The impact of internal and external loads on player performance in Chinese basketball association

Abstract

Background

Limited research has investigated the association between training load and performance of basketball players during games. Little is known about how different indicators of player performance are affected by internal and external loads. The purpose of this study was to determine whether external and internal loads influence basketball players’ performance during games.

Method

This longitudinal study involved 20 professional male basketball players from a single team, classified as first-level athletes by the Chinese Basketball Association. During 34 games, external load was measured as PlayerLoad using micro-sensors, while internal load was assessed using session rating of perceived exertion (sRPE). Player performance was quantified using three metrics: Efficiency, Player Index Rating (PIR), and Plus-Minus (PM). Pearson correlation coefficients were calculated to assess the strength of the relationships between training loads and performance metrics. Linear mixed-effects models were applied to further analyze the influence of internal and external loads on basketball performance.

Results

Pearson correlation analysis revealed moderate positive correlations between both sRPE and PlayerLoad with Efficiency and PIR. Specifically, sRPE (r = 0.52) and PlayerLoad (r = 0.54) were both significantly correlated with Efficiency. For PIR, sRPE (r = 0.50) and PlayerLoad (r = 0.56) also demonstrated moderate correlations. These correlations were further substantiated by linear mixed-effects models, which showed that sRPE (β = 2.21, p < 0.001) and PlayerLoad (β = 1.87, p = 0.004) had significant independent effects on Efficiency. Similarly, sRPE (β = 2.15, p < 0.001) and PlayerLoad (β = 2.36, p < 0.001) significantly predicted PIR. Additionally, a significant interaction effect between PlayerLoad and sRPE was found on Plus-Minus (β = -2.49, p < 0.001), indicating that the combination of high physical and psychological loads negatively impacted overall team performance. However, the correlation strengths for Plus-Minus were relatively low (sRPE: r = 0.16; PlayerLoad: r = 0.10).

Conclusion

Both external and internal loads positively contribute to performance, the integration of objective (accelerometry) and subjective (sRPE) measures of load provides a comprehensive understanding of the physiological and psychological demands on athletes, contributing to more effective training regimens and performance optimization.

Background

Training load is the cumulative stress placed on an individual from one or more training sessions (structured or unstructured) over time [1]. Training load can be divided into external load and internal loads [2]. External load is an objective measure of the work done by an athlete during training or competition, independent of internal load [3], it is determined by the organization, quality, and quantity of exercise [4]. Internal load refers to the relative physiological or psychological stress on an athlete during training or competition. Indicators such as heart rate, blood lactate, oxygen consumption, and session rating of perceived exertion (sRPE) are commonly used to assess it [5]. As new technologies for measuring athletes’ training loads become more available, sports scientists and basketball coaches are eager to adopt these innovations. External load measurement tools like electronic performance tracking systems (EPTS) [4] have become more prevalent in sports science [6]. Meanwhile, it is crucial to recognize that internal load determines the effectiveness of training [4, 7]. Integrating training load monitoring methods is essential, as external and internal loads provide different information about the demands of training and competition [8]. They should be used in combination to gain a deeper understanding of training stimuli [2].

Since training aims to enhance competition performance [9], previous studies have shown that basketball players’ performance relates to training load through metrics like game statistics [8, 10,11,12], neuromuscular performance [13,14,15], aerobic capacity [16, 17], and coaches’ ratings of player performance [18]. Evaluations of players’ aerobic capacity or neuromuscular performance are typically done during training and testing sessions since measuring these indicators during games is challenging. In contrast, game statistics offer an objective way to assess player performance during games, capturing technical and tactical actions, which is common in research. Some research [8, 11] shows a significant correlation between external load and athletic performance. The technical statistics of players during matches are associated with players’ external load (PlayerLoad, PlayerLoad/min) to varying degrees. Different perspectives have also been proposed. Some research [19] argued that there is a lack of consistent association between external training load measurements and performance in team sports, suggesting that internal training load might correlate more strongly with performance, depending on the metrics used [19]. Similarly, some research [10, 20] found no correlation between optimal performance and higher training loads in basketball players.

These findings highlight the need for further research to quantify the physical demands of basketball players and to better understand the relationship between training load and performance. Although these studies provide partial insights into the relationship between athlete load and performance, no research has analyzed the load conditions of athletes in the Chinese U19 Youth Basketball League. The players in this league all come from the youth teams of Chinese Basketball Association (CBA) clubs, some of these players will eventually play in CBA. The lack of research on the physical demands of the athletes in this league is regrettable.

Therefore, given the limited research on the impact of internal and external loads on basketball performance and the lack of studies on their interaction, this study aims to determine whether external and internal loads influence basketball players’ performance during games.

Materials and methods

Participants

This study involved 20 professional male basketball players from the same team, all classified as first-level athletes by the Chinese Basketball Association. Written informed consent was obtained from all participants prior to their inclusion in the study. Before providing their written informed consent, all players and coaches were fully informed about the study’s purpose, protocol, associated benefits, and potential risks. Approval for the study was granted by Tsinghua University Institutional Review Board (IRB 20210170). There were no players under the age of 18, and the study conformed to the Declaration of Helsinki [21].

Design

A longitudinal, observational study was conducted to monitor the external and internal loads of players. Data were collected during the preliminary and final stages of the 2023 Chinese U19 Youth Basketball League (14 games), the preliminary stage of the 2024 Chinese U19 Youth Basketball League (7 games), and 13 team scrimmages, totaling 34 games. The Chinese U19 Youth Basketball League, organized by the Chinese Basketball Association, is a professional basketball league that divides each season into preliminary and final stages. To develop more elite professional basketball players, the league requires each team to field six different players in the first and second quarters of the first half of the game. In the second half, the coaching team is permitted to allocate players according to their strategic decisions.

Procedure

External load

Before each game, all subjects wore upper-body garments with microsensors placed between the scapulae (Vector s7, Catapult Innovations, Melbourne, Australia). During the monitoring period, data analyses excluded rest periods during breaks and substitutions. Microsensor data were continuously recorded throughout all games and downloaded to a personal computer after each game for analysis using proprietary software (Openfield, Catapult Innovations, Melbourne, VIC, Australia). Microsensor data were recorded at 100 Hz using IMU devices, and external training load was quantified using a 3-dimensional accelerometer-based formula (accumulated PlayerLoad [PL]) calculated by the manufacturer [22]. This formula calculates the square root of the sum of the squares of the instantaneous rate of change in acceleration in three orthogonal planes (anterior-posterior, medial-lateral, and vertical), divided by a scaling factor (100), and reported in arbitrary units (AU). The calculation formula is as follows. Due to the frequent and rapid changes in activity and direction in basketball, current research indicates that PlayerLoad can accurately measure the instantaneous rate of change in acceleration across three planes of motion [23]. Its reliability and validity have been confirmed [24, 25].

$$\:\text{P}\text{l}\text{a}\text{y}\text{e}\text{r}\text{L}\text{o}\text{a}\text{d}=\sqrt{\frac{{\left({\varvec{a}}_{\varvec{y}1}-{\varvec{a}}_{\varvec{y}-1}\right)}^{2}{+\left({\varvec{a}}_{\varvec{x}1}-{\varvec{a}}_{\varvec{x}-1}\right)}^{2}+{\left({\varvec{a}}_{\varvec{z}1}-{\varvec{a}}_{\varvec{z}-1}\right)}^{2}}{100}}$$

Internal load

Internal measures were evaluated using the relative (min^− 1) sRPE values. sRPE was used as a perceptual indicator of the internal load based on the following formula [26].

$$\:sRPE=RPE\times\:Duration$$

where RPE is Borg’s category-ratio scale (1–10) and Duration is time in minutes.

Game performance

In-game performance was determined using the individual player efficiency, performance index rating (PIR), plus-minus (PM). Player efficiency is an individualized measure of in-game performance that combines positive and negative components to determine contribution to a game [8].

The Performance Index Rating (PIR) was chosen as performance indicator as it is the primary metric used by the Euroleague Basketball league (highest level competition in Europe), and has previously been used in basketball research [27].

Plus-Minus (PM) is a statistic that measures the point differential when a player is on the court. A positive value indicates that the team outscored the opponents while the player was on the court, while a negative value indicates that the team was outscored by the opponents. Plus-Minus metrics are widely used in well-known basketball leagues such as the NBA (https://www.basketball-reference.com/) and CBA.

Individual game-related statistics used to quantify performance for each player were officially recorded by qualified personnel after each game. The performance-related metrics and their formulas for basketball players are shown in Table 1.

Table 1 Game Performance Statistical Metrics and Formulas in basketball

Statistic analysis

An a priori power analysis was conducted using G*Power (Version 3.1.9.7, Heinrich Heine University Düsseldorf). Based on a linear bivariate regression model, the analysis used a two-tailed alpha value of 0.05, an effect size of 0.5, and power of 0.95 [28]. This determined a minimum sample size of 42 data points, which were met and exceeded in the current study. Data were collected across 34 games, including the 2023 and 2024 Chinese U19 Youth Basketball League and team scrimmages. Given the repeated measures across multiple games, the total data points exceeded this requirement.

All statistical analyses were conducted using R statistical software, utilizing the “lme4” package for linear mixed-effects models [29]. Prior to conducting mixed-effects modeling, Pearson correlation coefficients were calculated to assess the relationships between key variables, including sRPE, PlayerLoad, and the performance metrics (Efficiency, PIR, and Plus-Minus). The strength of correlation coefficients (r) was classified as small (0.10–0.30), moderate (0.31–0.50), large (0.51–0.70), very large (0.71–0.90), and extremely large (0.91–1.00) [30]. Key variables, including sRPE and PlayerLoad, were standardized prior to further analysis to ensure comparability and reduce potential bias from differing scales. Standardization was achieved by subtracting the mean and dividing by the standard deviation for each variable.

We adopted a stepwise approach to model selection, starting with a null model containing only random intercepts for player and match. The main effects of PlayerLoad and sRPE were then sequentially added, with each model’s fit evaluated using likelihood ratio tests (LRTs) [31]. Interaction terms between PlayerLoad and sRPE were also considered based on theoretical relevance, and were retained only if they significantly improved model fit. Model fit was assessed using AIC, BIC, and conditional R² values, with lower AIC and BIC indicating better fit. Residual diagnostics were performed to ensure model assumptions were met, enhancing the robustness of the final models. In the final models, coefficients were interpreted to assess the impact of internal and external loads on the performance metrics (Efficiency, PIR, and PM), with particular attention given to the significance and direction of the interaction effects between PlayerLoad and sRPE. In all models, Efficiency, PIR, and PM were treated as dependent variables. Descriptive data are reported as mean ± SD for each variable. For mixed linear models, data are presented as the estimate, the 95% CI, and the p value. The level of significance (alpha) was set at 0.05.

Results

Players’ descriptive characteristics are presented in Table 2. PlayerLoad, sRPE, EFF, PIR and PM are presented in Table 3.

Table 2 Players’ descriptive characteristics

Table 3 Descriptive analysis of physical demands and player Efficiency according to game conditions

Pearson correlation coefficients were calculated to assess the relationships between PlayerLoad, sRPE, and the performance metrics (Efficiency, PIR, and PM). The results indicated moderate positive correlations between PlayerLoad and Efficiency (r = 0.54), as well as between sRPE and Efficiency (r = 0.52). For PIR, similar moderate correlations were observed with PlayerLoad (r = 0.56) and sRPE (r = 0.50). However, the correlations with PM were relatively low, with PlayerLoad (r = 0.10) and sRPE (r = 0.16) showing weaker relationships.

To evaluate the influence of PlayerLoad and sRPE on Efficiency, PIR, and PM, we developed a series of linear mixed-effects models. We began with null models containing only the random effects (Match and ID) and then gradually added PlayerLoad and sRPE as fixed effects. We also tested their interaction to identify the optimal model for each dependent variable.

Efficiency model

Comparing the null model to the model with only PlayerLoad showed a significant improvement (χ² = 98.51, p < 0.001), indicating that PlayerLoad significantly predicts Efficiency. Including both sRPE and PlayerLoad further improved the model fit (χ² = 11.73, p < 0.001), suggesting that sRPE also has significant predictive value. The interaction between sRPE and PlayerLoad did not significantly improve the model (χ² = 0.001, p = 0.976). Therefore, the main effects model was optimal for interpretation and reporting. In this model, regression coefficients indicate that both sRPE (β = 2.21, p < 0.001) and PlayerLoad (β = 1.87, p = 0.004) significantly affect Efficiency (see Table 4).

Table 4 Parameter estimates of the linear mixed model based on the effect of sRPE and PlayerLoad on Efficiency (main model)

PIR Model

The null model served as a baseline. Adding PlayerLoad significantly improved model fit (χ² = 118.69, p < 0.001). Including sRPE further enhanced model fit (χ² = 13.17, p < 0.001). The interaction term did not significantly improve the model (χ² = 0.01, p = 0.913). The main model was optimal for interpretation and reporting. In this model, the regression coefficients indicate that both sRPE (β = 2.15, p < 0.001) and PlayerLoad (β = 2.36, p < 0.001) significantly affect PIR (see Table 5). This suggests that improvements in sRPE and increases in PlayerLoad are associated with significant increases in PIR.

Table 5 Parameter estimates of linear mixed model based on the effect of sRPE and PlayerLoad on PIR (main model)

PM Model

The null model served as the baseline. Adding PlayerLoad resulted in a slight, non-significant improvement (χ² = 2.47, p = 0.115). Including sRPE significantly improved the model (χ² = 6.96, p = 0.010). The interaction term between sRPE and PlayerLoad further enhanced model fit (χ² = 10.20, p < 0.001), indicating a significant interaction effect. Given these results, the full model was considered optimal for interpretation and reporting. In this model, the regression coefficients indicate that sRPE (β = 5.65, p < 0.001), PlayerLoad (β = -4.44, p = 0.009), and their interaction (β = -2.49, p < 0.001) significantly affect PM (see Table 6). This finding implies that improvements in sRPE and changes in PlayerLoad are associated with significant changes in PM, with a notable interaction effect. Data from 21 official games were collected for Plus-Minus.

Table 6 Parameter estimates of the linear mixed model based on the effect of sRPE and PlayerLoad on PM (full model)

To further explore the interaction between sRPE and PlayerLoad on PM, a simple slopes analysis was conducted(see Fig. 1). When PlayerLoad is -2 SD, the differences between levels of sRPE are significant (p < 0.001), as sRPE increases, PM also increases. However, when PlayerLoad is + 2 SD, the differences between levels of sRPE on PM are not significant (p = 0.70). Besides, when sRPE is high (+ 1 SD), an increase in PlayerLoad is associated with a decrease in PM. When sRPE is at the mean level, the negative relationship between PlayerLoad and PM still exists but is less pronounced. When sRPE is low (-1 SD), the effect of PlayerLoad on PM is the weakest, and the slope is nearly flat.

This suggests that under high internal load conditions, an increase in external load leads to a decrease in player performance, and under low internal load conditions, the impact of external load on player performance is minimal.

Fig. 1

Interaction plot of PlayerLoad and SessionRPE on Plus-Minus

The main effects model for Efficiency and PIR, along with the full model for PM, were optimal. These models highlight the significant contributions of sRPE and PlayerLoad to player performance metrics, with varying degrees of interaction effects. Detailed model comparisons and parameter estimates provide robust insights into the relationships between physical demands and performance outcomes in basketball players.

Discussion

The purpose of this study was to determine whether and how external and internal loads of basketball players affect their performance in games. The analysis revealed significant independent effects of both sRPE and PlayerLoad on Efficiency and PIR, with correlation coefficients indicating moderate to strong relationships (sRPE: r = 0.52, 0.50; PlayerLoad: r = 0.54, 0.56). Notably, the interaction between these variables was found to be statistically significant for Plus-Minus (interaction: β = -2.49, p < 0.001), This suggests that while sRPE and PlayerLoad independently influence Efficiency and PIR, their combined effect is crucial for understanding variations in Plus-Minus. This highlights the complex interplay between physical and psychological factors in influencing overall player performance. The tournament structure of the Chinese U19 Youth Basketball League, involving multiple games in a short period, poses unique challenges for monitoring external and internal loads. This demanding schedule complicates load management, requiring careful balance between training and recovery.

This is the first study to provide a comprehensive overview of internal and external loads in the Chinese U19 Youth Basketball League. The data includes both the preliminary and final stages of two seasons, as well as training data from instructional games. Additionally, this study is novel in exploring the impact of both internal and external loads on various performance metrics in basketball players. Prior research has mainly focused on metrics like efficiency [8], performance index rating [10, 12], points scored and field-goal percentage [11] to evaluate player performance. Nevertheless, it is vital to acknowledge that Plus-Minus metrics, extensively utilized in leading basketball leagues such as the NBA and CBA, warrant consideration in sports science research. Including Plus-Minus as a performance metric, this study provides a broader evaluation of player impact on team success, revealing new insights into the relationship between training load and game performance not fully examined previously. PlayerLoad, representing external load, showed a significant positive effect on Efficiency and PIR (r = 0.54, 0.56). It was suggested that a causal relation between game loads and game stats, however, was difficult to draw due to the different constructs of game stats and PlayerLoad [32]. The relationship between external loads and performance metrics has been examined across various levels of competition, including women’s collegiate basketball [33]. It is also noteworthy that only a slight positive correlation between external load variables PlayerLoad/min and high-intensity IMA events with EFF among semi-professional male basketball players [8]. Differences in league structure and game frequency may account for the discrepancies between these results and the current study. Unlike most national basketball leagues, which feature 1–3 games per week, the Chinese U19 Youth Basketball League follows a tournament system, with players facing six or seven games over seven days per season. The intensive schedule limits the opportunity for tactical adjustments through training, necessitating on-court performance improvements through playing rather than training. This result supports the notion that increased physical demand during games is beneficial to a point where it contributes positively without leading to fatigue or errors, ensuring that athletes are neither undertrained nor overexerted. Furthermore, previous research has shown a significant correlation between physical qualities and match-related performance statistics. Studies have found significant links between physical attributes like speed, agility, and anaerobic strength and assists and steals per game in elite Spanish adolescent female basketball players [34]. This aligns with the current study’s findings, where external load positively influenced Efficiency and PIR, especially critical for young athletes whose physical development is vital for enhancing their overall competitive performance. Moreover, our study found that U19 male basketball players might not fully execute the coach’s tactical intentions on the court. Technically, as players’skills evolve, they require increased running and jumping to bridge the tactical and technical gap with elite athletes. Consequently, effort and enthusiasm may be important factors in helping players achieve better performance. Athletes transitioning to the next level will need to prepare for greater training stressors and higher accumulated loads over the course of the basketball season [11].

In contrast to external demands, internal loads reflect players’ physiological responses to training or competition stimuli. The significant positive effect of sRPE on Efficiency and PIR indicates that internal load, as perceived by players, crucially influences their game performance (sRPE: r = 0.52, 0.50). This finding is not entirely consistent with previous research, possibly due to differences in the metrics used for analysis. In a previous study [8], the correlation between players’ acute load variables over the past seven days and game efficiency showed no significant relationship between internal load (Summated-Heart-Rate-Zones and sRPE) and efficiency. However, this study analyzed the actual load during the game in relation to efficiency. Notably, some research [19] reported that the relationship between internal training load and performance in team sports could be strong but varies depending on the selected metrics. Additionally, another research [12] found that lower internal training load during the week prior to a game resulted in better performance. Besides differences in indicator selection, inconsistencies of results between this study and other research may also be related to factors such as athletes’ level and age. In this study, although athletes need to cope with a dense schedule during the season, U19 basketball players face league regulations that require each team to field six different players in the first and second quarters of the first half of the game. Even if the post-game reported RPE values are high, the limitation on playing time makes their sRPE much lower than that reported by elite adult athletes [35]. The increase in sRPE may not reach the threshold that causes fatigue in young athletes. Therefore, there was a significant positive effect of sRPE on Efficiency and PIR. These differing findings encourage further basketball-related research to investigate load impacts in various competition and training contexts on athlete performance. This underscores the importance of psychological and physical readiness, reflecting how well-prepared athletes feel mentally and physically before entering a game. The findings suggest that coaching strategies that effectively manage athletes’ internal load before competitions can enhance their overall performance, possibly by optimizing their focus, stamina, and strategic engagement during games.

The exclusive analysis of external and internal game load variables might limit the understanding of basketball performance [36].This study integrates internal and external loads by combining objective data with subjective measurements to summarize basketball loads during games. The lack of a significant interaction between sRPE and PlayerLoad on Efficiency and PIR suggests that the benefits of psychological readiness and physical activity accumulation during games are additive rather than synergistic. However, the significant interaction effect on Plus-Minus indicates that the combined influence of sRPE and PlayerLoad is crucial for understanding this metric. Compared to Efficiency and PIR, Plus-Minus is not only based on individual performance statistics (such as scoring, rebounds, etc.) but also considers the overall team point differential when the player is on the court. Therefore, it may more sensitively reflect the interaction of multiple factors. Additionally, as a composite metric, Plus-Minus may be more easily influenced by the game context and opponent strategies. In our research, under high internal load conditions, an increase in external load leads to a decrease in player performance. External load reflects the physical demand on the player, while sRPE represents the player’s subjective feelings and psychological stress. A narrative review emphasized the importance of monitoring both external and internal loads to optimize performance and minimize injury risks among basketball players at all levels, from elite to youth [37]. The interaction of high load and high stress may exhibit different effects in different game contexts. For instance, during a game’s tense final moments, players under high load and stress may struggle to maintain peak performance, impacting Plus-Minus scores. Plus-Minus is also influenced by factors like tactical execution, team collaboration, and game context. This renders Plus-Minus a sensitive and complex performance metric capable of capturing the interactions between physical load and psychological stress. Given the significant interaction between PlayerLoad and sRPE on Plus-Minus, coaches can adopt various practical strategies to boost player performance. By monitoring PlayerLoad and sRPE in training and games, coaches can adjust training intensity, optimize recovery periods, and rotate players to avert performance drops from excessive load. In critical game moments, tactical adjustments like strategic substitutions or timeouts can aid players in managing stress and sustaining optimal performance levels. Personalized training programs that account for individual load responses and incorporate psychological training and recovery strategies are essential for players facing high-load conditions. Enhancing psychological readiness via stress management can help players maintain focus and deliver consistent performances in high-pressure situations. Our study identifies a significant interaction between PlayerLoad and sRPE affecting Plus-Minus, indicating the importance of both internal and external loads in understanding player performance. However, other research presents a contrasting view. A study investigating the relationship between training loads and game performance in elite basketball players discovered that higher training loads did not necessarily improve performance [10]. This variance underscores the complexity of the load-performance relationship, influenced by factors including player experience, training environment, and sex- and league-specific demands [38], which can alter how training loads affect performance.

While this study identifies moderate correlations between game loads (both internal and external) and basketball performance metrics, several limitations should be noted. Firstly, the study focused on a single basketball team, limiting the generalizability of the results to other age groups, competitive levels, and female players. Secondly, the study overlooked the impact of contextual variables such as player positions and the distinction between starters and non-starters, which could provide more practical insights. Integrating internal (sRPE) and external (PlayerLoad) loads, and incorporating the Plus-Minus metric, broadens our analysis and offers new insights into the relationship between training loads and game performance. Additionally, focusing on elite Chinese basketball players, a group previously underrepresented in research, provides essential context for optimizing training strategies for elite athletes. Future research should investigate the thresholds for beneficial versus detrimental internal and external loads to precisely describe their impact on performance. Extending this research to various sports or player positions could uncover unique dynamics that inform more specialized training and game strategies.

Conclusion

Overall, our study provides valuable insights into the individual and combined effects of sRPE and PlayerLoad on basketball performance in the Chinese U19 Youth Basketball League. The main findings suggest that both external and internal loads positively contribute to performance, with distinct patterns of influence on Efficiency, PIR, and Plus-Minus. Specifically, sRPE and PlayerLoad independently affect Efficiency and PIR, while their interaction significantly influences Plus-Minus. The integration of objective (accelerometry) and subjective (sRPE) measures of load provides a comprehensive understanding of the physiological and psychological demands on athletes, contributing to more effective training programs and performance optimization. Recognizing the distinct roles of internal and external loads, basketball practitioners might manage athletes’ loads at different stages of competition (preliminary and final stages) to help players achieve optimal performance.