Batting averages are asked to carry more meaning than they can hold

A summary statistic used as a verdict

An average takes runs and divides them by dismissals, which produces a stable and easily communicated summary of how long a batter tends to survive while scoring. The trouble begins when the number is used to rank players who faced different bowling, in different conditions, at different positions and in different eras. None of those differences are inside the calculation, so comparing two averages compares two summaries of situations that were never comparable.

This is not a flaw in the statistic itself, which does exactly what it says, but in the weight placed on small differences between them. A gap of a couple of runs is treated as a ranking when it is well inside the noise created by the factors the number ignores.

How the denominator behaves

Dividing by dismissals rather than innings means a batter who is often stranded unbeaten receives a systematic lift relative to one who is not. That is not cheating, and there is a real argument that an undismissed innings should not be counted as a completed failure. But it does mean that batting position and team role feed directly into the number in a way that has nothing to do with skill.

A lower-order batter who regularly finishes an innings unbeaten accumulates an advantage that a top-order batter can never access. Any comparison across positions therefore has to correct for the denominator before it can say anything at all.

The conditions problem that will not go away

Runs scored on surfaces offering seam movement and uneven bounce are harder to make than runs scored on flat, true surfaces with fast outfields. Averages do not distinguish between them, so a career spent in demanding conditions is penalised relative to one spent in generous ones. Attempts to adjust for this exist, but every adjustment requires choosing which factors matter and how much, which imports the analyst assumptions into the result.

The adjusted number is often more useful and is also less objective, which is a trade that its users do not always acknowledge. There is no version of the statistic that is both fully comparable across contexts and free of judgement about what to correct for.

Why the average survives despite all this

It survives because it is cheap to compute, unambiguous in definition and available across the entire history of the sport in a consistent form. Those are genuine virtues, and no alternative metric has matched them while also solving the context problem the average ignores. The practical result is that a statistic designed as a summary has become the default unit of argument about greatness.

Arguments conducted in that unit inherit its blind spots without anyone having to defend them explicitly. The number is not the problem; the habit of treating a difference in it as a settled question is.

Using it without overclaiming

The average is well suited to describing a single career at a coarse level and to flagging that two players operated in very different ranges. It is poorly suited to separating players whose numbers sit close together, because the ignored variables are larger than the gap being discussed. A reader who wants to know who was better needs innings context, opposition strength and role, none of which the number can supply.

Holding both facts at once, that the statistic is sound and that most of its popular uses are not, is the whole of the discipline here. Everything else is a preference dressed up as a measurement.