Look, I know I probably write too many articles. I've had people tell me it's tough to keep up with everything I contribute podcast- and article-wise every week, and I tend to agree. I like writing and I like that I've had the opportunity to do a lot of it.

One way I've managed to write so much is to keep my subject matter segregated enough that I can avoid repeating myself. Another way? I get inspired because I read a lot.

I imagine you read articles every day and I do, too. There is a lot of free, quality content out there to supplement the superlative Insider articles on QS.

• Brainstorm Brewery
• Channel Fireball
• Empeopled for Corbin's "Market Watch" that I pretend I don't read
• Gathering Magic for Ryan's article
• MTG Headquarters where Reuben writes something like my old Alticle series but, better
• MTG Stocks for updates
• MTG Price, a site I used to contribute to and still read daily

Travis Allen at MTG Price has inspired some other pieces I have written and I think he's a snap-follow on Twitter (@wizardbumpin). Before this turns into a "Who to Follow" situation and gets treated like an installment of "Who to Follow" and is accordingly read by no one, I will set my sights on the point I was going to make, but I want to mention someone else who is doing quality work.

Jared Yost (@gildedgoblin) wrote about a topic that is in my purview as someone who writes about buylisting here on QS Insider, so naturally I was all over it. I have a lot of praise for the article, a little criticism and a strong desire to get some discussion started about how to approach data analysis like this in the future.

I may do some of this on my own in this column, so stay turned for future installments where I take my own whack at it. If you haven't read the piece already, I would urge you to do so right now before continuing with this one.

My Take

I am not going to shy away from using the word "criticism" because I think most of the criticism I have is due to the limitations of the technological platform used to scrape the prices. There were a few assumptions that I think can lead to some misinterpretation, but some of the finer points we can pick at are outside the scope of the project.

He says early on that the intention is to look at trends, and while I think some of the trends are misleading and the data set is incomplete in some serious ways, overall the data and the way they are presented are valuable. Let's just dig in and go section by section.

Data Collection Method

This is where I think a lot of the limitations associated with the project arise. The data was collected using MTG Price's Pro Trader tool. It did a good job of scraping the sources it targeted--my issue is with the sites omitted.

I think the most glaring omission is Card Kingdom, a site with which I couldn't be happier. Card Kingdom's buylist is the gold standard as far as I'm concerned. It is one of three sites featured as defaults in Trader Tools which is heavy praise already.

Add to that the anecdote that someone who has been buylisting for subsistence for years, Ryan Bushard, doesn't even use Trader Tools because he's found a faster way to list directly on the Card Kingdom site and doesn't bother listing elsewhere unless Card Kingdom isn't buying the card at all.

In a given buylisting order from me, I'd say 50% of the cards go to Card Kingdom. Leaving them out and including smaller sites like Hotsauce (a fine buylist, don't get me wrong, and I have recommended Adam as someone to sell to at events and in person) is a real limitation of the data set we're going to get.

He controls for some of the variation between buylisting and selling on eBay by calculating a real eBay value that has the 14% eBay and Paypal fees subtracted, but doesn't control for other values like listing fees and shipping fees.

When I buylist 1,000 $1 cards, I pay $8 shipping. When I sell 1,000 cards for $1 each on eBay, my shipping costs approach $300. This makes higher-value cards more attractive on eBay but not lower value ones. This is probably fine in the context of a theoretical exercise, but I feel like this exclusion from consideration leads to conclusions being drawn, specifically in the "% loss" section. I don't mean to jump ahead.

Also, excluding blanks, or cards not purchased by a given buylist, will give a misleading value when prices are averaged. Since the numerators will vary but the denominators will also not be constant, you lack a basis for comparison.

A dealer who pays $80 for Tarmogoyf and buys no other cards will have a modal buylist value of $80, which explains why it appears CCG House is incredibly generous when it comes to buying mythics but in reality, they probably only buy high-value Modern cards.

It's highly unlikely they are paying the most or even close to it, but a misleading mode value graph leads even the author of the article to extrapolate that conclusion. Their median value is exactly the same as the mode value and that tells the tale in my view.

Calculations Explained

Seems fine. I made my point about shipping fees above. This is incredibly tricky to control for, but anyone who does a lot of selling and a lot of buylisting knows these things. If eBay and TCG Player look attractive to you based on this article, look into all that it entails. It's not as simple as graphically comparing the "prices" may suggest, but all that means is that we need to take care when drawing conclusions.

Data Listed By Rarity

Another fine section. Doing both an "All" and a "Non-common" calculation was savvy. I feel like the "Average Loss %" calculation is a little loose when you compare eBay to buylisting, but I think it has a lot of value when you compare different buylists to each other.

So What Does This Tell Us?

I feel like this graph tells quite a lot. Channel Fireball and ABU are clearly outperforming the other buylists by sheer volume of cards accepted. I'd love to see Card Kingdom represented on this graph. It is surprising to see Strikezone offering on so few cards, also.

Briefly, I think it was incorrect to conclude that since CCG House has such a high mode value for mythics that this is a good place to look at unloading mythics. Quite the opposite if you ask me; the mode is nearly identical to the mean, and that indicates a small number of cards being purchased.

If you could scale this value against the number of cards on the buylist (where they are buying the third-fewest out of everyone) you might get a value that could tell us a bit more. I feel like CCG House buys a lot of $7 mythics for $3 and not many others. You really can't tell from this graph either way, so I wouldn't extrapolate anything.

The Money Shot

This graph is what it's all about. Everything else was interesting--this graph is the payoff. The entire project set out to take a look at how much money you're losing selling to various buylists and whether you're better off not losing quite so much selling via a different method.

I don't think commons should be included here, frankly. I think their sale price has too much overhead built into them. Determining the "true" cost of a common is next to impossible.

Even cards that are 4x in popular decks like Pillar of Flame sold for nearly the same amount as a card like Typhoid Rats just because websites aren't going to list commons for sale for a penny. After a certain point, the labor costs don't justify doing the work to sort them.

Commons have a certain amount of "I'm making someone find this for me and I'm going to pay extra and I'm okay with that." This project extends back to the beginning of Modern, but a majority of the cards encompassed by this project are available in booster packs for around $4 each. When you consider that, uncommons are a far more interesting measure given their lack of these limitations.

What I'd Like to See

This graph can tell us a lot about whether we should be buylisting, but the most useful thing would be determining the "cut-off point" where it makes more sense to sell on TCG Player or eBa. This seems to indicate that it's more attractive to sell on eBay or TCG Player than to just buylist, but only as an average.

There is a value where I am not going to buylist a card, which I have to calculate with my gut most times. Trader Tools tells you the spread on a card when you bring it up, and I will snap-buylist cards with a spread under 10%.

If the card is worth around $50-$100, though, at what point does it make sense to just TCG Player a card with a 35% spread?  We want to know the retail value of a card for which your TCG Player or eBay fees, shipping fees, etc., will be completely negated by the percentage above buylist you receive.

Drawing Conclusions

Okay, so I spent three fourths of an article looking up someone else's article's butt with a flashlight. Why did I bother?

I think it was important to me to understand the implications of all this data because I still think it's worth it to buylist cards 90% of the time. I want to make it clear that while the article seems to indicate there are more scenarios where you'd be better off selling on eBay or TCG Player, that paints an incomplete portrait of reality.

Again, this is not a result of carelessness or stupidity on Jared's part--in fact, this is one of the most scientifically-responsible pieces of analysis I've seen lately.

I think that if the analysis could be modified to control for higher shipping costs associated with selling cards piecemeal compared to sending off one big package to a dealer or two, it might paint a different picture. I think the graphs of percentage loss might be a lot closer. I think this was an excellent thought exercise and I really love having data to back up conclusions.

Taking off my scientist's lab coat for a second, there is one big factor that the data can't measure, but it's something intangible and therefore not super scientific.

When you buylist cards, they're gone. I've never sent to a buylist on Trader Tools and had them say "we don't need these cards anymore" so the result of me sending the cards is that they are as good as sold.

You have to wait for cards on eBay and TCG Player to sell. You have to constantly update prices to make sure yours are competitive. Sometimes you won't sell unless you're absolutely the cheapest price on there, adding to your percentage loss. Sometimes cards tank in value as they sit there not selling. You have to go to the post office all the time. You have to stuff envelopes every day.

You have to spend a ton of time listing. When I was selling on eBay hardcore, I got so fast I could list 20 items an hour. There are automated ways to update to eBay--Crystal Commerce is a popular way to update your eBay, Amazon, TCG Player and personal website inventory all at once. That's a $500 initial cost. How much of your percentage loss savings are eaten there?

I'm not saying don't sell on TCG Player--I do. I'm saying that it makes sense to sell a far smaller percentage of cards on TCG Player than you might think based on some of these graphical representations. For me, I think the fact that I send one package to each dealer twice a month or whatever, can jam hundreds of items in Trader Tools per hour, and know how much I am getting and when, makes buylisting the far more attractive option.

This got me thinking about calculating the card value, spread, etc. where it makes more sense to try and sell on TCG Player. If I can come up with a decent algorithm, we could program a useful tool. I'm sure it's been tried, so for now I'm going to keep shooting from the hip in this regard.

I'd like to thank Jared for writing such a compelling article. I'm still going to buylist primarily and I think buylisting is far more attractive than his piece would seem to indicate, but I am really looking forward to the conversation it will generate.  Let's get one started below.