In Part 3 we explained how to identify the line (in our data) between real and fake Yeezys by use of a “minimum”. A similar exercise was done to eliminate auctions with multiple pairs by use of a “maximum”. The minimum was calculated to be $1,000 (inclusive) and the maximum was $3,250 (inclusive). Any sale of a Yeezy 2 which did not fall within that range was excluded from our data analysis.

With a confirmed minimum and maximum we re-calculated the Campless data points for the Yeezy 2, like average price and volume. We can now compare them to the statistics we calculated in Part 1, which were calculated without using a min or max (i.e., including fakes and auctions with multiple pairs).

To refresh your memory, here are the original stats for DS Yeezys:

**Yeezy 2** **Stats (including outliers)**

- Total Pairs Sold (DS): 4,944
- Average Price Sold (DS): $1,035
- High Price Sold (DS): $90,300
- Low Price Sold (DS): $0.01
- Total Pairs Unsold (DS): 3,284
- Average Price Unsold (DS): $1,058

Before we show the new stats, take a look at those numbers and ask yourself if they make sense. Logically, does it make sense that the average price of a Yeezy 2 – a real Yeezy 2 – is $1,035? Is $1,035 a reasonable price when the average price is $2,489 at Flight Club (available before they removed historical pricing)? Even if you were on the fence about $1,035 – what about $90,300 or $0.01? It takes neither a sneakerhead nor a data scientist to agree those data points are suspect.

So now let’s look at the new numbers, side by side with the old:

**Yeezy 2** **Stats (including outliers –> excluding outliers)
**

**Total Pairs Sold (DS): 4,944 –> 2,182****Average Price Sold (DS): $1,035 –> $1,732**- High Price Sold (DS): $90,300 –> $3,250
- Low Price Sold (DS): $0.01 –> $1,000
- Total Pairs Unsold (DS): 3,284 –> 1,375
- Average Price Unsold (DS): $1,058 –> $1,941

The two key insights to call out here are 1) the decrease in pairs sold, and 2) the increase in average price.

1) The first thing we notice is that the total number of DS pairs sold has decreased by more than half, from 4,944 down to 2,182. This seems right; we expect there to be more sales of fake Yeezys than real ones bc a) fake Yeezys cost less; and b) fake Yeezys are not limited in quantity like real ones.

2) The next thing we notice is that the average price went up about 70%, from $1,035 to $1,732. This not only seems right compared to the Flight Club prices (see Part 2, second to last paragraph, for detailed comparison to Flight Club), but is also almost exactly what we predicted the average would be when we applied the normal distribution curve to the Yeezy 2 histogram in Part 3 (first paragraph after the below chart in Part 3).

So there you have it.

If we were to accept the eBay data as is, without taking a critical look and removing outliers, we might report that the average price of a Yeezy 2 is $1,058. And you would think we’re crazy. But, based in no small part on our detailed approach to sneaker data, we are able to correctly report that the average price is $1,732.

Ummmm…. so, in brief, you look at the data and see two maxima. You fit the normal distribution on the higher maximum, remove the outliers based on this normal curve (which removes the lower maximum), and so your new average (without outliers) is the same as the center of the normal distribution. Well, duh! How is that a key insight? That’s exactly what you get because you removed the data that didn’t fit your normal curve. The real question (I don’t know the answer, and I’d like to know) is how did you know which distribution to fit and how to fit it in the first place?

Thanks. I’m glad that there are sneakerheads that can think with their heads.