Leading researcher Ed May has generously offered his ARV software for use by the APP. The software uses a combination of a very carefully constructed photo set of 300 photos, an MS Access database, and the MatLab program. The software generally obtains 80 to 90 percent success when a pick is made, which is an unprecedented rate of success in the field. One group trial has been run in the APP. This was the CAS-OAK series, which had 4 hits, 0 misses, and 16 passes. That is, 20% results (all hits) and 80% passes. This pass rate is somewhat higher than the pass rate Ed May has reported (70%+). A formal pick is made only when the Figure of Merit is above .4519. The FoM is a measure which the computer assigns by comparing the viewer’s session with two prospective targets from the photo set. We are still in the process of obtaining all components of the software for use by several APP groups.
FoM for Target A corresponds to the Over, or a Win by the Away team. FoM for Target B corresponds to the Under, or a win by the Home team. Pick was made only if one Figure of Merit was greater than threshold of .4519.
Official hits (4) have higher ABS FoM differences . Low FoM difference of these is .24. (Trial 13 has a higher FoM difference (.3005)). Columns X to AB were not picks. But picking the higher FoM each time would have resulted in 11 hits and 9 misses. Betting the higher FoM when the difference is >.2 would have resulted in 7 total hits and 1 miss; or resulted in 3 hits and 1 miss if you exclude the 4 actual picks (which were all hits).
- Jon Knowles (CAS-OAK group manager)
Papers by Ed May and associates on which the CAS software is based are available here.