Understanding Cricket Statistics: What Batting and Bowling Averages Really Tell You

What batting average actually measures: the batting average formula is: total runs ÷ number of dismissals. The 'dismissals' denominator (rather than 'innings') means not-out innings are excluded from the count — they contribute runs but not a dismissal to the calculation. This creates a known inflation: players who frequently finish not-out (typically lower-order batsmen, players in high-scoring partnerships where declarations come, or batsmen in successful run-chases) see their averages inflated relative to players who bat through their innings until dismissed. Examples of not-out inflation: (1) Michael Bevan (Australia limited-overs) averaged 53+ in ODI cricket with an extraordinary number of not-outs — his ODI average is considered inflated by not-outs (he frequently finished games not-out after match-winning partnerships). (2) MS Dhoni's ODI average (50+) includes a significant number of not-outs at the end of successful chases — his impact in those chases is unambiguous, but strictly speaking his 'dismissal rate' would be different. (3) Lower-order Test batsmen (Nos. 9-11) frequently accumulate not-outs as last men standing in declared innings — their averages often look artificially inflated relative to their actual batting quality. The position adjustment problem: a batting average of 40 for an opening batsman (facing the new ball, hardest conditions, from the first delivery) is more impressive than 40 for a No. 6 (arriving with the ball older and easier, pitch familiar from watching, the team's position partially established). Raw averages do not adjust for batting position — they treat all 40s as equal.

Bowling Average — What It Tells You and What It Doesn't'

Bowling average formula: total runs conceded ÷ total wickets taken. Elite bowling averages: under 20 is extraordinary (McGrath 21.64, Walsh 24.44, Marshall 20.94, Anderson 26.45 from 700 wickets). The bowling average reflects cost per wicket — a low average means the bowler takes wickets cheaply. The context problem with bowling averages: (1) Team quality: a bowler for a very strong team (Australia 2000s) has catches taken by world-class slip fielders — edges that would be dropped by weaker fielding sides contribute to the strong-team bowler's wicket count. A bowler for a weak team may generate the same edges but see them dropped — same balls bowled, fewer wickets, higher average. (2) Format: McGrath's career average is from Tests — it doesn't include ODI data. A bowler with excellent Test average but expensive ODI economy may appear less impressive if only the Test figure is quoted. (3) Pitch conditions: a spinner who plays all their cricket on spinning subcontinental pitches (more wickets at lower cost) will have a lower average than an equivalent spinner who plays on slower English pitches that don't turn. The pitch condition adjusts the bowling average without the statistic noting it. (4) The tail-wicket inflation: some bowlers are more effective at dismissing lower-order batsmen (tail-enders) than top-order — their wicket count includes proportionally more 'easy' wickets from Nos. 9-11. A bowling average doesn't distinguish a wicket taken against Brian Lara from one taken against a No. 11.

When Statistics Mislead — The Selective Data Problem'

Common ways cricket statistics are selectively presented: (1) Small sample size: a player who averages 60 from 10 Test innings is interesting but statistically fragile — one or two ducks can dramatically reduce the average. A 50+ average from 100+ innings is far more statistically stable. (2) Home vs away split: a bowler averaging 20 at home but 35 away looks very different when the average is quoted without the split. The combined 27 average sounds excellent; the home-only reliance is a significant caveat. (3) Opposition quality weighting: averaging 40 against Australia and South Africa is more impressive than averaging 40 against Zimbabwe and Bangladesh. Raw averages don't weight by opposition quality. (4) Era differences: comparing Bradman's 99.94 (1928-1948) with modern averages requires acknowledging different pitches (uncovered pitches were more variable), equipment (no helmets), and field restrictions (none). The statistic is still extraordinarily impressive but the era context matters. Modern metrics that improve on raw averages: (a) Runs Above Replacement (RAR): compares each player's scoring rate to what an average player at the same batting position would score in the same conditions. (b) Bat-speed adjusted averages: some modern analytics systems account for the era of play. (c) Opposition-weighted averages: weight each innings by the quality of the opposition's bowling attack. None of these have displaced batting average as the public-facing metric because the newer metrics require proprietary data and complex explanations — 'batting average' is simple and immediately interpretable even if imperfect.