, we observe clear differences between typical weekdays (Monday to Friday) and weekends (Saturday and Sunday). On weekdays, average stress levels started to increase at around 6 a.m., whereas on weekends, they did not rise before 8 a.m. Furthermore, average stress levels on regular Saturday nights were noticeably higher than on weeknights. Overall, Saturdays appeared to be the most stressful days, showing higher average stress levels than other days of the week during time awake, even in weeks without a football match. In contrast, weekdays tended to be less stressful, with average stress levels on Sundays comparable to those observed on weekdays.
We now compare the week of the cup final (right panel) with averages across regular days (left panel). Overall, the two panels show broadly similar patterns for most of the time. However, the day of the cup final, May 24, stands out clearly. On that day, the average stress level was considerably higher than on any regular day, including regular Saturdays, starting from typical wake-up times. Notably, an elevated average stress level was already apparent during the preceding night. It continued to rise over the course of the day, particularly after lunch, and peaked between 6 and 8 p.m., right before kick-off. Although substantial differences in average stress levels were evident during the match, it remained elevated afterwards, exceeding averages observed at any time on regular days.
Average stress level (scale from 0 to 100) across participants over the course of the cup final matchday (May 24, 2025; orange) compared with the average stress level (scale from 0 to 100) across participants and across all regular Saturdays within the observation period (May 14 to July 31, 2025; blue). The shaded areas represent the 10th- and 90th-percentiles of averages across participants for regular Saturdays. For the graphical representation, we smoothed averages and percentiles to reduce volatility and increase interpretability.
Next, we evaluate the difference between the Saturday on which the cup final took place and regular Saturdays — the days that were, on average, the most stressful during the week anyway — in more detail. Figure 3 illustrates the average stress levels across participants on the day of the cup final compared with the average stress levels across participants on regular Saturdays, with the latter being averaged across both individuals and multiple Saturdays. In addition, it displays the 10th- and 90th-percentiles across Saturdays: for each Saturday and each time point of the day, stress levels were first averaged across participants, yielding one value per time point and Saturday; the percentiles were then computed across these values at each time point. Similar to our findings in Fig. 2, we observe higher average stress levels on the day of the cup final than on regular Saturdays at every time of the day. This difference was particularly pronounced before kick-off and during the match. Specifically, during the match, i.e. between 8 and 10 p.m., the average stress level was higher by 43% compared with regular Saturdays. Moreover, during the typical time awake, the average stress levels on the day of the cup final exceeded the 90th-percentile of regular Saturdays at every time point.
Vital parameters over the matchday and the course of the match
The final of the German cup competition on May 24, 2025, attracted considerable national and international attention, primarily due to the participation of third-division club Arminia Bielefeld. Given the exceptional significance of this event for Arminia Bielefeld fans, this section takes a closer look at the cup final itself and the responses of supporters who attended the match. We first outline the course of the cup final, then present fans’ vital parameters over the match period in relation to the uncertainty of the match outcome and across different viewing contexts, and finally consider the impact of the day of arrival in the host city on fans’ stress levels for those who attended in the stadium.
According to pre-match betting odds, provided by a major European bookmaker, VfB Stuttgart, the runner-up of the previous national league season and opponent of Arminia Bielefeld, was deemed the clear favourite. Specifically, translating these betting odds into odds-implied winning probabilities as a predictor for the match outcome17, the estimated probability of a VfB Stuttgart victory was 70%, a draw after 90 minutes was about 20%, and an Arminia Bielefeld win was approximately 10%. Although the initial phase of the match was quite balanced, with even promising scoring opportunities for Arminia Bielefeld, VfB Stuttgart scored in the 14th, 22nd, and 27th minutes. This 3-0 lead increased VfB Stuttgart’s odds-implied winning probability to over 95%, rising to more than 97% by halftime. In the second half, VfB Stuttgart scored again in the 67th minute, before Arminia Bielefeld was able to reduce the deficit with goals in the 84th and 87th minutes, eventually resulting in a 4–2 victory for VfB Stuttgart. Despite these being the first goals ever scored by a third-division club in the cup final, they had minimal impact on the odds-implied probabilities: Prior to Arminia Bielefeld’s first goal, the probabilities were 99.5% for a VfB Stuttgart win and 0.25% each for a draw and an Arminia Bielefeld victory. Following the 4–2 scoreline, VfB Stuttgart’s winning probability only slightly decreased to 97.7%, with a 2.0% probability of a draw and 0.3% for an Arminia Bielefeld win.
Average heart rate (bpm) across participants over the course of the match (black solid line; left y-axis) and objective uncertainty of the match outcome derived from betting odds measured by \(1-\text {Gini}\) (grey dotted line; right y-axis) during the cup final. For the graphical representation, we smoothed the average heart rate to reduce volatility and increase interpretability. The grey shaded area marks the halftime break. Goals are indicated by vertical lines (red: VfB Stuttgart; blue: Arminia Bielefeld), together with the resulting scoreline.
Figure 4 illustrates the average heart rate of those study participants who provided data throughout the cup final, starting from kick-off at 8 p.m. (minute 1) until the end of the added time (around 10 p.m.). To relate fan involvement to uncertainty about the match outcome, the figure additionally depicts the measure \(1-\text {Gini}\) to quantify (objective) uncertainty (grey dotted line, right y-axis). The Gini coefficient is calculated based on odds-implied probabilities and ranges from 0 to 1, with higher values of \(1-\text {Gini}\) indicating a higher degree of uncertainty. For example, odds-implied probabilities of one-third for each of the three possible outcomes would yield a \(1-\text {Gini}\) value of one.
Results presented in Fig. 4 illustrate that the heart rates of Arminia Bielefeld fans were highest during the first 15 minutes of the match, with the average reaching up to approximately 96 bpm. After each goal scored by the opposing team, VfB Stuttgart, we observe a decrease in the heart rate, with the average value falling below 90 bpm until the beginning of the halftime break. The average heart rate increased during the first few minutes of the break; however, when the match resumed, it remained relatively low between 86 and 90 bpm. The average heart rate dropped to its lowest point of the match shortly after the fourth goal by VfB Stuttgart (below 86 bpm around the 70th minute). Especially for the first half, average heart rates aligned with the development of objective uncertainty, represented by \(1-\text {Gini}\): Both lines peaked within the first 15 minutes of the match and then steadily decreased until halftime. While the average heart rate declined more gradually, the objective uncertainty based on odds-implied probabilities decreased in steps following each goal. The results in Fig. 4 are particularly interesting for the last 15 minutes of the match, where Arminia Bielefeld scored two goals. While the objective uncertainty of the match outcome remained very low (the value of \(1-\text {Gini}\) increased from 0.007 to only 0.026 after the two goals), the average heart rate increased by around 10 bpm and reached similar values after the second goal by Arminia Bielefeld as during the first 15 minutes of the match.
Average heart rate (bpm) across survey participants with the same viewing context over the course of the match (blue: stadium; orange: public gathering; pink: TV). The shaded areas represent the heart rates’ 10th- and 90th-percentiles. For the graphical representation, we smooth averages and percentiles to reduce volatility and increase interpretability. The grey shaded area marks the halftime break. Goals are indicated by vertical lines (red: VfB Stuttgart; blue: Arminia Bielefeld), together with the resulting scoreline.
To further contextualise the fan experience, Fig. 5 shows the average heart rate over the course of the match, while distinguishing between participants who watched the match in the stadium, at a public gathering, or on TV. These values are based on different numbers of participants, determined by the location of the match attendance, and include only those 37 participants who completed the additional survey. The average heart rate was highest for participants in the stadium (average: 94.2 bpm), followed by those watching on TV (79.4 bpm) and at a public gathering (73.8 bpm). On average, heart rates of stadium attendees were 23.1% higher than those of participants watching elsewhere. After the first goal of Arminia Bielefeld, this difference increased to up to 35.8%, with a maximum average heart rate of 108.0 bpm in the stadium. Beyond the location where the match was watched, alcohol consumption was also associated with elevated heart rates: The average heart rate of participants who reported alcohol intake was on average 5.3% higher throughout the match, 7.4% higher during the second half, and even 11.7% higher following Arminia Bielefeld’s first goal.
Average stress levels (scale from 0 to 100) across survey participants who attended the final in the stadium with the same arrival in Berlin (orange: Friday (or earlier); blue: Saturday) over the 24 hours prior to kick-off. The shaded areas represent the stress levels’ 10th- and 90th-percentiles. For the graphical representation, we smooth averages and percentiles to reduce volatility and increase interpretability.
For the study participants who attended the cup final in the stadium, we further gathered information on whether they arrived in the host city on Friday (or earlier) or on Saturday, i.e. the matchday. Figure 6 illustrates the average stress levels across these fans, differentiated by the day of arrival, along with the 10th- and 90th-percentiles. For both groups, the average stress level steadily increased over the course of the matchday, peaking just before kick-off. Notably, there were distinct differences depending on the day of arrival. Fans who travelled to Berlin prior to the matchday exhibited higher average stress levels late Friday evening. In contrast, those arriving on Saturday displayed increased average stress levels from 6 a.m. on the matchday, while the increase of the Friday group started around 8 a.m. only. While the average stress levels were relatively similar between 11 a.m. and 2 p.m., fans arriving on Friday or earlier exhibited lower average stress levels from 2 p.m. until kick-off.
