Back when I added blacksmithing to my spreadsheet I ended up discussing how to implement inscription in my spreadsheet. I’m not quite there yet, but one of the things we did discuss in the reddit thread was the possibility of simulating card crafting to figure out how many cards we would expect to waste.
Recently Darkmoon decks sales are way up. With the possibility of upgrading them to item level 900 some of them are very desirable, particularly the tank version, which sells super well.
When you are crafting darkmoon cards you won’t usually get full value until you get a full deck. Due to the 1/32 chance of getting any one card there is a large bit of RNG involved. What we want to know is how many wasted cards would we expect at various crafting samples and what is in essence the actual crafting cost. This will help us make a better TSM operation and a better minimum price.
I have sold 88 Darkmoon decks since I got the rank 3 Darkmoon Card recipe and started stocking them. In addition to this I have 51 decks in my inventory currently. Which means I have crafted at least 1112 cards. Which is a lot. But I also have a lot of cards left over. This means that my actual crafting cost is quite a bit higher than 8 times the crafting cost of a single card. My wasted cards amount to 267 cards.
This means my effective crafting cost so far is 24% higher than the cost of 8 cards. The question is: Is this unique? And what sort of effective crafting cost can we expect?
Enter Monte Carlo simulation
Monte Carlo simulation is a powerful modeling tool. Essentially you set up a model and simulate it thousands of times. This will give a good idea of what some number or size converges to. Luckily WoW RNG is very suitable for simulation. We assume that each card has 1 in 32 chance of appearing (equal chance). We can then use a random variable to generate a long list of cards. I decided to do samples of 250, 500, 1000, 2000 and 5000 cards. For each sample we find the number of total decks and then the number of cards left over. We then do 5000 simulations of each sample size.
After finding the number of wasted cards and decks we can easily calculate the amount of cards you need to craft per deck and the effective crafting cost as a percentage of the normal crafting cost.
Let’s look at the results
First let’s take a look at the expected number of cards per deck. For each sample size I present the best sample, the 10th percentile (the 10% best results used this number of cards or less), the average (or expected value), the 90th percentile (The 10% worst results are above this in cards) and the absolute worst result.
So, we can see immediately that we expect quite a lot of waste. Even if you craft 5000 cards you would need on average 9 cards per deck. That’s a total of about 550 wasted cards. In effect you would need to craft an ungodly amount of cards to reach the 8 cards per deck theoretic crafting cost.
At 250 cards we can see that the crafting cost is horrible. The worst sample had 36 cards per deck. On average you would only be able to turn half your cards into decks. At 1000 cards, which is where I am rouhly the expected cost is 10,6 cards per deck. My minimum price uses 9 times the crafting cost, so I will have to increase it (you can find my updated TSM group at the bottom).
Now let’s look at it in terms of percentages, below is the same table, but it has been divided by 8 to find the percentage increase in crafting cost. 8 cards per deck equals 100%. As we covered for me 24% of my cards have been wasted.
At 1000 cards the expected value is 33% cards wasted so we can see that I have actually been getting quite lucky. I am actually in about the 10% luckiest crafters of Darkmoon Decks.
Expected card waste is high, but profits are still higher
I was very surprised at how high the expected card waste was. At 250 cards you can not really expect to make a profit unless decks are selling for 16 times the crafting cost of a single card. And even then you would need it to be 20 times to feel reasonably safe that you won’t lose money.
It seems to me that the 1000 card sample is a good base for setting minimum prices. I think sizes of 1000 cards is attainable for a lot of crafters. Smaller sample sizes can be very risky. We can easily see that the worst cases quickly get less bad as we increase the sample. Going from 250 to 500 cards halves the crafting cost per deck in the absolute worst case.
At current prices on my realm you will be profitable in almost all cases if you craft at least 500 cards. This might not be the same on yours, but crafting less than that is a high risk proposition, so be aware if you are planning to test this market. That being said the market is very hot at the moment and the profit can be extremely good (I have 300% profit margin on the immortality decks at the moment).
Get in and go crazy!
My sales from the last week are a testament to how good the market is. At 12 cards per deck my crafting cost is about 25 000 gold. Still very healthy margins as you can see below.
Modifying the TSM setup
Based on this I have increased the minimum price of my TSM settings to 12 times the crafting cost of a single card. This should guarantee a profit at sample sizes that are 1000 cards or above.
My updated group can be found here. Time to mass buy herbs and start canceling and reposting my decks more aggressively.