Anthony Capece – Modeling Uncertainty in MTGO (How Much Do You Trust Worth?)

There was a pretty significant change to MTGO last week. Wizards shut down all daily, premier, PTQ, and MOCS events indefinitely. Where there is change there is uncertainty, and where there is uncertainty there is opportunity. We’re all on the same page here – we want to take advantage if card prices drop. Unlike most Magic specs, though, there is a non-zero chance of this one having really bad outcome, an outcome where we buy in and prices crater, costing us a bunch of money. How do we approach it?

Here’s what I’m not going to do in this article: I’m not going to try to convince you I know what will happen. Nobody knows exactly how things will unfold, probably not even Worth and the Wizards MTGO team, and least of all me. If you want opinions on what will happen, Twitter is lit up like a Christmas tree.

An uncertain situation like this calls for examining probabilities instead of trying to predict a specific outcome. When you flip a coin, you have no idea what side will land up, and no one would ever claim to be able to predict coin flips. But you know that the probability is exactly 50% of either side coming up and that gives you enough information with which to work. Poker is the same – good poker players don’t try to guess what the flop will be, they think about all the possible outcomes and the likelihood of each and bet accordingly.

You probably already have an opinion about what will happen with MTGO. But do you have a plan? Or are you just going to stare at all the red on until “it feels right” and pull the trigger? This isn’t like watching a card bounce up and down in its range – price history goes out the window if the fixes don’t go well. In this case, you may see a lot of people take the “dump my MTGO collection for whatever I can get and buy, you know, real cards” approach. I have to tell you, it makes a lot of sense. Then again, maybe it will just blow over and MTGO events will be up again soon. The point is, we really don’t know.


In this article, I’m going to show you how to build a very simple expected value (EV) model that takes what you think will happen and turns it into a plan for buying or not buying into MTGO (actually, I’ve already built the model for you). You are going to input the parameters, they are going to be your guesses based on the information that you have. This works out best because if you get it right you will be super proud of yourself, and if you miss there is really no one to blame but yourself. It’s your money, after all.

Disclaimer: if you are familiar with expected value and good at Excel models, you probably already have something like this. And if you are a seasoned trader, you might not need something like this. Your intuition is probably good enough. If you are thinking, “I really don’t have much confidence in the MTGO product, but at some point these cards have to be a good buy. I just don’t know when that is,” well then, bingo.

This model is going to tell you what your buy price is for any particular card based on the possible outcomes of the MTGO situation and the probability (that you assign) of each scenario. I like this model best for non-redeemable cards. Redeemable cards are going to be linked to the paper price, at least loosely. No matter how bad things go on MTGO, you can always cash a set of digital Theros in for paper cards and go play, and that will provide a floor for the digital versions. MTGO will continue to serve this function regardless of whether or not the events are stable. Our approach will work much better for Modern sets (other than Innistrad block, which is still redeemable). Their only value is their playability on MTGO. It’s not to say you couldn’t adjust the numbers to better reflect redeemable sets, just that I didn’t have that in mind when I made it.

Potential Outcomes

We’ll start by listing the potential outcomes of the current MTGO situation and how they would affect card prices. If any of these scenarios seem off to you, don’t worry. You can add different outcomes and change the numbers as you see fit.

My take on it boils down to three possibilities:

1) Wizards gets MTGO fixed and back online quickly. In this scenario, Worth’s update at the end of the year is basically “it’s fixed and coming back very soon.” All the players return and the prices bounce right back to where they were prior to the shut-down announcement. You sell anything you bought on a dip at full retail and enjoy the profits.

2) Wizards gets it fixed, but it takes longer than anyone really wanted. In this scenario, Worth’s update is something like “we’ve figured it out, but we need more resources and it’s going to take a few more months.” MTGO actually loses some players as a lot of people decide they aren’t going to wait it out, dump their cards, and officially move on to Hearthstone or League of Legends or just paper Magic. Prices recover to about 75% of their original price when the events come back. If you bought low enough, you make some money. If you jumped the gun, you lose some money.

3) They don’t figure it out in any reasonable amount of time. Maybe they hire new developers and they all quit after a month, putting them back to square one. The end of the year announcement is “we’re still working, please stick with us, but we have no idea when this is going to be fixed.” I’m sure it would be accompanied by some underwhelming consolation prize like cheap cube drafts until events are back (actually, I wouldn’t mind that). This is the doomsday scenario where everyone but the most die-hard players lose confidence and bail. Prices plummet to 25% of their original price because no one is going to leave money tied up in a system with no real timeline for return. If you bought in at anything but rock bottom, you probably lost a lot of money.

Maybe you think there should be another scenario between two and three, a really-bad-but-not-doomsday scenario. If you want to go deep, add even more scenarios. Or maybe you think prices settle at 50% in scenario two instead of 75%. If you want to simplify, you can take it down to two scenarios: “MTGO lives, prices recover to 100%,” and “MTGO dies, prices plummet to 10% original value.” Make your notes, you can always change the model later.

By they way, if I was trying to do a version for Standard (redeemable) prices, I would want to make the floors a lot higher to reflect redemption (so even in the worst case scenario, prices only drop to 80% of their original price or something along those lines). Just don’t mix redeemable and non-redeemable, I think they are going to behave very differently if things get bad.

Now we estimate how likely each scenario is. This is where it becomes your model. If you are an optimist and you think there is a really good shot they fix it by the end of the year, give scenario one an 80% likelihood. If you are a developer and understand just how hard this is going to be but still think it’s doable, maybe you weight the second scenario higher. If you are risk averse and really can’t afford to lose money on MTGO, be conservative and rate scenario three at 50%. Your call. Just make sure they add up to 100%.

Personally, I’m guessing there is a 50% shot at the first scenario, and about a 45% at the second. I’m not sure how long it will take, but I think they’ll fix it. It won’t surprise me if some players never come back, but I think they’ll get it done before players flee in large numbers. So I think the doomsday scenario is only 5%.

Now open the spreadsheet (you have to enable macros). The card I’m looking at is Misty Rainforest (23 tix prior to the announcement) and it has all the assumptions I made above. Obviously, you can change the card, just find the pre-shutdown price on Hit the “Run It” button and presto, it spits out 19.55 tix at the bottom.

19.55 tix is the price at which my expected value is zero. If my assumptions are right, on average, I will make money if I buy below 19.55 tix and lose money if I buy above it. So I know that once Misty Rainforest falls below that price, I’m in my buy zone.

A word about expected value if you are unfamiliar: EV tells you what you can expect to gain or lose on average if you repeat the scenario many times. Again, think poker. You get dealt pocket aces and it means that you are a favorite to win that hand. You usually make money on that hand, but not always. There is only one outcome to each hand, but if you average all the hands where you are dealt pocket aces, you usually win a good bit of money (though you do occasionally lose when someone hits a straight or something). You have a high expected value with pocket aces, but that does not mean you are guaranteed to win money. This is a crucial point.

In poker you get to play hand after hand, giving you a large sample size – we only play this game once. Because of that, I would strongly advise against using this model as a hard buy/sell indicator. It is going to give you one piece of information to consider when making the decision, it is not going to make the decision for you.

You’ll find a lot of value in just plugging tons of different assumptions in and running the model over and over. It will help you put everything in context. What’s the worst that could happen (take scenario three up to 95% likelihood)? Misty would have to get all the way down to 6.6 tix to break even on your expected value. What if Magic players really are addicts and don’t sell off their cards no matter what (set cards to drop to only 95% of original value even in the worst scenario)? The model says 22.4 tix, meaning buy on any small dip. Even as I write this, though, cards are dropping more than that.

After trying lots of iterations, you will get a feel for what is reasonable. You either have faith in Worth and Wizards, or you are willing to bet against them. If your assumptions say that you feel good about MTGO getting fixed, the model will tell you to buy on small dips (break even price will be high). If you are pessimistic about it, it will tell you to wait until cards get very cheap to hedge against the worst case. Remember, it’s giving you the EV = 0 price. You actually want to buy in cheaper than that.

All the spreadsheet does is calculate the profit in each scenario you input, weight that profit based on the likelihood of the scenario, and then solve for a card price. I say “all it does,” but I personally find these things incredibly valuable. It’s a great tool for quantifying the mess of information floating around in my head. It stops me from running too far in one direction and from making buys that aren’t consistent with what I think will happen in the future.

Taking Action (Or Not)

After you’ve finished with the model you just wait and watch. When new information comes out, update your model. If Worth gets fired in mid-December and flames Wizards on Twitter, you might want to bump up the likelihood of that third scenario and run it again. If you hear that they are bringing in some MTGO grinders to help test, there’s a pretty good chance they are making progress. Adjust accordingly. If we don’t hear anything, well, your guess is still as good as mine. By the time you read this, prices will have done something. They are either dropping fast or holding firm or something in between, but they are doing something, and it’s going to give you valuable information. It’s telling you what other people think the outcome is going to be.

Wait, watch, update your assumptions, and then when something gets pretty out of whack with your model, buy. If your model tells you 20 tix is break even on Misty based on your assumptions and the current price is 15 tix and dropping, consider buying. The model is telling you that the market is less confident in a fix than you are. If you are right and they are wrong, you will profit.

Leave a margin of safety (don’t buy right below your break even price). Understand that cards might not get as cheap as you need them to in order to feel good about investing in a bunch of digital objects on a buggy platform. In that case the model is telling you to sit tight. Also understand that if prices don’t drop at all, there just won’t be any opportunity. Spending ten minutes running every scenario you can think of is really going to help you picture these outcomes. If the opportunity does come, you should be able to recognize it.

Thanks for reading.

Download the model here–MTGO Breakeven

(Reminder – you have to enable macros for the model to work. The excel sheet probably isn’t that hard to break, either. Just keep a clean copy saved and re-open it if the macro stops working for any reason.)

About the Author
@acmtg   -    Articles Anthony is your typical started-during-Revised-then-quit-then-came-back-years-later Magic player. He enjoys the financial aspect of the game the most, mainly because it lets him use his analytical side but also because it makes up for the money he hemorrhages drafting on MTGO.

7 comments on Anthony Capece – Modeling Uncertainty in MTGO (How Much Do You Trust Worth?)