Over the course of the season, it has been noted that our offense, while good on average, appears to be "fragile", an idea that subjectively seems to be an issue, though it may not be readily quantifiable. I would like to explore the idea more, but before trying to quantify it, I think it would be helpful to try to define what fragility means. I can see it defined one of two ways. The first is an offense that generally plays well, but is prone to bad games. I might separate it a bit from a team that plays inconsistently (that's why although I think looking at variance in offensive output, however you measure that, is on the right track but probably not the right approach). The idea would be a team where most games are good or very good, but some games are markedly deficient. One way to do this would be to compare the output in the median game to the worst performing games (you pick the median game, not the average, because it would not be affected by the low performing outliers). The other question is what would your offensive efficiency metric be. I'm kind of partial to points per drive, though simple stats like yards per play, total yards, or total points are more accessible, and more advanced stats (like offensive efficiency) are available. I would think, however, that whatever stat you chose, you would not want to adjust for opponent strength (or opponent defensive strength). You could imagine an offense being considered "fragile" if it feasted on poor defenses, but was shut down by good ones; adjustment for defensive quality would tend to modulate that effect.
The other way to look at fragility is on a drive by drive basis. In this sense, a fragile offense is one that has a lot of three and outs, even if it is also able to rack up yards on big plays (or on sustained drives). So I thought one way to measure this was to look at the proportion of the time that you have a first and ten and can advance the chains (or score). A "boom and bust" offense, as we saw earlier in the year, would score poorly. Thus, a three-and-out, three-and-out, 75-yard-TD run, would score a 33% on this metric, while a long drive that racks up 4 first downs before stalling out at the opponent's 28 and gets a 45-yard-field goal would score an 80% on this metric (4 first downs out of five sets of downs). The problem I see for this is that it would fail to differentiate a "fragile" offense from one that was just plain bad.
Anyway, I'm willing to do some research into these metrics, if people think they'd be helpful, but I'd like feedback first. Is the concept of offensive fragility helpful? How is it best measured?