I think the reason for this, is that strategies that are older than a year have a way to measure annual return, whilst those that are not have to be measured some way as well. One cannot rely on both cumulatives or both annuals. For one reason, cumulative measurement would deny the algorithms with a long history to show their progress (but also regress) from their past years. Low-risk algorithms that have been here for years rely on that. Cumulative on younger algorithms would seem smart, if those did not blow up before their one-year anniversary. They only look good on scrappers and high-risk algorithms. Next, annuals would be impossible, as it would mean that young algorithms must be predicted upon by the data given. This would mean that a 3-month old algorithm might be wrongfully multiplied by 4. Most algorithms with high risk start off very well and therefore would imply generally high returns, but from experience, most fall within 6-8 months. That would be offsetting potential customers.
So… No, there is no good way to settle it.