While Running Performance and Marathon Level are two related metrics, they are not perfectly correlated. This means that despite it seeming contradictory, it is completely normal and common to see Marathon Level decrease even with a good Running Performance score.
The purpose of the Running Performance score is to compare your measured VO2 Max of your most recent run to your 42-day average VO2 Max measurement. Running Performance represents the performance of a singular run, rather than being a measurement of your overall fitness. This score will have a much higher correlation to how fast (pace) your most recent run was, compared to your average run speed (pace). You can also think of this as an equation (Todays VO2 Max/42-day avg. VO2 Max = Running Performance), this is a short-term, acute, measurement.
You can think of Marathon Level, on the other hand, as a long-term indicator of your overall running ability. Marathon Level is a measurement or indicator measuring how your heart rate reacted to a given pace, and what the trend is for your heart rate (ie: exertion) at a steady/stable pace.
Seeing a change of +/-0.1 or 0.2 to Marathon Level is incredibly common, and is simply a recalibration of the algorithm measuring your overall running ability. Since this is a long-term measurement, it is possible that there were runs within the last six weeks which may have caused the estimation to be higher or lower than expected, and this recalibration is now trying to correct this. In addition, if there was a significant workout (good or bad) that fell out of the 42-day analysis window, Marathon Level may increase or decrease even with an average run - as that workout is no longer included in the assessment. It is more common to see longer runs have a more significant impact to Marathon Level, while shorter runs will play a smaller impact - as shorter runs make it more difficult to estimate overall fitness level as compared to long runs (13mi/21km being long runs).
Ultimately, it's important to remember that these numbers are all estimations. They are entirely dependent upon the "Good data in, good data out" principle. For the most accurate numbers, follow best practice tips for recording heart rate data. While we are highly confident in the degree of accuracy in our measurements, these data points cannot take into account many external factors such as altitude, weather, stress levels, psychological approaches, and more.
For more information about these metrics, please take a look at these articles below: