eBay prices are a lot better than you think

The most common criticism we receive about our Sneaker Price Guide is that some prices are too low.  While it’s possible that we’ve made a mistake (and so we double check every sneaker that someone questions), more often than not the problem is not with our price, but with the sneakerheads’ perception of true value.  We agree that many prices in our Guide are lower than we expected them to be.  In fact, when we first started doing this work – way back in the summer of 2012 – we feared there was a problem with our data or methodology.  So we reviewed thousands of sneakers by hand until we were convinced that many sneakers do, indeed, sell for much less than the “expected price”.

So we decided to prove it, mathematically.

Hypothesis:  Sneakerheads think eBay is more expensive than it actually is

ConclusionTrue. Perception is $46 more than reality for Jordans on eBay

The following is a complete explanation of the work done to reach that conclusion, broken down by our four key findings.

Data Criteria:

  • We analyzed all Jordan Retros, 3 to 14
  • We collected all eBay auctions (sold and unsold) for all DS mens Jordans over a 14 month period ending in May of 2013
  • Each sneaker had to have at least 20 sales and 20 non-sales, or was excluded
  • We identified six “time bins” related to the length of an auction:  0-1 days, 2-4 days, 5-7, 8-10, 11-20, and 20+
  • Each sneaker had to have at least one sale and one non-sale in each of the 6 time bins, or was excluded
  • 143 different Jordans qualified, resulting in 219,353 eBay auctions included in the analysis (119,976 sold; 99,377 unsold)

FINDNG ONE:  THE BEST DEALS ON EBAY DISAPPEAR INSTANTLY

The first step was to study the price of auctions over time. To do so, we calculated the aggregate average price for the Jordan auctions (both sold and unsold) at six different points over time (“time bins”), correlated with the time it takes an auction to end.  We also calculated the average price – the single “true market value” for all Jordans.

Average Prices of Jordan Auctions Over Time:

  • 0 to 1 days:  $278 (sold)     $298 (unsold)
  • 2-4:                 $284                  $320
  • 5-7:                 $286                   $352
  • 8-10:               $305                  $361
  • 11-20:            $334                   $379
  • 20+:                $341                  $398
  • Total              $289                    n/a

When we plot these data points on a graph we can see a clear pattern and an interesting visual hypothesis as to why eBay auctions are perceived to be so expensive.

Auctions over time chart ONE

Key Insight:  This chart shows that the lowest prices are found in eBay listings which end quickly.  The first three points on the “Sold” line – those priced at or below average – actually account for 89% of all eBay sales.  But the average sales price increases the longer it takes the eBay listing to end, whether the shoe is ultimately sold or not.  The very clear upward slope is evidence that these deals disappear almost instantly, leaving “stale digital inventory” sitting for weeks.  This has clear implications for three people:  Buyers, Sellers and You

  • Buyers:  The lesson here is to ignore any auction which has been sitting for more than 7 days – odds are it’s overpriced and a better deal will come along soon.  In fact, if you’re looking to save more than $10 you should really stop looking after the first day.
  • Sellers:  If your auction hasn’t sold after a week, it’s probably overpriced and you should consider lowering the price.
  • You:  If you’re reading this article, waiting for us to prove that eBay prices are lower than you think, the previous chart may have sparked your curiosity, but as an astute sneakerhead you are certainly thinking, “That’s all well and good Mr. Campless, but we still need a way to measure perception.”

FINDING TWO:  OVERPRICED AUCTIONS ARE SEEN 5X MORE OFTEN THAN UNDERPRICED AUCTIONS

The way to measure eBay perception is not the volume of auctions at each price point, but rather how often people see a particular priced auction.  This is best represented by the total number of days each auction spends on eBay.  We can calculate this by taking the volume of auctions at each price point, multiplied by the average number of days the auction takes to close.  The result is a percentage of total auctions seen for each price point/time combination.  This allows us to measure perception.

Total Days on eBay = Avg. # days  an auction is on eBay (time bin) * Volume of auctions at the average price

Auctions days data table v2

At an aggregate level, the 219k auctions we analyzed spent a total of 1.22 million days on eBay.  When we plot the % of total days seen on the chart, and add up the time bins greater than the average, we get very interesting results:

Auctions over time with percentages TWO v3

Key Insight:  Quickly adding the percentages in various locations relative to the actual market price provides several fascinating statistics, all of which support the hypothesis

  • Overpriced auctions are seen 69% of the time, or 5.3 times more often than underpriced auctions
  • Auctions priced at least $50 more than the average are seen 56% of the time, or 4.3 times more often than underpriced auctions
  • Only 13% of all auctions seen are underpriced, and these are underpriced by, at most, $11 less than average

Starting to connect the dots . . . we’ve proven that low priced sneakers disappear quickly . . . which leads to “stale digital inventory” . . . so people see overpriced auctions more than five times as often as underpriced ones . . . presumably creating the perception that eBay prices are higher than reality . . .

The next step, then, is to actually quantify “perception price”.  How much greater is the perception of eBay pricing than actual?

FINDING THREE:  PERCEPTION IS $46 GREATER THAN REALITY

The statistical definition of “perception” is the price at which exactly half of the auctions seen are more expensive, and exactly half are less.  We can quickly identify this point by creating a table from the data used to calculate “days on eBay”.  When we sort this table in ascending order of price we can then calculate the cumulative % of auctions seen which – at 50% – will tell us the exact dollar amount of the “Perception Price”.

Perception Price vs Reality

Key InsightThe Perception Price is $335.  The actual price is $289.  The difference of $46 (16%) represents how much more expensive people think eBay pricing is than reality.

  • $46 is almost exactly how much the eBay & Paypal fees (14%) would be on a $335 sale (perception price).  Many people already think sellers tack these fees onto the sneaker price, so this lends even more credence to the notion that sneakerhead perception is in line with the auctions they see.  (Or it’s an absolutely ridiculous scary coincidence).
  • This chart shows how we calculated the 89% of sales volume priced at or below average referenced previously in this article.  We add the number of auctions from the first three rows (110,074) and divide by total number of auctions sold (123,443).  Note, these 89% of auctions are only seen 31.7% of the time (3rd row, last column).
  • This analysis does not take into consideration many other factors which could further increase a sneakerhead’s perception of eBay pricing, such as the shock factor of seeing one or two of these ridiculous auction stories, or any latent recollection of Flight Club’s pricing (which has a 20% commission tacked on top), or simply the constant chorus of talking (sneaker)heads who lament eBay by involuntary reaction.
  • Bottom line:  We at Campless believe the eBay price perception gap is much greater than $46, but that is the extent to which we have been able to show statistically.

FINDING FOUR:  OTHER FACTORS IMPACT PRICE BUT DON’T CHANGE PERCEPTION

When considering the accuracy of our Price Guide, which is the calculation of eBay average pricing, there are a few other reasons why the data-driven price we report may be different than what you expect to see.  Here are the top four:

Fakes vs. Lots:

  • Fakes:  We take significant measures to remove fake sneakers from our data including:  statistical outlier analysis, minimum & maximums, spot checking and, in the case of significant sneakers like the Yeezy, in-depth auction-by-auction analysis.  Our four-part series on using data to spot fake Yeezys goes into great detail on the process.  That said, it’s still possible that the rogue Fake finds its way into our data.  In general, the inclusion of Fakes lowers the average price.
  • Lots:  “Lots” are auctions which contain multiple pairs.  They are most often an issue when trying to price the individual sneakers from a pack.  We use many of the exact same methods to eliminate Lots as we do for Fakes.  And like Fakes, we eliminate the large majority of Lots, but some still slip through the cracks.  When they do, the result is an increase of the average price.
  • IMPACT:  It’s possible that the high level impact of Fakes and Lots offset each other – one drives the price down, the other drives it up – leaving the price right back where it started.  That is probably the case for any sneaker which is both part of a pack and highly replicated.  But we err on the side of exclusion – if there is any question, we exclude the auction.  So the reality is that when looking at 8.8 million auctions, the very few Fakes and Lots which slip by have a negligible impact on average price or other Campless statistics.

Volume vs. Volatility:

  • Volume:  The average price of a high volume sneaker (the Gamma 11s, for example, sold 12,066 pairs since release) is, in general, more reliable than a sneaker with low volume.  Simply put, the more auctions, the lower the chance the price is negatively impacted by “bad” data.
  • Volatility:  Price volatility (defined as standard deviation divided by average price) is a measure of how much you can expect the price to deviate from the average.  Simply put, the greater the volatility, the wider the range of expected price.  The lower the volatility, the more confidence you can have in the exact average price.
  • IMPACT:  When reviewing the Campless Price Guide, which now includes sales volume and price volatility, remember that higher volume and lower volatility means you should have higher confidence in the exact price listed.  Lower volume and higher volatility are indicators to expect a larger range around the price listed.  For more information on volume, volatility and other statistics we use to extract meaning from sneaker data, see: “Sneakerhead Statistics Defined“.

Fakes, Lots, Volume and Volatility are important concepts for our work.  The manner in which we account for these factors is just one of the reasons the Campless methodology can be trusted, and why data analysis is inherently more accurate than the “expert knowledge” of any individual.  It also helps explain why your perception of eBay pricing is not as accurate as our data-driven analysis of it.  How many auctions have you looked at?  100?  1,000?  Even 10,000?  Let’s say you do nothing but search eBay all day, every day and you’ve actually seen 100,000 auctions.  Great, just 8.7 million more to go.

IN SUMMARY:

  1. The best deals disappear immediately.  89% of eBay sales are priced at or below average, but these auctions disappear within a few days, leading to “stale digital inventory” which sits for weeks, some that never sell.
  2. Expensive auctions are seen the most.  The result of “stale digital inventory” sitting for weeks is that overpriced auctions are seen 5 times more often than underpriced auctions.
  3. The perception gap is at least $46.  Quantifying the “stale digital inventory” phenomenon into dollars, perception is that Jordans are at least $46 more expensive than reality, not counting the unmeasurable impact of extreme auctions.
  4. External factors don’t impact perception.  Fakes, Lots, low volume and high volatility all play a part in the accuracy of any one auction.  However, traditional data analysis techniques combined with almost 9 million auctions ensures that these factors have nominal impact on aggregate sneaker statistics.

Finally, in the interest of boiling it all down to one simple chart, here’s one to show your friends:

eBay pricing perception chart

All that said, please keep checking our work.  If you see a price that doesn’t look right, let us know.  We’ll double check it.  I mean, we’ll probably refer you to this post again, just to make sure you’ve done your homework, but we’ll always double check ours.

Advertisements

10 comments

  1. […] If you have any concerns whatsoever about the accuracy of Campless data, please read: “eBay prices are a lot better than you think“.  We have gone to great lengths to prove – mathematically – that public […]

  2. Cool! Can’t wait to see more.

  3. Cheech · · Reply

    Wow, this seemed very painstaking. Awesome data, can’t wait to share this.

    1. That’s one way to put it. There’s no question it’s time consuming and exacting, but lots of fun, so that’s ok

  4. […] Josh Luber is the founder of Campless, a sneakerhead data company.  For a complete explanation of his eBay perception analysis, please visit the Campless post: “eBay prices are a lot better than you think.”  […]

  5. […] Josh Luber is the founder of Campless, a sneakerhead data company.  For a complete explanation of his eBay perception analysis, please visit the Campless post: “eBay prices are a lot better than you think.”  […]

  6. TJ Heller · · Reply

    Considering that this is the foundation on which you base your price guide, etc, this “analysis” is pretty unhelpful and the methodologies used are insufficient to substantiate the claims your blog post titles make.

    Besides the analysis itself – the whole “perceived price” notion doesn’t seem to make sense. In an auction market for an asset – the “perceived price” isn’t an asking price any more than it is a bid price. Transactions for the asset, despite whatever “perception” exists, occur where a buyer agrees with a seller on a price.

    Besides that, here are my issues and feedback regarding how you analyze your data:

    1. Your analyses output feeble, irrelevant information. The claim that “welp, thats what the data tells us” is handwaving nonsense. I’m sure that you do know (not being sarcastic) that you can design an analysis to output whatever flimsy garbage you like. If the goal of the site or claim of a post title is to provide a robust analysis of data (that is otherwise rock solid), do the actual work. For example, run a well designed regression analysis testing more than just a single (likely biased) hypothesis. A correlation analysis to “prove” something about asking prices and baskets of time reveals nothing except that they simply covary and nothing else.

    2. A handle on how volatile a pair has been would be very helpful – but again – the information from your analysis of volatility presented on this site is useless. Again, were this data presented better (i.e. simply providing a chart of volatility instead of just the average would be a great start) it would indeed be useful if only for the purpose of getting a handle on the trend in volatility.

    3. Half of your more recent posts are purely about the site’s accolades or totally superfluous, freakonomics-style musings on other irrelevant topics (i.e. sensitivity to inflation, what?). My suggestion would be to redirect that energy into providing analyses that have some substance.

    I say that in it’s current state – the value of the information on the site is really nothing more than a curiousity. Something that Complex’s sneaker blog and a bunch of site visitors look at and say “neat, charts and stuff about shoes”. The truth is that it looks kind of legitimate on the surface, but when you lift up that rock it’s not pretty and certainly not something that you will ultimately regard as an authoritative source for robust information on a sneaker collection.

  7. Thanks for taking the time to study our work and share your thoughts. If you have any specific questions or critiques, or any explicit alternative solutions to our work, I’m happy to discuss them and to share as much of our methodology and data as necessary in order to have a productive and respectful debate.

    On the other hand, what you you have is the thinly veiled equivalent of “I don’t like you” and or “I want to start a troll war” – both of which are to be expected from anonymous blog posters, so by that standard you have certainly succeeded.

    But lest there be no confusion as to what constitutes an actual and specific critique, allow me to respond to the one comment which most closely resembles a substantive point.

    You said: “the whole “perceived price” notion doesn’t seem to make sense. In an auction market for an asset – the “perceived price” isn’t an asking price any more than it is a bid price. Transactions for the asset, despite whatever “perception” exists, occur where a buyer agrees with a seller on a price.”

    First, as to the concept of “perceived price”, in general, the article is clear that it’s about sneakers on eBay, in the aggregate, and not about a single pair. So assuming that you or any one person doesn’t KNOW the exact price of every single pair sold on eBay (a valid assumption, I’m sure you will grant me), then all that’s left is a person’s PERCEPTION about the prices of sneakers on eBay. The article then makes an assumption about how that perception if formed – i.e., by number of auctions seen. (This is the one area you might want to focus on if you want to debate this topic).

    Your comment preface “in an auction market” is irrelevant because any eBay LISTING that ends with no sale – which is what we base our high perception values from – means the LISTING was not an AUCTION at all, but rather a “Buy-it-Now” or Auction with reserve, either of which would not have the same characteristics of a straight auction.

    Finally, by bringing the argument back to the transaction and the agreement between buyer and seller, you are essentially making my point. The actual sale is the ACTUAL price, not PERCEIVED price. We report the actual sales price for eBay sneaker auctions in our price guide. And as the introduction to the post explains, people often think (i.e., perceive) our prices to be lower than reality when, in fact, we are simply reporting the agreed upon price by the buyer and seller.

    So the question is: Why? Why do people perceive the actual price of eBay sneakers to be higher than it actually is? Well, we did a big analysis and wrote a long blog post documenting what we believe the answer to be. Please feel free to publish a competing analysis.

  8. TJ Heller · · Reply

    Believe what you will – but I had a feeling your immediate response would be an ad hominem attack based on my comment being anonymous. You categorize it as a “thinly veiled” troll post or some kind of “i dont like you” comment. I’m not sure how to respond other than to say that there is no veil and illuminates your unwillingness to accept what I believe are valid points. I must also say, while reading your response – your tone is so pretentious and condescending almost to the point of being humorous or as if it’s being written as a satire of the manner in which some kind of “shoe oracle” would write.

    I made the choice to try and be discrete with my arguments because I have seen how you have treated others who have chosen to try and argue or challenge this site or the authors of it. I’ll avoid that drama.

    Anyhow, I’ll address some points:

    * The article then makes an assumption about how that perception if formed – i.e., by number of auctions seen. (This is the one area you might want to focus on if you want to debate this topic).

    I will focus on it, thank you for pointing me in the right direction. An auction “seen” on eBay is no different than seeing an asking price in an auction market. In the eBay auction market – bidders enjoy the ability to not reveal their willingness to pay until a bid is made. Until then, all that is “seen” are asking prices, and to use that as a proxy for “perceived price” is a very confused way to look at it.

    *Your comment preface “in an auction market” is irrelevant because any eBay LISTING that ends with no sale – which is what we base our high perception values from – means the LISTING was not an AUCTION at all, but rather a “Buy-it-Now” or Auction with reserve, either of which would not have the same characteristics of a straight auction.

    A fair point – I did assume that your data took that into account one way or another.

    *Finally, by bringing the argument back to the transaction and the agreement between buyer and seller, you are essentially making my point. The actual sale is the ACTUAL price, not PERCEIVED price. We report the actual sales price for eBay sneaker auctions in our price guide. And as the introduction to the post explains, people often think (i.e., perceive) our prices to be lower than reality when, in fact, we are simply reporting the agreed upon price by the buyer and seller.

    Isn’t the above blog post not just about reporting the prices, but moreover the “spread” that exists between a “perceived price” and actual sales prices? Even going so far as to place a dollar figure on that spread?

    *So the question is: Why? Why do people perceive the actual price of eBay sneakers to be higher than it actually is? Well, we did a big analysis and wrote a long blog post documenting what we believe the answer to be. Please feel free to publish a competing analysis.

    Again, your whole analysis is predicated on a “perceived price”. It’s a pseudo-analysis really. Again, sellers in an auction market are neither price-setters, nor do they obfuscate buyers willingness to pay.

    Thank you for your permission to publish a competing analysis (effectively saying “well you go ahead and try it”. I’m not sure if that’s a serious suggestion or not – but I’d be happy to (despite the fact that it is typically the burden of an author or whomever asserts some theory to substantiate it) – if, for the sake of fairness, your captive audience were able to view my critique, and make their own minds up, etc. That’s not what I’m after – but won’t just do your job for you for the sake of doing it…

    1. *“There is no veil”

      Are you agreeing with my assessment that the purpose of your note was either “I don’t like you” or “I am trying to start a troll war”?

      *“your unwillingness to accept what I believe are valid points”

      You’ve made no valid points, only generic insults. The one arguable specific critique I answered in detail.

      *“your tone is so pretentious and condescending almost to the point of being humorous or as if it’s being written as a satire of the manner in which some kind of “shoe oracle” would write.”

      Absolutely. I’m glad my condescension wasn’t lost in written form. What exactly did you expect when you make not one substantive point but say things like: “Your analyses output feeble, irrelevant information . . . handwaving nonsense . . . information presented on this site is useless . . . totally superfluous . . . irrelevant topics . . . nothing more than a curiousity . . . it looks kind of legitimate on the surface, but when you lift up that rock it’s not pretty”

      *“I made the choice to try and be discrete with my arguments because I have seen how you have treated others who have chosen to try and argue or challenge this site or the authors of it. I’ll avoid that drama.”

      Vague and generic is not the same as discreet. You’ve already started that drama, and don’t try to pretend like your comment was anything other than a purposeful attempt to wade deep into it. I challenge you to find a public comment I’ve made to someone who wanted to legitimately discuss or question our work where I didn’t fully and completely engage in respectful debate. I only “treat others” that way when they make similarly troll-like comments such as yours.

      That said, I LOVE a good debate and welcome people to challenge our work – respectively. Obviously I should just ignore you and your comment, but with no good debate to chew on right now, I’ll gladly reply to your response to my substantive points.

      * “An auction “seen” on eBay is no different than seeing an asking price in an auction market. In the eBay auction market – bidders enjoy the ability to not reveal their willingness to pay until a bid is made. Until then, all that is “seen” are asking prices, and to use that as a proxy for “perceived price” is a very confused way to look at it.”

      We don’t use bid prices in any way. As mentioned previously, we use auctions that end with no sale. The overwhelming majority of auctions which end with no sale are buy-it-now listings, sitting there with an asking price. (The few others are where the final bid price of an auction with reserve doesn’t hit the reserve, but in that case the final end-without-a-sale price would be too low, and not relevant). The reason the buy-it-now price (aka asking price) is a proxy for perception is because if a consumer is scrolling through eBay, this is the price they see.

      * “Isn’t the above blog post not just about reporting the prices, but moreover the “spread” that exists between a “perceived price” and actual sales prices? Even going so far as to place a dollar figure on that spread?”

      Absolutely. You were the one who brought up the actual transaction, which is not the focus of the post, nor the primary driver of perception. As you can see in the post, those eBay listings which end in a sale stay on eBay less time than those which do not. And that is our posited driver of perception – length of time on eBay.

      * “Again, your whole analysis is predicated on a “perceived price”. It’s a pseudo-analysis really.”

      No idea what this means. We absolutely created the notion of perceived price, if that’s what you mean. We do not attempt to hide that, nor do we posit that our definition of perception is without reproach. I specifically suggested that you focus your critique there for that reason. But the appropriate way to question our definition of perception is not to throw bombs from the sideline that our work is “feeble” and “useless”, but to present your own hypothesis for a better way to measure perception, and explain why. I am always very interested in alternative solutions and have worked offline with formerly anonymous commentators who had put forward innovative ideas in a respectful and professional manner. If you don’t care about, or don’t believe in the idea, of “perceived price”, then why are you writing here if not to simply start drama? I have strong anecdotal evidence that there is a difference between actual and perceived price. You are welcome to not believe me, but that should make for a short argument.

      * “Again, sellers in an auction market are neither price-setters, nor do they obfuscate buyers willingness to pay.”

      Again, any use of “auction” in any argument here is irrelevant. And any mention of “buyers willingness to pay” brings this right back to actual sale, which is also not relevant to this post, as explained above, several times now.

      *”Thank you for your permission to publish a competing analysis (effectively saying “well you go ahead and try it”. I’m not sure if that’s a serious suggestion or not – but I’d be happy to”

      It is absolutely a serious suggestion. Or, at the very least, a suggestion to make a specific and analytic based argument from which to discuss the topic.

      *”(despite the fact that it is typically the burden of an author or whomever asserts some theory to substantiate it)”

      Not sure what you think the threshold for “burden of authorship” is –particularly for a personal blog – but the post itself is very descriptive and more than sufficient to substantiate the hypothesis we set out to prove (as confirmed by non-anonymous, independent third parties with impeccable and unimpeachable credentials whom review all of our high level analyses), including the documentation of all necessary assumptions and limitations (of which all analyses have, in some regard).

      *”– if, for the sake of fairness, your captive audience were able to view my critique, and make their own minds up, etc.”

      I will absolutely publish your competing analysis to answer the same hypothesis on Campless, provided you reveal who you are (and provide proof to the same), and make it specific, analytic, respectful and professional. The specific hypothesis, by the way, is: Why do people perceive eBay prices to be greater than they actually are?

      *”That’s not what I’m after – but won’t just do your job for you for the sake of doing it…”

      No idea what that means either. This isn’t my “job”; I do it for fun. And the reason you would be doing it is to prove that you have a genuine substantive critique and enjoy engaging in true intellectual debate, as opposed to using my comment board as a place to spread baseless, troll-like insults.

      The floor is completely yours. Put up or shut up.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: