Mining the training load metrics in basketball: a systematic review

1. Introduction

Basketball is a team sport marked by frequent and dynamic interactions both within and between players, requiring a blend of technical, tactical, physical, and anthropometric components [1-3]. For coaches, the use of strategies to track athletes’ responses to training stimuli is essential for effectively managing both training and competition [4]. Monitoring training load offers objective data to guide training adjustments, aiding in the precise periodization of physical preparation and reducing the likelihood of overtraining [5], 6]. Consequently, numerous methods have been developed to monitor internal and external training load in team sports [7], [8].

Internal training load refers to the physiological stress experienced by the athlete in response to a training stimulus, such as perceived exertion, heart rate, or biomechanical data [9-11]. External training load, on the other hand, measures the physical effort imposed, independent of the athlete’s internal response – for example, training duration, distance covered, running speed, and body accelerations [12].

Monitoring strategies should be tailored to each sport, considering its specific performance demands [13]. In basketball, time-motion analyses highlight the importance of high intensity. These intermittent actions require sport-specific movements like dribbling, positioning, and cutting, placing unique physical and physiological demands on players [4], [14]. Therefore, evaluating the effectiveness of training load models within basketball settings is necessary to ensure proper monitoring practices [13]. Offensive and defensive drills during practice elicit similar physiological and physical responses, although competitive games are significantly more demanding [15]. Tools such as accelerometers and heart rate monitors can help distinguish between the demands of practice and actual competition [8], [9].

Internal training load responses have been widely documented in isolated training sessions [8] and games [16]. While studies like that of Manzi et al. [6] offer new insights into internal load models, comparisons between internal and external load in basketball are still lacking [13]. The measurement of external training load is justified by its association with internal responses [13], and these two types of loads are often described as dose (external) and response (internal) [12]. The limited research on external load models in basketball may be due to challenges associated with popular monitoring tools used in other sports [13]. Time-motion video analysis is labor-intensive [17], while GPS technologies face signal issues and accuracy limitations in indoor environments [15], [18], [19].

Accelerometry has emerged as a practical alternative, overcoming many of these drawbacks and proving effective for monitoring external load in team sports [20]. Tri-axial accelerometers have been developed to assess load by calculating the vector magnitude of acceleration changes across three planes of motion [15]. Since basketball involves full-body movements in multiple directions – forward, backward, lateral, and vertical – accelerometer-based models are considered well-suited for external load monitoring in this sport [15], [20]. Edwards [21] introduced a heart rate zone-based model that estimates training load based on time spent within predefined HR zones. This approach has since been applied in sports like Australian football [22], rugby [23], and soccer [24].

There is currently a lack of research on training load among basketball players at different competitive levels [25], [26], and no systematic reviews have been conducted to analyze and compare training and game loads, despite the clear need for such assessments [27], [28]. Internal training load models that incorporate perceptual [29] and physiological measures remain the most widely used in team sports [11]. Training load monitoring is vital for effective periodization, understanding cumulative training effects, and optimizing athlete development [30]. Recognizing seasonal variations and the relationships between different metrics is key to selecting the most effective monitoring strategies [31]. Accordingly, this systematic review aims to offer a comprehensive overview of internal and external training load metrics in basketball, explore their application in both training and competition contexts, and discuss their implications for performance monitoring and enhancement.

2. Methods

3. Results

3.1. Search results and study selection

A total of 95 articles were found by searching the PubMed search database. No articles from complementary sources were recognized with the potential to be implemented in the review. Subsequently, the type of articles intended (observational, clinical trials, and RTCs) were selected and 57 articles were eliminated. A total of 38 articles were analyzed based on title and full text according to the inclusion and exclusion criteria. No article was eliminated after analysis. After the screening procedures, 7 articles were included in the present systematic review. A detailed representation of the screening procedures is presented with a flow diagram shown below.

Fig. 1PRISMA flow diagram

4. Discussion

The present systematic review aimed to provide a comprehensive overview of the internal/external training load metrics used in basketball training/competition. The results of the analyzed studies were organized into weekly analyses of load distribution and relationships between internal and external loads. Importantly, the heterogeneity of the included studies must be recognized as a central finding of this review. The seven studies varied considerably in terms of competitive level (elite, semi-professional, university), sample size, monitoring period (ranging from 3 to 15 weeks), and training load metrics employed (HR-based measures, RPE, accelerometry, vertical jump testing). This diversity makes it difficult to establish direct comparisons or unified conclusions. Rather than concluding which metrics are definitively “the most applied,” the review highlights the range of methods explored within this limited body of literature. Metrics such as RPE, HR-based models, accelerometry-derived indicators (ACC, DEC, PL, IMA), and neuromuscular tests (CMJ, COD) have all been applied. Their frequency of use in the available studies reflects emerging trends rather than established consensus. Coaches and practitioners should therefore interpret these findings as indicative of possible approaches rather than prescriptive best practices.

Therefore, the present discussion was conducted following the independent variables of age group, match-related contextual factors, periodization structures (mesocycle, and/or season phases), playing positions, weekly schedule format (week of 1, 2, and 3 games), and training load indicators. This comprehensive analysis provides insights into how these variables influence training load, highlighting practical applications for coaches and practitioners in adapting training regimens to the specific needs of basketball players. The main results showed that mesocycles present a load variation and a variation along the season phases, influenced by the type of training, playing positions, age group, training mode, and contextual factors. For example, understanding these variations can help optimize workload management to improve performance and reduce injury risk, especially during critical phases of the season. This evidence had previously been reported for other team sports [12], [35]. The training load monitoring strategies were semi-professional contexts, followed by elite and university contexts. The studies were conducted on adult and youth male basketball players, and any application to female basketball contexts has been reported. This highlights a significant research gap, emphasizing the need for more studies focusing on women’s basketball to ensure equitable insights across genders. Women’s basketball training load monitoring continues to be under-explored compared to men’s basketball [36], [37], [38].

The training load variables presented in this systematic review are commonly used in several team sport environments such as basketball, specifically the perceived exertion (e.g., RPE, sRPE) [37]. These metrics provide valuable, accessible tools for coaches to monitor and adjust training loads, especially in resource-limited settings. It was significantly correlated with two HR-based methods in professional basketball players and in football, showing to be more significant with external accelerometry markers [20]. Berkelmans et al. [39] stated that individualized HRmax based on field-based tests supplemented have the greatest accuracy in heart rate response when calculating the summated heart rate zones. The training load indices based on sRPE are also linked to injury risk and fatigue status in women’s basketball players [36], [40]. Also, well-being status should be considered when monitoring the training load [4], [41]. Total weekly training and acute-to-chronic workload ratio demonstrated high week-to-week variation (i.e., spikes up to 226 % and 220 %, respectively) [41]. Such variations underline the importance of closely monitoring acute workload spikes to prevent overtraining and injury, especially in athletes transitioning between competition and rest phases. According to Scanlan et al. [9], elite basketball players performed significantly more total movement changes and experienced greater activity workloads (i.e., jogging and running) [9]. Players’ starting status also represents a key independent variable for training load, considering that starting players experienced a higher total weekly and similar well-being status compared with non-starter players [41]. Multicollinearity has been reported for perceived effort and heart-based measures [37], [12].

A position-specific preparation has been emphasized by several authors [1], [42], [43]. Williams et al. [43] reported that the backcourt positions experienced similar demands amongst training and games, reporting greater demands across all variables during games than training for frontcourt basketball players. These findings suggest that tailored training programs are essential to meet the distinct physical demands of different positions, potentially enhancing performance during competition. Moreover, short-term fluctuations in performance should be equally considered to promote longer-term improvements [44]. In youth basketball, the specificity of the positions must consider the players’ maturation conditions [1], [42] and the training periodization [30], [31]. Specific training tasks, such as small-sided games (SSG) with differential tactical aims and regimes, can influence external and internal demands in basketball [45]. These results are in agreement with previous reports in training load monitoring [46], [47].

The ability to rapidly change direction and jumping performance are key components in basketball [43]. Neuromuscular physical performance composes important information for better training planning that is more effective in improving player performance [48]. Studies conducted on the topic used COD testing in basketball players, and their results did not allow definitive conclusions to be drawn [44]. According to Ferioli et al. [48], the preparation period may induce negligible changes in the vertical jumps (i.e., CMJ), when there is an improvement in the ability to sustain repeated COD efforts. Otherwise, Heishman et al. [42] described insignificant differences between the guard and forward/center positions. These inconsistencies point to the need for further studies exploring neuromuscular adaptations in basketball players across positions and performance levels.

Current research established a window of opportunity for training monitoring the following considerations: 15 weeks, at least 247 overall training sessions, and 2 to 12 training sessions per week. The literature considers that effective monitoring should consider at least one week of continuous control of the training load [5], [12], [49]. This highlights the importance of adopting consistent monitoring protocols, ensuring actionable insights for coaches to optimize load management.

The present review has some limitations that should be acknowledged. Firstly, the low number of studies directly addressing external training load metrics in basketball limits the ability to generalize findings across different competition levels, genders, and age groups. Secondly, most studies included in the review focused on male players, leaving female basketball largely underrepresented. Most critically, the review relied solely on PubMed as a search source. PubMed, while comprehensive for biomedical literature, does not index a wide range of applied sport science research commonly found in databases such as Scopus, Web of Science, and SPORTDiscus. This narrow database selection likely contributed to the identification of only seven eligible studies, meaning the present synthesis cannot be considered fully representative of the available evidence. Consequently, the conclusions drawn should be interpreted with caution, and the finding that only a limited number of studies exist may reflect search constraints rather than a true absence of research. Additionally, methodological heterogeneity among the included studies, particularly in the use of different training load metrics, impairs direct comparison between findings. Additionally, the exclusive use of the PubMed database in the systematic search process may have contributed to the limited number of studies identified, as other relevant databases might contain complementary evidence. Finally, the review did not consider contextual factors such as psychological or environmental influences on training load, which could provide additional insights into load monitoring. Future research should address these gaps to improve the understanding and practical application of training load metrics in basketball.

5. Conclusions

The present systematic review provides information on training loads in basketball players. Current research suggests that variation in perceived exertion, HR-based measures and time motion analysis were de most applied training load metrics to quantify internal and external load in Basketball. Most of the data have been taken from studies for semi-professional, adult, and male basketball players. However, given the restricted database coverage and the limited number of included studies, these findings should be interpreted with caution. Broader database searches are recommended for future systematic reviews to ensure a more comprehensive synthesis of the available literature.

The findings of this systematic review highlight key practical applications for basketball coaches, sports scientists, and practitioners in optimizing training load management. Internal (e.g., HR, RPE) and external (e.g., accelerometry, GPS) training load metrics can be effectively used to tailor individualized conditioning programs, minimizing the risk of overtraining and injuries. Position-specific demands should be considered, as backcourt and frontcourt players experience different physiological and mechanical loads, necessitating tailored training approaches. Monitoring acute-to-chronic workload ratios and well-being indicators can enhance injury prevention strategies and recovery management. The integration of small-sided games (SSGs) provides a sport-specific method for modulating external and internal loads while maintaining game-like intensity. Furthermore, the current literature highlights the need for more research on female and youth basketball players to establish evidence-based monitoring strategies for these populations. Wearable technology, such as inertial measurement units (IMUs) and heart rate monitoring systems, offers valuable real-time feedback, improving load optimization and player performance. Future studies should aim to refine periodization models and explore multi-sport comparisons to enhance training load management in basketball and other team sports.

\