Gavin Floyd has been extremely lucky this year. He sports the sixth-lowest BABIP of any starting pitcher (.255), and, not coincidentally, his actual ERA (3.78) has been far lower than his FIP (4.95). In fact, if we use FIP as a standard, only three starting pitchers have been luckier than Floyd this season (Joe Saunders, Shaun Marcum, and Armando Galarraga). Conventional sabermetric wisdom suggests that Floyd is very likely to regress.

But if we look more closely at Floyd's season, we can see that the "conventional wisdom" is misleading. Yes, Floyd's overall numbers are indicative of good luck, but a closer look yields two interesting results: first, Floyd has improved significantly over the course of the season; second, Floyd has already regressed.

Gavin Floyd was incredibly lucky during his first nine starts this year. This stretch culminated in a complete game where he allowed three runs against the Angels, actually raising his ERA to 2.93 at the time. Despite this stellar ERA, Floyd had a miserable 25/27 K/BB ratio (and struck out more batters than he walked only twice in his first nine starts). Floyd was not a ground ball pitcher, but he had allowed only 6 homers in 58 innings – not an incredibly low rate overall, but much lower than he should have allowed. Most importantly, Floyd sported an absurd .172 BABIP during this stretch. 

Here are some more stats from Floyd's first nine starts:

Floyd followed up his complete game on May 23 with a relatively poor outing against the offensively-challenged Indians on May 28, in which he allowed five runs (four earned) in six innings. However, one important thing changed: Floyd struck out seven, while walking zero. Up until that start, Floyd had walked at least two batters in every single start, and hadn't struck out more than four in any start. 

His May 28 start in Cleveland would prove to be the start of a trend. Floyd followed this start up with another start in which he walked zero batters. In fact, since May 28, Floyd has struck out more batters than he has walked in 14 of his 16 starts (remember, he only did this twice in his first nine starts). While his walk rate has remained high, it has come down from where it was at the beginning of the season. Furthermore, his walk rate has been offset by a drastic improvement in strikeouts.

In his last 94 innings, Floyd sports a much-improved 85/34 K/BB ratio, thanks to an improvement in the overall amount of strikes he has thrown. Floyd has given up 17 homers during this span – a very high total which can partly be attributed to bad luck. His BABIP has been completely reasonable (.292), and his 4.31 ERA has been better than league average. 

Here are some stats from Floyd's most recent 16 starts:

It appears that the only kind of luck in Gavin Floyd's last 94 innings is bad luck: the rate at which his fly balls are becoming homers is rather high. Other than that, his line looks very normal, and can easily support his 4.31 ERA during this span. 

There is no doubt that Floyd's early-season success was driven almost entirely by an unsustainably low HR/FB rate and absurdly low BABIP. Since then, however, Floyd has significantly improved his strikeout rate and lowered his walk rate. Thus, his luck has run out, but he has made up for it by improving his skills.

So what are we to make of Floyd's season, overall? I suppose it depends on how much weight you give to Floyd's apparent in-season improvement. We cannot simply pretend that Floyd's first nine starts never happened; Floyd pitched terribly, and should have had a MUCH higher ERA to show for it. But, looking forward, Floyd has made significant improvements in his game, and if he can maintain these improvements, he may very well last as an above-average major league pitcher.