The Cut Conversion Metric

I want to make a brief aside about the cut conversion metric—the number of players who successfully made the Top Cut with an Identity, as a proportion of the total number of people playing that Identity at the tournament.

I am a big fan of this metric. It leverages the natural mechanics of the swiss pairing system to “solve” a number of otherwise thorny analytical problems—which is a solution that is both elegant and beautiful, and therefore true. However, it is not a metric that is as easy to express as it’s counterpart—the humble win rate.

For a while, I settled for expressing cut conversion as a percentage, but this number is meaningless unless compared to a baseline %—the % of players that will make a Top Cut regardless of ID choice. If you have a 100 player tournament with a cut to top 16 then 16% of players will make the cut regardless of ID—so that’s a 16% baseline. Anything higher than a 16% conversion indicates a strong performance for an ID, anything lower indicates poor performance.

This baseline, of course, changes from tournament to tournament. In order to be able to compare conversion rates across the season, and in the hopes of improving the readability of these articles, I now try to express Identity cut conversion in a different, and hopefully clearer, way.

In our example of a 100 person tournament with a cut to top-16, if 16% of PD made the Top Cut, then that would be exactly average. We can express this as 1x (16/16)—meaning that PD cut conversion followed the baseline exactly as expected. If 32% of PD made the Top Cut, we could express that as 2x (32/16)—that is to say, if you played PD at this tournament, then you were twice as likely to make the Top Cut than the average Corp. Likewise, if only 8% of PD made the Top Cut, that would be a cut-conversion of 0.5x (8/16)—in other words, you were half as likely to make the Top Cut with PD than the average.

Too long—didn’t read? Anything more than 1x is good. Anything less than 1x is bad.

Hopefully this simplifies the expression of a somewhat complicated, but I think important, statistic. If we look at this metric in combination with win rate, then I think together that gives us a pretty solid performance indicator for an ID. If win rate for an identity is above the side average, and the conversion rate for that ID is above 1x, then I would say that ID can be safely considered to be “highly competitive”.