 To recap: The box plot below shows the relationship between a collectible’s grade and its selling price.

As expected, when grade increases, price increases. This is a basket of 15 collectibles, with prices normalized so we can see the relationship between price and grade.

As previously discussed, below the grade of 7, the impact of grade on price is not as drastic as above 7.
9 and 10 seem to have an exponential relationship.
We knew this from tracking the market, but now we have a nice visual to show the relationship.

Let’s go one step further with something interesting.

Let’s answer this next question: What does this price/grade relationship look like for low value collectibles vs. high value collectibles?

Again, we used the same 15 cards from our previous test: Each of these has over 1000 public transactions, so we can make a nice model with it.

The range of sales prices are below:

A wide range of sales prices from just under \$85 to over \$7100. Ranges are from PSA 1 to PSA 9.

We normalized prices, as shown in the last post.

Let’s determine the relationship between grade and price for items BELOW \$1000 and items ABOVE \$1000.

``````top15_norm %>%
group_by(spec_desc) %>%
mutate(avg_price = mean(price)) %>%
ungroup() %>%
mutate(price_bucket = ifelse(avg_price < 1000, 'low', 'high')) -> top15_norm

top15_norm %>%
filter(date > as.Date('2016-09-01')) %>%
ggplot(aes(x = as.factor(grade), y = normalized_price)) + geom_boxplot(outlier.alpha = .1) + facet_wrap(~price_bucket)``````

Here are the results:

The first graph is for cards above \$1000.
The second graph is for cards below \$1000.

Cards above \$1000: