Of course lying and scheming doesn't work. You have to also have diplomatic skills to deflect the fact that you lie and scheme. If you nail somebody, you have to do as much work as you can to keep them and their associates liking you. Big blowhards become targets, like Chill Town in season 2, or are blind to what they need to do, like Howie.
But most of all, the sports analogy breaks down here because of the way luck enters. If the Steelers have bad luck and a Troy Polamalu gets injured, that's a bad break, but in the long run it becomes another test of how well their organization planned for adversity by developing their bench. The Steelers have more than a dozen total contests to prove themselves.
Big Brother is entirely different because it has a lot more luck. The best possible team is two people who mow down everyone to get to the final two and win the entire purse. No matter how you slice it, though, those two people will eventually reach a final three where the third wheel can, by pure luck, limp into a spot in the final two. The final HOH is too much of a random factor. That means the best strategy a 2-person team can have is to work their way to a final three position where they have a 67% chance of winning. But even if you get there, you have a 33% chance of Lady Luck taking you down.
Poker makes a better analogy to Big Brother. You can only play the odds, plus the best players excel at reading other players. In the long run, the best players always win. But even in his best days one sour turn at the river would lose Oil Brunson even the best played hand, maybe on occasion the tournament. In Big Brother, there is no long run, no re-shuffle with another play at the odds. The best players can always play really well, but then have a sour turn at the final HOH and lose the whole enchilada to luck. The best players know they're working within that limitation. So there is a real difference between judging someone on their gameplay and voting based on personal distaste for one of your equals on the playing field.