🔬 condensed matter

Gap distributions between successive personal bests in cricket: Data and Models

This paper analyzes cricket personal best records to demonstrate that career progression and temporal evolution significantly alter gap distributions between successive records, causing them to deviate from classical universal predictions and follow truncated power laws that reflect nonstationary, path-dependent systems.

Original authors: Priyanka D. Bhoyar, Prashant M. Gade

Published 2026-08-04

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Priyanka D. Bhoyar, Prashant M. Gade

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are watching a marathon, but instead of just tracking who is winning, you are obsessed with the exact moments when a runner sets a new personal record for speed. In the world of science, there is a branch called "record statistics" that studies how often these milestones happen. If you were flipping a perfectly fair coin or rolling a fair die over and over, math tells us there is a predictable rhythm to when a new "best" score appears. It's like a clockwork machine: the longer you play, the harder it gets to break the record, and the gaps between new records follow a specific, simple pattern. But real life isn't a fair coin. Athletes get better with practice, they get tired, they age, and the rules of the game change. This paper asks a fun, curious question: Do real athletes follow that simple, boring clockwork rhythm, or does their career have a secret, messy, and exciting story hidden inside the timing of their best moments?

The researchers behind this study decided to crack open the data of cricket, a sport famous for its obsession with statistics. They looked at the careers of top players in three different versions of the game: Test matches (which can last five days), One Day Internationals (50 overs), and T20s (a fast-paced 20-over format). They didn't just look at the scores; they looked at the "gaps." A gap is simply the number of games a player had to wait between one personal best score and the next one. If a player hits a new high score, then waits 5 games to beat it, then waits 20 games to beat that, those numbers (5 and 20) are the gaps.

The paper's main discovery is that real cricket careers are not random. When the scientists measured these gaps, they found a pattern that looked like a "truncated power law." In plain English, this means that while short waits between records are common, there are also surprisingly long waits that happen more often than simple math would predict. The data showed that the "exponent" (a number that describes the shape of this pattern) was around 0.80 for Test matches and 0.84 for One Day matches. This is significantly lower than the "1.0" you would expect if the players were just rolling dice. A lower number means the "tail" of the distribution is fatter, implying that players can go on long streaks without breaking a record, but then suddenly break one again in a way that defies simple randomness.

To figure out why this was happening, the authors played a clever game of "shuffle and see." They took the actual scores of every player and scrambled them, mixing up the order so that a player's early-career scores were mixed with their late-career scores. This created a "fake" career where the player had the exact same set of scores, but no memory of when they happened. When they analyzed these shuffled careers, the pattern changed dramatically. The exponent jumped up to nearly 1.0 (around 0.94 to 0.98), which is much closer to the simple, random prediction. This suggests that the order in which a player plays matters immensely. The specific journey of a career—learning, aging, and adapting—is what creates the unique, non-random pattern of records.

The team also tried to build computer models to see if they could recreate the real-world data just by adding in things like different player skills or different career lengths. They tried models with "heterogeneous" players (where everyone has different abilities) and even models with some time-based connections. However, none of these synthetic models could reproduce the low exponent numbers seen in the real data. This rules out the idea that the pattern is just caused by having players of different skill levels or careers of different lengths. The authors suggest that the real secret lies in the "temporal organization" of the career—the way a player's performance evolves over time, perhaps improving slowly, hitting a peak, or dealing with the changing conditions of the sport.

Interestingly, when they reversed the careers (playing the scores backward from the last game to the first), the pattern didn't fully return to the random "1.0" either, though it did get closer. This hints that the way records are set isn't just about getting better over time; it's about the specific, non-symmetrical way a career unfolds. The paper concludes that cricket players are not just rolling dice; their personal bests carry the fingerprint of their entire career's history. The study suggests that to understand how humans improve and set records, we have to look at the story of time, not just the numbers on the scoreboard.

Technical Summary: Gap Distributions Between Successive Personal Bests in Cricket

Problem Statement
Classical record theory posits that for a sequence of independent and identically distributed (i.i.d.) observations, the probability of setting a new record at the $n$ -th trial is $1/n$ , leading to an inter-record gap distribution $P(g) \sim 1/g$ . However, sporting careers are inherently non-stationary and path-dependent, influenced by learning, aging, injuries, and changing external conditions. While cricket possesses a rich statistical culture allowing for the reconstruction of careers inning-by-inning, the statistical properties of the intervals (gaps) between successive personal best scores have not been rigorously analyzed. The study investigates whether the gap distributions in cricket careers conform to classical i.i.d. predictions or if they exhibit distinct scaling behaviors driven by the temporal organization of performance.

Methodology
The authors analyzed career data for leading run-scorers in Test, One Day International (ODI), and T20 cricket formats, sourced from ESPN Cricinfo.

Data Construction: The study included 108 Test players, 98 ODI players, and the top 101 T20 run-scorers. Gaps were defined as the number of innings between successive personal best scores.
Null Models: To isolate the effects of temporal ordering, the authors employed several null models:
1. Bootstrap-Shuffled Careers: Innings scores for each player were randomly reordered. This preserved the career length, batting average, and full score distribution while destroying all temporal correlations.
2. Reversed Careers: The chronological order of innings was reversed to test the symmetry of record occurrence.
3. Synthetic Models: Simulated careers were generated with heterogeneous player abilities, variable career lengths, and various correlation structures (weak short-range and long-range power-law correlations) to determine if these factors alone could reproduce empirical deviations.
Statistical Analysis: The gap distributions were fitted using Maximum Likelihood Estimation (MLE) to a truncated power law model: $P(g) \propto g^{-\alpha}e^{-\lambda g}$ . Model selection was performed using the Akaike Information Criterion (AIC), comparing the truncated power law against Poisson, log-normal, and negative binomial distributions.

Key Results

Empirical Scaling: The empirical gap distributions for all three formats (Test, ODI, T20) exhibit heavy-tailed behavior that deviates from the classical $1/g$ prediction. The data are best described by a truncated power law with scaling exponents in the range $0.799 \leq \alpha \leq 0.843$ . The truncated power law provided a significantly better fit (lower AIC) than alternative distributions.
Impact of Temporal Ordering:
- Bootstrap-Shuffled Data: When temporal ordering was destroyed, the fitted exponents increased substantially to $\alpha \approx 0.939 - 0.979$ , approaching the classical prediction of $\alpha = 1$ . This indicates that the specific sequence of performances, rather than just the distribution of scores or career length, is the primary driver of the observed scaling.
- Reversed Careers: Reversing the career timeline also resulted in larger exponents than the original data (though not always reaching unity), suggesting that record-setting performances in real careers tend to occur later than expected under a time-symmetric process.
Synthetic Models: Synthetic models incorporating heterogeneous abilities, career lengths, and even temporal correlations failed to reproduce the small exponents ( $\alpha < 0.85$ ) observed in real data. This suggests that the observed statistics arise from complex career dynamics not captured by these simplified stochastic constructions.

Significance and Claims
The paper claims that the progression of personal bests in cricket retains significant information about the temporal organization of a player's career, which cannot be explained by simple stochastic record processes or static score distributions.

Deviation from i.i.d.: The systematic departure from classical record statistics ( $\alpha \approx 1$ ) demonstrates that cricket performances are not independent draws from a stationary distribution.
Career Evolution: The "fatter tails" (smaller $\alpha$ ) in the empirical data imply a higher probability of surpassing a personal best after long intervals compared to random expectations. The authors attribute this to non-trivial temporal structures such as gradual skill improvement, adaptation to changing playing conditions, and the evolution of equipment and rules over time.
Methodological Contribution: The study establishes that record statistics serve as a sensitive probe for non-stationarity and path dependence in complex systems. The use of bootstrap-shuffling effectively isolates the contribution of temporal ordering, showing that career progression involves a specific "memory" or trajectory that simple null models cannot replicate.

The authors conclude that while extreme "record ceilings" (e.g., a player never surpassing a specific high score) exist, the observed heavy-tailed scaling reflects a broader pattern of career evolution where the timing of milestones is intrinsically linked to the non-stationary nature of athletic performance.

Technical Summary: Gap Distributions Between Successive Personal Bests in Cricket

More like this