Scouting for the amateur draft is an incredibly difficult exercise for a whole host of reasons. One of the most well-documented ones is the relative age effect (RAE for short).

Summed up, the idea is that within any given age cohort, relatively older individuals tend to hold an advantage over their younger peers. This has been observed almost everywhere it's been studied, including within hockey and other sports along with standardized test scores in school.

So, within a given group of prospects, like those taken in the 2023 NHL Draft, the older individuals will hold an advantage over their peers because of the extra time they've had to develop their game.

Take Connor Bedard and Adam Fantilli, for two noteworthy examples. Bedard and Fantilli were both part of the same draft class and even played on the same team at the 2022 World Juniors and 2022 U18 World Hockey Championships, but Fantilli was born in October 2004 whereas Bedard was born in July 2005.

So when comparing the performance between the two players, even if they were equally as good, Bedard would still be the better prospect because he is nine months younger. This difference can be quite meaningful, especially at younger ages. Nine months at 18-years-old is like four percent of your life. That is a significant amount of extra development time.

Now, it can be incredibly difficult to adjust for these differences intuitively. Off the top of your head, how much more impressive does Fantilli have to be relative to Bedard to make up for the nine-month age gap? This question is basically impossible to answer with any degree of accuracy or confidence. It’s not any one scout or team’s fault either – the human brain just isn't well equipped to answer such questions.

As a result of either scouts or NHL decision-maker’s inability to answer such questions intuitively, relatively young players have on average, outperformed their draft position historically. 

So how can we try and properly adjust for relative age differences? We can turn to statistics, which it turns out, are quite good at answering these sorts of questions. At least on average, and when we have a lot of data. Luckily in hockey, we get to observe hundreds of prospects per year. With statistics, we have a few ways to isolate player performance from their relative age, but most people will use regression analysis. 

Regression analysis shows us the average change in some Y variables given a one-unit change in some X variables. So to use regression analysis to help us account for relative age differences, we predict scoring rate as a function of age. This is usually where “Age-Adjusted Scoring” metrics come from.

How does this work? Well, let’s use the first example I ever saw on the subject by Carolina Hurricanes scout Rhys Jessop as an example. Jessop predicted scoring as a function of age among CHL players and found for every year a player ages, we expect their scoring rate to increase by 0.1672 points per game.

Now we can try and answer the question, “How much more impressive does Fantilli have to be relative to Bedard to make up for the nine-month age gap”. Pretend they were both CHL players for the sake of simplicity, and note nine months is three-quarters of a year. Take our 0.1672 multiplied by 3/4 (0.75), and we get 0.1254. With that, we can estimate Fantilli would need to score the equivalent of 0.1254 CHL points per game more than Bedard just to be equally as impressive of a scorer because Fantilli is nine months older.

It’s easy to see how and why this kind of analysis can be useful. Once we know our age adjustment, we can easily apply it and compare players of all sorts of ages. This way we can know who has a more impressive scoring rate independent of their different ages.