As CM stated, this topic has been wrung out exhaustively over this thread.
I also agree with CM (because it's clearly the truth) that small changes in the ground rules and assumptions in a simulation can bring about large changes to the outcome.
It's also a question of fact, not opinion, that the F-35 doesn't bring any improvement in firepower, speed or agility to air combat; consequently (and this is backed up by many briefings and statements) its claimed loss-exchange-ratio advantage must stem from stealth and fusion creating a highly asymmetric "see first, shoot first" situation.
Although properly that should be "see first, identify first, shoot first". And that's where the questions arise. DRFM jamming, digital passive EW and selective RCS reduction make LPI radar much more challenging than it used to be. IRST has made some very important advances.
It makes things more complicated, and I think it's ambitious to assume in the 2020s that you'll be able to track an aircraft on radar without being detected.