This week's topic is about taking a journey into the bot universe. We, as a team, decided to make a move towards a bot system, and I will cover the many implications behind such a decision. I think one must carefully assess the pros and cons of this time-consuming task, which can also end up being quite expensive as you will see. As of now, we are six weeks into the process of running two bots which share credits. In the end, I will tell you whether it is worth it or not, based on my personal (although limited) experience.

Think It Through, Then Think About It Some More

If you are serious about running a bot, you should consider the few options that are available to you. If you think you can program it yourself, well, that's great for you, go ahead and use AutoIt. I'm into psychology. I am about to receive my Ph.D., and I never failed a class. The only time I almost failed was when I took an optional course on Introduction to Microsoft Excel. This is just to tell you how super bad I am with technology.

So to me, options were much more limited. We first tried to hire three computer engineers, who knew nothing about Magic. They all gave up after a few days. Given this failure, I also tried contacting a few major bot chains I enjoyed trading with, to see if they would rent their technology, but I received negative answers. I recently approached Goatbots as well, but same story. I was left with just two possibilities:

1. NS5 bots (mtgons5.com)
2. Mtgolibrary bots

I ended up choosing Mtgolibrary. One reason was that NS5's website has been down for several weeks, and when it came back up again and I tried dowloading their software, my firewall went nuts detecting malware. Did I tell you I am super bad with computer stuff? That was enough to freak me out, and only Mtgolibrary was left as a final option, even though I find these bots slow like hell compared to engines like Goatbots. As I will tell you, there are ways to make the trades faster, but you have to sacrifice some automation to achieve that.

The Costs of Setting Up a Bot

Mtgolibrary bots are free to use as long as you don't buy or sell anything. Whenever you buy or sell, the owner (a guy named Albert) collects 2.5% of the total amount of the transaction. In order to get away with this fee, you must buy a license (various periods are available, from 1 month all the way up to lifetime). To run more than one bot like we wanted to, you must buy a separate license for each bot.

Now, I had to figure out how to run this software from my laptop, as I have no other computer at home. I'll make this simple, because that's what I would have liked other people to do for me. It's impossible without managing Virtual Machines, because of the limited screen resolution a laptop can handle (the bot requires a resolution of 1200x1024 or higher). If you own a PC, you will probably be OK running just one bot. But then again, if you want to run more than one, you still have to get familiar with virtual machines.

I used VMWare Workstation's 30-days trial version, and honestly it was a pleasant experience. For a noob like me, having a virtual machine running Windows 7, a virtual mouse and a virual keyboard, and software that handles everything for me was awesome.

They have a super nice feature, called "clone." You click a button with that label, and after a ten minute wait you get an exact replica of your first virtual computer. So you set up everything you need, make all the necessary updates for MTGO (including installation of DirectX and .Net framework, whatever those are). Then, you clone the thing and call it a different name (Windows 1, Windows 2, Banana, you pick), and you do this as many times as you want... or can you? Well, it all depends on your memory. I own an Intel Core i5 2.5 GHz, 6GB of RAM. Given these features, I can only support two bots. A third bot makes everything laggy and buggy. There are ways to make the virtual engines less demanding, but I haven't mastered them as of now.

At the end of the trial version, I tried the free edition of VMWare Player, but it was unstable, inconvenient, and the bots didn't run properly. I ended up buying the Workstation licence at a 25% discount, paying 200$.

So far, we are down $200.

Running only two bots is limiting in many ways. First, you only get two ad spaces, which reduces your visibility on the market by a lot. Second, people get more confident when they see a large bot chain with accounts sharing credits. Since I am unable to enhance my laptop's performances, running more bots implies buying a new PC with loads and loads of memory. We are talking about investing an additional $750 here.

Feeding the Bots with Tickets

Then we had to feed our bots with tickets. If not, people open a trade, and they turn away the minute they see we don't have anything available. We decided to allocate 500 tickets per bot, thinking that would last a while. Two days later, I was disappointed. We had bought a bunch of cards we wanted, at prices we had set, in quantities we were comfortable with. The only problem is that all our tickets were already gone.

So I put up an additional 300 tickets, waiting for the right moment to sell our RTR boosters and unlock a nice bunch of tickets (900 tix). Even this additional input didn't last long. I lowered my buying prices again, thinking that whatever we'd end up buying, I'd be super happy with. As I am writing this, we're already out of tickets again (Hey Jeff, I hate having to tell you this via a published article, sorry buddy).

So we are down for an additional $2200

What's Wrong with Our Buylist?

Really, nothing is wrong with the prices we have set. There are a few Standard staples we wanted to acquire at a good price, so we considered the 2.5% and we offered a competitive buying price. But we also wanted to post a buy ad for cards that not many bots post for, because we think those cards are either Modern playable or underpriced Legacy staples. Given that few bots are posting for them, we were offering decent buy prices, but these prices still represented a significant discount compared to mtgotraders' selling prices. In fact, those buying prices we were offering were so low that we would never dare solicit human sellers with such bad offers.

So we used our limited ad space to represent some popular Standard targets we needed (RTR removals and shocklands for example), and devoted the rest of the ad space for cards such as Grove of the Burnwillows, Wasteland, Linvala, Keeper of Silence, Jace, the Mind Sculptor, Birthing Pod or even Goblin Guide.

Here is a sample of our advertised cards with their buylist prices:

Believe it or not, in all of these examples we were successful and acquired a bunch of copies. We were then free to either manually sell the cards off for a quick flip, or hold them for the longer term.

Should We Also Start Selling?

Given the rate at which our tickets were disappearing, we had to decide if we also wanted to sell some cards through our bots. Quickly, a ton of new questions arose. What cards can we sell, if not our long-term targets? If we must sell unowned cards, do we have to first buy more cards? Did we want to buy four copies of every card ever printed, like Aboshan does, in order to be well stocked and encourage players to come back regularly? Did we want to use mtgolibrary's pricelist? Did we only want to buy rares and mythics, like Supernovabot does? Do we want our bots to transfer their collections? Do we want one seller bot and one buying bot?

As of now, some answers are still unclear to me, and we have not fully tested all the possible avenues.

Ready or not, I wanted to give it a try, so I decided to extend my wishlist to buy even more cards, in order to include some bulk rares and other newly available cards from DGM. My rationale was that some DGM rares' value would be less well known to some players, and they would agree to sell them at bulk prices to round up their trades with our bots. For example, I thought about buying Notion Thief and Plasm Capture at 0.10 each, and matched my sell price with Cardbot (roughly 60 cents each). This worked out just fine, and I ended up buying and selling these a few times over the week, with a pretty fair margin.

Some oddities happened when setting up my sell prices. I felt lazy when I got to the Modern cards, so I decided to set prices at a level I would be happy to sell at when the next Modern season arrived. I told myself these selling prices would also act as reminders, like my spreadsheets. For example, I set up my sell prices for Birthing Pod and Razorverge Thicket at 5.00 and 2.00 respectively, knowing I had bought my copies at 2.49 and 0.25 each. I was astonished this weekend when I found out I was sold out on both cards. I had to check back on mtgotraders, and the cards were available there for 4.15 and 0.55 respectively!

Having a bot ready to sell is useful for the future too. I will have to unload a few targets, for which we own large quantities. Doing so manually will be painful. Owning a bot means that I can mark my cards tradeable only when I feel the time is right, and at the price I want. The bot will then slowly unload the cards for me while I'm away.

A Few Pieces of Advice

Running a bot can be really time consuming. (Almost as much as having to rewrite an article because of some misunderstanding on my part of WordPress, but that's a different story.) We made a strong effort to follow our own rules and avoid the mistakes mentioned in our second article. I feel those pieces of advice apply to this part of speculation too, so don't hesitate to refer back to that article.

Stay Focused -- If you are tight on tickets, as we ended up being, make sure you reduce your wishlist to the most important targets you want to acquire. It's easy to be enthusiastic and write a large list of cards you would like to buy at a discount. But if you end up running short on tix all the time, you will find this experience quite expensive. You can also decrease your buy prices, so whatever you buy, you can turn it into instant profits.

Careful With the Quantities -- I have no problems with buying 100 copies of each shockland if the price is close to 2 tix, as I believe it's a safe position. During DGM's prerelease, I scanned the set's mythics and rares to find bulk cards I could buy for cheap. But I neglected to adjust my prices for some of these cards over time. As prices decreased rapidly, some player randomly found out that my bots were buying Savageborn Hydra at 2.00, while mtgotraders was selling them for 1.55. My quantities were set up to 12 per bot, and rest assured the guy made sure I would own all 24 in no time. Reducing your buying quantities to four or eight per bot sounds much safer.

Adjust your Prices Regurlarly -- It may be a little obvious. Since I don't use mtgolibrary's pricelist, I have to check in routinely on mtgotraders and make sure I am not overpaying. With MM's spoilers, I also had to decrease the buy price of many cards. After each completed trade, I also go back to my wishlist and decide if I want to change my price or my quantities, or leave them as is.

Make Sure You Have Enough Money -- Each bot should represent an investment of anywhere between 500 to 1000 tix. If you have that floating around, you should be fine. For any less, I have no idea how you could manage a bot, other than perhaps a bulk bot.

Committing to the Work Required -- This feels quite important. We are six weeks in and I'm just starting to feel more comfortable with the fact I'm running a bot. Fees are adding up quickly. It takes a lot of time to adjust everything every day. Make sure you like handling such things daily before acquiring a free lifetime licence. You wouldn't want to invest this kind of money and get fed up after six months.

Make Sure You Are Visible -- Space in ads is so limited and you have to be clever. Making yourself visible with a few keywords is helpful. Having a few popular cards on your buy list, with a decent premium, will attract many players who will then dump other cards you want, even though you didn't advertise for them on the Classifieds. Selecting which cards should appear on the Classifieds requires some thinking; Jeff and I had to sit and discuss at length why we wanted some specific cards to show up in our ads. The more bots you run, the more likely you are to appear in people's search. Getting trades because you successfully dragged people with clever ads feels quite rewarding.

Was It Worth the Effort?

To conclude, as I get more used to the settings of my bots, I spend less time rearranging my features and my prices. Fewer flaws can be found in my wishlist. We have good visibility because we are strategic with our postings. Some harsh decisions lay ahead of us, such as buying a new PC, or getting licenses for our bots to save on the fees.

Overall, even if it takes some efforts and some time to set up the virtual machines, I think there are good profits to be made. We opted for a strategy where we focus on specific cards, and we intend to gradually unload our cards using our bots when the time comes. Our profits so far fully cover the fees we were charged. We have also covered the license fees for VMWare Workstation.

Starting the bots meant that we had to inject additional capital, but we had the leverage to do so. I will still be able to pay my house and my tuition fees and my income taxes. So stress levels are under control.

Eventually, I could write a detailed guide to help people set up their own bot, in a step-by-step description. Mtgolibrary has a blog for that, but I find their guides confusing at times. Even though I do not feel like an expert in the field of botting, I will gladly help you figure out what your strategy should be, given your bankroll and the time on your hands. And I could certainly ask you a few hundred questions you might never have considered before giving it a try.

The Psycho,
Sébastien Morin