METHODOLOGY
EXPLAINED
Given the small sample size, M2 Surface uses a weighting methodology that reverse ranks the data-point submissions based on their relative mean deviation. With reference to the number of submissions, data-points with the lowest absolute mean deviation are weighted highest.
EXAMPLE (Theoretical)
Control
Mean: 19.03
M2 Surface: 19.00
A difference of -0.03.
Outlier
‘Institution F’ submits a data-point with a mean deviation of 2.72.
Change of mean: -0.31 (19.03 vs 18.72)
Change of M2 Surface: -0.13 (19.00 vs 18.87)
Rejection Method (Old)
Original mean change: +0.23 (19.03 vs 19.26)
By rejecting the outlier, 1/6th (17%) of available data, the new mean 19.26 over-values the data-point by +0.23; the rejection method results in significant valuation swings.
CONCLUSION
Given an illiquid market, M2 Surface recognises that institutions may over- or under-value their data-point submissions relative to other institutions. Rather than further reduce an already small sample size, M2 Surface weights the most accurate submissions accordingly. This results in a more consistent and therefore more reliable set of results, or consensus.