by Bob Smith & George Russ
We want to know the likely demands of both a competitive season and the training required to maintain skill and relevant fitness qualities throughout the season, This dataset is from the 2017 season. This analysis will provide the master-number that gives every bowling loading programme the perspective as to how far we are off full 1st Class Cricket Workloads. The number of overs required for Championship Cricket and the likelihood of workload exposure over days/ weeks/ months and seasons.
Given the infinite workload scenarios that we can expect to see in Championship cricket it is necessary to display this in a different way to the one-day matches which are capped at 10 and 4 overs for the 50 Over and T20 Formats respectively. George created a workbook of all the match-overs bowled throughout the country in the 2017 Championship season. We then created a filter for players characterised as fast bowlers, fast-medium, medium and spin as we anticipate that these players are used differently in matches so it is of little use to lump them all into the same analysis. We will analyse fast and fast-medium in one category, medium pace in a second category and spinners in a third category.
Fast and Fast-Medium Bowlers:
5218 overs were recorded across X fast and fast-medium paced bowlers in the 2017 season. The distribution of daily workloads (on days where there was bowling) is presented in figure 1
.Figure 1: Incidence of overs bowled per day in a Championship matchday
The average is 10.0 (SD 5.47, range 1.2-28.2) overs per day. There are a few limitations in this basic analysis that make it miseleading for our needs analysis:
- Often an innings will be interrupted by the end of play for that day or bad weather and therefore that high workload is split between 2 days.
- To create a practical guide for developing a bowling workload system, the average bowling load will be exceeded 50% of the time and therefore place players at some risk for half of the bowling days. Just logically preparing players for 10 overs per day is too low.
To address the first limitation we will present the incidence of number of overs bowled in innings 1 and innings 2. This analysis will provide us with the distribution of workloads across all fast bowlers without considering if there was an overnight stoppage in the middle of that innings.
.Figure 2 shows the incidence of overs bowled in the first innings of Championship matches in 2017 by all fast and medium-fast pace bowlers.
282 First-Innings were bowled across all fast and fast-medium paced bowlers in 2017. The average number of overs bowled in the first innings is 17.0 (SD 6.8, range 11.5-34.5).
Figure 3 shows the distribution of number of overs bowled in the second innings of Championship matches. Fast and medium-fast paced bowlers participated in 255 second innings. The average number of overs bowled is 12.7 (SD 6.5, range 1.2-34).
Figure 3: The incidence of overs bowled in the second innings of Championship matched in 2017 by all fast and medium-fast paced bowlers.
The average number of overs bowled in a match is 26.0 overs (SD 10.5, range 1.0-54.0) per match.
Figure 4: The incidence of number of overs bowled in a Championship match combining 1st and 2nd innings.
We thought there might be a relationship between the number of overs bowled in the first innings and the number of overs bowled by the same player in the second innings of that match. Figure 5 shows every match overs for each fast and fast-medium bowler for the 2017 season. There is massive variability around the mean for match (1-54 overs), the 1st innings (0-34.5 overs) and the 2nd innings (0-34 overs). There is no correlation between overs bowled in the 1st innings and overs bowled in the 2nd innings (R2=-0.041).
Figure 5: Match overs and contributing 1st and 2nd innings overs for each player in each Championship match in 2017.
This basic descriptive analysis provides us with some useful information but cannot underpin our bowling load system. It is not low workloads or even average bowling workloads which are likely to lead to under-recovery and injury in our players. It is the exposure to infrequent high bowling workloads delivered in a single bout of extreme exercise or a dense period of bowling that they are unaccustomed to. We need to find a way to better represent potentially harmful workload in our data set.
Through the analysis of daily, 1st innings/ 2nd innings and match overs we have presented average (range) showing a large amount of variance amongst demands of cricket bowling. The end goal is to inform our bowling workload surveillance and training prescription to provide information as to the magnitude and frequency of workloads that may exceed the player’s current capabilities. The average is obviously made up of all data, which is given an even weighting. However we know it is the high relative workloads that we are interested in preparing players to be able to cope with. But how high do they need to be? We need to better represent potentially injurious workloads by giving greater weight to high workloads which we think have a greater likelihood of leading to injury. One such method (adapted from cyclic endurance sports) is the Normalised Average calculation (Allen and Coggan, 2010). This calculation involves taking the raw data, raising it to the fourth power (in order to emphasis larger numbers), take an average of this data, then taking the fourth root of that score to provide a new average. These authors did this to take into account the disproportionate relationship between intensity and fatigue in cyclists undertaking a hill based interval session. The average and normalised average scores are shown in figure 5.
Figure 5: Comparison of bowling overs for 1st, 2nd and 4 day Match calculated as an average of raw data or as a Normalised Average.
We feel this goes some way to better represent workloads that are more likely to cause under- recovery and injury. This may form a more realistic picture of the magnitudes of chronic single day and 4-day workloads that we should be maintain through Championship cricket periods.
Please note, all averages quoted after this section are not Normalised Averages. Normalised Averages are displayed to represent a method of better representing high workloads.
Weaknesses of a ACWL using total workload to analyse fast bowling:
We propose that the same method should be applied to Total Workload Calculations in cricket bowlers. Research into the ACWL has commonly used total workload (workloads collected from all types of activity which are ascribed an intensity metric and collated together) to calculate acute and chronic workloads. The issue with doing that for a cricket innings is that fast bowling spells are interspersed with periods of very low intensity fielding activity. Using the ECB RPE rankings (unpublished data) for cricket activity we can create a daily workload but it is misleading.
Championship fielding = RPE 3.6 Fast Bowling = RPE 9
Player A participates in 96 over day of fielding in which he also bowls 26 overs (70 overs of fielding + 16 overs of bowling. According to our workload calculations:
70 x 4.5mins x RPE3.6 = 1134 au
26 x 4.5mins x RPE9= 1053 au
In this example the low intensity fielding workload accounts for more than 50% of the daily workload but we know that it is the bowling workload that is responsible for the majority of fatigue and injury risk associated with the day. It is far higher intensity, will create a greater amount of fatigue and take longer to recover from than the fielding workload. It is difficult to quantify and compare different modalities of exercise and the physiological impact that they induce (both positive and negative). Bannister’s RPE system goes someway to addressing this in that the 1-10 scale is non-linear in it’s ratings (i.e., 1-4 is degrees of easy and moderate-hard, 5 is hard 6-10 are scales of hard to impossible) so it shows higher workloads are disproportionately more taxing than lower workloads. We need to find a method of displaying the data to account for the massive disparity between low and high workloads in cricket.
Distribution of workloads across a 4 day match:
There are countless scenarios of how workloads can be distributed across a 4 day match, the number of overs bowled in each and how many days between innings. In order to best
prepare players for all extreme scenarios we also wanted to know how innings were distributed across a 4-day Championship match. This is important as it shows us the patterns of workloads that we should expose our players to in a bowling workload training plan or a return to play programme to ensure the player is well prepared for the variety of bowling scenarios that the game could provide.
Bowling on one day of a Championship match occurs 16% of the time and averages 13 overs. Bowling on 2 days in the match on consecutive days (day 1 and 2, 2 and 3, 3 and 4) occurred 32% of the time and averaged 21 overs. Bowling on 3 consecutive days (1-3 or 2-4) occurred 32% of the time and averaged 32 overs. Bowling on each of a 4 days occurred in 10% of Championship matches averaging 36 overs. Bowling loads representing what we might imagine as a stereotypical distribution of bowling (day 1 and 3 or 2 and 4) occurs only 7% of the time averaging 21 overs. Bowling 3 out of the 4 matchdays with a day off between the 1st and 2nd innings (1,2,4 or 1,3,4) occurs on 3% of occasions averaging 30 overs.
Figure 6: The distribution of workloads across a 4 day match.
Building a System
The next step is to run a series of Bowling Workload scenarios and see how much risk we are likely to encounter with a given chronic workload. Blanch and Gabbett 2015 presented a polynominal relationship (figure 7) between the acute chronic workload ratio and likelihood of subsequent injury using data from cricket, rugby league and Australian Rules Football. It is important to point out some of the differences and departures for our next step up of analysis and how we might be able to address these limitations in the future.
Figure 7: Blanch and Gabbett, 2015’s representation of the relationship between acute chronic workload ratio and likelihood of subsequent injury.
It is important to point out some of the differences and departures for our next step up of analysis and how we might be able to address these limitations in the future.
- We will be ascribing risk to acute-chronic workload ratios using the exact equation and curve proposed by Gabbett and Blanch despite this being derived from 3 different sports and not just cricket. We will be doing our own injury vs bowling exposure analysis on the data we have available (that of our own players in the last 3 years) but accept this might not be a large enough cohort. What we would need is all injury data across all counties as well as workloads in training and matches but at this stage this is not a possibility.
- Gabbett and Blanch applied their analysis to total systemic workloads across all 3 sports. We will be using the same curve to ascribe risk to bowling volume.
- Bowling intensity is assumed to be fixed in all competitive matches which is unlikely to be true.
We will use the bowling data from the 2017 season to create a series of charts that plots the likelihood of exposure to different daily, 1st and 2nd Innings and Match overs alongside risk of injury with a given chronic workload. For the sake of practical simplicity (and replicating the methodology of Carey et al., 2018) we will create buckets of 25% ACWL ratios and ascribe a median amount of risk to that bucket (100-125%, 125-150%…). Figure 8 shows the incidence of total overs bowled in Championship match and the extent to which that would exceed a 4 day chronic workload built up through training and matchplay. This provides us with data as to the likelihood of exposure to different ACWL depending on how much chronic workload (4 day) we are willing to build through the off-season.
Figures 8: The incidence of ACWL ratios with a range of chronic load tolerances (4 day) leading into Championship matches.
Example 1 (referring to figure 8):
In the 2017 season, if a fast/ fast-medium bowler has developed a 4 day chronic load of 20 overs leading into a match the likelihood of exposure to a Championship match which required so many overs as to produce >100% ACWL is 83%. 29% of these matches would have fallen into the sweet spot of ACWL (100-125%) which is associated with a subsequent risk of injury of 4%. 32% of matches would create an ACWL of 125-150% (5% risk), 9% with an ACWL of 150-175% (8% risk) and 7% with an ACWL of 175-200% ACWL (12% risk). 6% of matches would have exceeded 200% ACWL with a subsequent risk of injury of 6%.
This provides us with data as to the likelihood of exposure to different ACWL depending on how much chronic workload we were willing to build through the off-season and maintain into the season beyond. It does not give us a definitive answer as to how much chronic load each player needs going into a Championship match rather we believe it is up to the coach and player to strike a balance between how much chronic load they are willing to do the work to build vs the magnitude of exposure to workloads in excess of their hypothetical load tolerance and associated risk of injury.
It could be suggested that building the highest chronic workload possible would leave the lowest possibility of exceeding an ACWL ratio of 1.25 and minimising risk of injury through a spike in workload. However, this must be weighted against a contrasting constraint in that building a high workload requires a greater exposure to fast bowling, which in itself comes with greater risk of injury.
Example 2 (referring to table 8):
If a fast or fast-medium paced bowler built a 4 day chronic workload of 40 overs in the 2017 season the liklehood of exceeding 100% ACWL in Championship match was 13%. In each occurrence, the ACWL would not have exceed the ‘sweet spot’ of 100-125% ACWL with an associated risk of 4%.
We must therefore find the balance point of both factors in order to minimise risk of injury in training to build a high chronic workload but still having high enough load tolerance to cope with the majority of bowling demands in the season. This will likely depend upon an array of variables that the coach and athlete must balance. I’m not sure a computer algorhythm can answer this.
Example contributing factors:
- Injury history
- Time of season (may accept greater risk at end of season when wickets are lessbowler friendly)
- Training age
- Chronological age
- Duration of time to build workload (return to play context)
- Player’s role for the team
- Individual’s previous bowling history, how their overs tend to be distributed
- Type of wicket (some grounds are more or less bowler friendly for wickets)Figure 9: Chronic exposure to single innings type scenarios and the likelihood of exposure to workload in excess of 100% of chronic workload.
- Figures 9 shows the different ACWL ratios with different chronic load tolerance (single day) leading into Championship matches. I do not believe that this analysis is entirely valid along ACWL principles. However, I do believe there is validity in exploring very high workloads on a single day or consecutive days as I believe that this has a place in training. By adopting the principles of ACWL, figure 9 shows how an exposure to a range of chronic loads (1 day) leading into a Championship match will prepare them for the variety of workloads expected in the 1st innings of the match. I have left the risk column in for comparison only but it should be considered that this is a very different analysis to Tim Gabbett’s research where this is derived from. The table simply adopts the ACWL framework to display the distribution of 1st innings exceeding 100% of chronic bowling exposure over the same single day timeframe.
Figure 9: The incidence of ACWL ratios with different chronic load tolerance for a single day of bowling in a Championship match.
Example from Figure 9: If a fast or fast-medium pace bowler has a daily load tolerance of 22 overs designed in such a way as to mimic bowling the 1st innings of a Championship match we can expect this workload to be exceeded in 36% of 1st innings of a Championship match. 21% of 1st innings workloads will be between 100-125% of 22 overs, 13% between 125-150%; 1% between 150-175%; and 1% between 175-200% of 22 overs.
I think this is interesting information to inform a training programme. If we have not exposed our bowlers to a simulation of a 1st innings in training with high workloads that they can be expected to perform in a Championship match do we place them at a greater risk of injury?
In the example above if we prepare players to a chronic load of 14 overs, 1st innings simulation in training, 83% of the time we can expect to see the 1st innings of a match exceed this practice workload with 62% of the 1st innings workload in excess of what is considered ‘the sweet spot’ in the ACWL literature.
Summary:This analysis shows the extremes in magnitude, density and frequency of fast and fast- medium bowling. Given the diversity of scenarios and magnitudes of bowling volumes we do not think that there is any one load progression programme that will prepare a bowler for competition. The athlete should be challenged with a range of scenarios that mimic the high workload situations of competition. The magnitude of “high workload” exposure depends upon a balance between the willingness to accept risk from either exposure to injurious workloads that exceed 125% ACWL ratio in competition or the risk of injury from exposure to the greater number of overs required to build such a high chronic workload. What is clear is that having a low bowling load tolerance leaves the athlete increasingly vulnerable to the unpredictable scenarios that the game provides.
Allen, H. and Coggan, A. (2010). Training and Racing With a Power Meter. Boulder, CO: V eloPress.
Blanch, P. and Gabbett, T.J. (2015). Has the athlete trained enough to return to play safely? The acute:chronic workload ratio permits clinicians to quantify a player’s risk of subsequent injury. British Journal of Sports Medicine, 50, 471-475.
Carey, D.L., et al. (2018). Optimizing Preseason Training Loads in Australian Football. International Journal of Sports Physiological Performance, 13, 194-199