Slightly incorrect. The law of averages says that the most likely result as you roll more and more times is that the average *approaches* 3.5 but doesn't quite reach it. Fine, so then it's a split of single hair. It's hardly a worthwhile application in this context. It's precisely worthy of application. The difference between the expected average after the bad start and the ordinary expected average shrinks over time, and with an infinite number of rolls it eventually becomes infinitesimal. But we don't have an infinite number of games left. We have 160. So we know how big the difference is expected to be over our sample: About one win's worth. But a single win is not a large enough unit to realistically hold any weight. Referring back to your example of 162 coin flips and the first two were tails - yes - I would predict that 80 head flips would follow. However, although that may be the best guess possible, would I realistically expect it to fall on exactly 80? No, because a single number prediction is next to impossible to predict. That is my point. Within a sample of 162, two outcomes are barely significant. So, again, I technically understand your point. The cubs are slightly less likely to finish .500 (assuming that was their projection). However, that percentage is so small that it is not plausible here.