A reasonable proxy for 'contested races' would be those with small margins, say where the winner gets less than 52%. Without an objective measure like that, such a statistic would be rather like 'unforced errors' in tennis, which turns out to be complete bullshit due to its subjectivity.