To expand on the “doesn’t exist”, NR pays out per railway period on an average of delays and cancellations over those 28 days. It is expected that there will be a “normal” level of delays and that acts as the threshold above which NR pays, based on the difference between the threshold and the average minutes delay figure for that period.
The rate at which the payment is made is a fixed sum, based on the revenue for that train service group, for that TOC. So there are a myriad of payment rates, all commercially confidential.
The TOC itself has a similar part of the delay regime by which it pays delay costs to NR. The payment rate and threshold will be different as they relate to the effect on other TOCs. All the payment rates and thresholds are usually reviewed and adjusted by the ORR as part of their periodic review process.
The whole regime was designed on the star model principle with NR sitting in the middle. This is to stop TOCs suing other TOCs (and NR) for every single delay. It is effectively a simple liquidated damages regime for revenue loss.