There's no way to spin the results if you do the calculations correctly. All you can say with small sample size is that the results you obtained are unreliable. Let me explain how to do the calculations correctly and you'll see what I mean. In post #12, using the 77 games Tannehill started, I said that 46.75% of his games at least 21 points were scored, 57.14% of his games at least 18 points were scored and 87% of his games at least 11 points were scored. OK, now remember that this season we gave up 17, 20, 20 and 10 points, in that order. So, looking at only one game at a time, let's say the probability of winning game 1 is P1 = 0.5714 (getting rid of the % sign means dividing by 100), the probability of winning game 2 is P2 = 0.4675, and of course P3 = 0.4675 and P4 = 0.87. So.. let's explicitly do two calculations: the probability of winning precisely 4 games, and the probability of winning precisely 3 games. The probability of winning precisely 4 is simply P1*P2*P3*P4 = 10.87% as stated in post #12. The probability of winning precisely 3 games requires a bit more thought. There are 4 possible ways to win precisely 3 games: either win games 1,2,3 and lose game 4 OR win games 1,2,4 and lose game 3 OR win games 1,3,4 and lose game 2 OR lose game 1 and win games 2,3,4. OK? So what is the probability of that first possibility: win games 1,2,3 and lose game 4? It's simply P1*P2*P3*(1-P4) because (1-P4) represents the probability of losing (technically: not winning) game 4. In probability theory, the word "OR" means "addition", while an "AND" means "multiplication", at least if events are all independent of each other which we can assume here. So the probability of winning precisely 3 games is: P1*P2*P3*(1-P4) + P1*P2*(1-P3)*P4 + P1*(1-P2)*P3*P4 + (1-P1)*P2*P3*P4 = 34.53% as shown in post #12. As you can see, there is only ONE way to do the calculations (correctly) so you can't spin this. Sample size comes into play because you have to estimate the distribution from which you get these probabilities, and the smaller the sample size the greater the potential error in your estimates of P1, P2, P3 or P4, and those errors propagate through the calculations. That's what I was referring to in posts #14 and #18. But you can't spin the results. All you can say is the results are unreliable with small sample size. That's also why I said in post #18 that IF the small sample size was truly representative, then you'd most likely get 3-1 in the situation described.