Monitoring training load (TL) is common practice in team sports with most practitioners using the data to aid training prescription and design. However, these practitioners will know the difficulty and frustration associated with achieving compliance from their players. Try as you might, it is inevitable that there will be occasions where individual player’s TL data will be missing. In this blog post we will look at some practical methods of minimising missing data, the impact missing data can have, and finally a possible approach to dealing with missing data.

Minimizing  Missing Data

As part of our recent study, we interviewed strength and conditioning (S&C) coaches and surveyed amateur Rugby Union players that used a training monitoring system (TMS) across a season [1] (open access: https://www.mdpi.com/2076-3417/10/21/7816). The TMS required the players to record six subjective measures prior to each training session and match on a coloured 1–5 Likert scale (fatigue, muscle soreness, sleep duration, sleep quality, mood, and readiness to train). Additionally, session type and rating of perceived exertion (sRPE) was gathered after sessions. A key finding derived from the qualitative data in this study was that both the strength and conditioning (S&C) coaches and players alike valued the TMS but there were several barriers to its effectiveness.

The greatest challenge for S&C coaches may be getting initial and then continued compliance from players to ensure consistent data. It appears that, for a training monitoring system to be effective, players need to see value in using the monitoring system to the extent that it offsets the burden of using it. A limiting factor highlighted by the coaches was the lack of consistency in the training data gathered, which resulted in difficulty supplying feedback to the players and adjusting the training prescription and design based on the data. Notably, players rated the regularity of both receiving feedback on their data and the coaches using their data poorly, with one player reporting he “would like more feedback on the data collected and this to be discussed very quickly as part of our training”. This causes a quagmire, whereby in order for S&C coaches to act on data it must be consistent, but players must receive regular feedback and perceive that their data is being acted upon in order to give it consistently. For players to engage with a TMS and give consistent data, they need regular feedback and evidence that the data are informing their training. This should aid in achieving consistent data from players and in turn improve the effectiveness of the TMS.

Tip 1: Give regular feedback!

The S&C coach interviews also highlighted the difficulties of gathering sRPE data after matches.  Another important advantage of sRPE is its temporal robustness. Crucially, sRPE has been shown to be temporally robust for up to at

least 24 h post-exercise [2] and a practical solution to alleviate the issue may be for practitioners to collect post-match sRPE the following day when the chaotic nature of match-day is not present.

Tip 2: If data are missing… follow up with players!

A TMS should be as efficient and succinct as possible to reduce player and data analysis burden and in turn, increase the consistency of the data gathered [3]. Consequently, only measures that will inform the coach’s training practices regularly should be included in the system. If you find you are collecting data that is not regularly influencing your training practices, remove it from your TMS.

Tip 3: Only collect data that you regularly use!

Impact of Missing Data

Missing TL data and the resulting incomplete dataset may alter the practitioner’s understanding of a player’s response to training and in turn, the accuracy of training strategies [4]. Without a complete dataset, all other metrics (e.g. weekly changes in TL, acute:chronic workload ratio [5], training monotony and training strain [6]) calculated using these sRPE data would be inaccurate. This also presents considerable challenges for the practitioner in terms of individualised TL prescription. Figure 1 below demonstrates the negative effect of just 5 missing TL data points over a 50 day period on the calculation of the exponentially weighted moving average acute:chronic workload ratio. This may result in a causal sequence whereby practitioners prescribe inappropriate TL and consequently increase injury risk and hinder sports performance [7]. It is therefore recommended that practitioners be cautious when calculating additional TL metrics if they fear there may be some data missing.

Figure 1. The effect of 10% missing data on the accuracy of the exponentially weighted moving average acute:chronic workload ratio (EWMA ACWR). The blue line represents the EWMA ACWR if no missing sRPE data was present prior to its calculation. The red line represents the EWMA ACWR if five of the fifty sRPE data points were missing prior to its calculation.

Tip 4: If you think there may be data missing, be cautious how you use the data you have!

Missing Data Imputation

In our recently submitted study, we explored various methods of addressing missing TL data. We collected sRPE data from 10 male professional soccer players over a 32-week season. Then, data were randomly removed at a range of 5 – 50% in increments of 5%.  We then replaced the missing data using thirteen different methods of missing value imputation (MVI). A simple method of MVI that proved very effective in our study was to use the daily team mean. This is essentially the mean of all the other players’ TL for that particular session. If we take the example from above and replace the 5 missing data points with the daily team mean you can see the effect this method has on the accuracy of the true exponentially weighted moving average acute:chronic workload ratio (see Figure 2 below). When the missing TL values are replaced by the daily team mean, any further metrics calculated will likely be closer to the true value.

Figure 2. The effect of daily team mean missing value imputation on the accuracy of the exponentially weighted moving average acute:chronic workload ratio (EWMA ACWR) when 10% missing data is present. The green line represents the EWMA ACWR if no missing data sRPE data was present prior to its calculation. The blue line represents the EWMA ACWR if the five missing sRPE data points were replaced using the daily team mean prior to its calculation.

Tip 5: If you have a small level of missing data, replace it using the daily team mean!

In conclusion, missing TL data is inevitable. As practitioners our best course of action is to use a TMS that will keep missingness to a minimum and then address any missing data using an appropriate MVI. Hopefully the tips outlined in this blog will help practitioners going forward in addressing this ongoing challenge.


  1. Griffin, A., Kenny. I.C., Comyns, T.M., & Lyons M. (2020). The development and evaluation of a training monitoring system for amateur Rugby Union. Applied Sciences, 10(21), 7816. https://doi.org/10.3390/app10217816
  2. Christen, J., Foster, C., Porcari, J. P., & Mikat, R. P. (2016). Temporal Robustness of the Session Rating of Perceived Exertion. International Journal of Sports Physiology and Performance, 11(8), 1088–1093. https://doi.org/10.1123/ijspp.2015-0438
  3. Halson, S. L. (2014). Monitoring Training Load to Understand Fatigue in Athletes. Sports Medicine, 44(S2), 139–147 https://doi.org/10.1007/s40279-014-0253-z  
  4. Weston, M., Siegler, J., Bahnert, A., McBrien, J., & Lovell, R. (2015). The application of differential ratings of perceived exertion to Australian Football League matches. Journal of Science and Medicine in Sport, 18(6), 704–708. https://doi.org/10.1016/j.jsams.2014.09.001
  5. Griffin, A., Kenny, I. C., Comyns, T. M., & Lyons, M. (2019). The Association Between the Acute:Chronic Workload Ratio and Injury and its Application in Team Sports: A Systematic Review. Sports Medicine, 50(3), 561-580. https://doi.org/10.1007/s40279-019-01218-2 
  6. Foster C. Monitoring training in athletes with reference to overtraining syndrome. (1998). Med Sci Sports Exerc, 30(7), 1164-1168.  https://doi.org/10.1097/00005768-199807000-00023
  7. Soligard, T., Schwellnus, M., Alonso, J.-M., Bahr, R., Clarsen, B., Dijkstra, H. P., … Engebretsen, L. (2016). How much is too much? (Part 1) International Olympic Committee consensus statement on load in sport and risk of injury. British Journal of Sports Medicine, 50(17), 1030–1041.  https://doi.org/10.1136/bjsports-2016-096581


Alan Griffin is a PhD Candidate on the Irish Rugby Injury Surveillance (IRIS) project in the Department of Physical Education and Sport Sciences in the University of Limerick. His primary research interests are in the area of Strength and Conditioning. He is currently undertaking research into the development of injury prevention measures in amateur Rugby Union through the examination of training load. Contact Alan via email at alan.griffin@ul.ie or view his profile on LinkedIn, Twitter and ResearchGate.


Tagged with: