The way most studies of weight loss work is by recruiting a pool of obese applicants. This is intrinsically biased: someone who has had a lifelong struggle with obesity can be recruited across 100% of their lifespan, whereas someone who spent two years obese and then lost the weight and kept it off can only be recruited for that 2 year window, or 2.5%. There are probably other factors that come into play that bias the sample even further.
The question these studies answer is "given a random obese person, how likely is this person to lose weight?" This is a relevant clinical question, and the answer is usually a pretty low percentage. For an individual who hasn't struggled with obesity their entire life, a more pertinent question is "given that I have just become obese, how likely am I to lose the weight again and keep it off?" The chances of that are much higher.
[0] By BMI, which has a pretty big margin for error, but I was visibly overweight.