Correct! But it isn’t further along in the season. It is the beginning of the season. The current rankings are mostly based on pure bullshit and wishful thinking, plus a huge dash of bias. Given that, the head to head victory absolutely should take priority over….fantasy nonsense bullshit. Currently we have THREE wins that are all better than any Duke win, including a win ver Duke themselves. Anyone ranking them ahead of us isn’t basing it on reality — they are basing it on personal bias.
That can change, further in the season, when there is more data. For now, base it on the days available.
I completely disagree with that.
There's a reason pre season human polls tend to be more accurate in predicting tournament success compared to finalized rankings.
I know that seems counter intuitive but it's true. Humans tend to get worse. The reason is because they weight the most recent results and tend to ignore everything else. Computers tend to get better with more data but something needs to be said about this as well since I've seen this said so many times here........
Many people will look at pre season rankings and say don't pay attention to them. This isn't true. In fact, if you look at the model diagnostics page on Kenpom you see this for example for last year:
Accuracy/bias of predicted margin
| Nov | Dec | Jan | Feb | Mar/Apr | All |
|---|
| MAE (Margin) | 9.24 | 9.00 | 8.71 | 9.05 | 8.95 | 8.98 |
| MAE (Totals) | 12.82 | 12.76 | 12.21 | 12.71 | 13.12 | 12.68 |
| Bias (Totals) | -0.38 | +0.42 | +0.80 | -0.07 | +1.24 | +0.35 |
| Bias (Home Team) | -0.78 | -0.10 | +0.24 | +0.39 | +0.84 | +0.10 |
There isn't a whole heck of a difference between Nov and March here.
Even in this time of the transfer portal where teams are turning over more than ever, you can still make a reasonable prediction even before a game has even been played. Obviously a team like UK with 100% turnover will be harder to assess and thus admittingly by Ken himself thought we were ranked too low because it still has data baked into it from the previous year (obviously pointless for us).
But in terms of overall, you can still very much get a clear picture even in November how teams will do.
