comparing data with different sample sizes is usually more complicate than just looking at the overall success rate.
for example, if you took a random selection of 8 the 29 games we've played without him there's a reasonable chance you would come up with 0 wins as well.
Or to give another example - if someone has scored their first 10 penalties they have a 100% success rate, however, someone who has scored 49/50 penalties can be argued to be statistically the better penalty taker.