It’s Time to Update the DCI Rating System

I’ve been harping on about the DCI ranking system on twitter (where I do all my best harping) for the past couple of weeks. This has largely been borne of my frustration that despite playing objectively well – with a consistent 4:1 win:loss ratio – I have barely been able to progress my DCI ranking.

There are numerous reasons why I’m finding it difficult to progress my ranking, which I’ll try to cover here. For reference, here’s a hand DCI ranking calculator.

Firstly, however, a quick look at the DCI system. The DCI is based on chess’ ELO system. While a natural fit in theory, there are some differences between Chess & the ELO, and Magic the Gathering and the DCI.

• Magic involves an element of luck that chess does not: This includes draft / sealed pools, card draw variance, deck selection and pairing, dice rolls and coin flips. A game of Magic, unlike chess, is never entirely about skill. Whereas a Chess Grandmaster will always beat the kid who just started, this is not true in Magic, where luck inevitably plays a part, no matter how small.
• The ‘K Factor’ in ELO rankings is not applied the same way in DCI rankings: In Chess, the K Factor is staggered based on the skill level of the player. In Magic, the K Factor is staggered based on the assumed skill level it would take to win the tournament. This means in chess – except in open events – players are generally restricted on playing against opponents on a similar rating to themselves. In Magic, players can come up against opponents with vastly different ratings, thereby facilitating large rating swings (both positive and negative).
• Chess maintains one ranking, DCI maintains several: In Chess, you have a single ELO number within a particular tournament group. Within the DCI, you have several rankings – eg. Limited, Constructed, Total. This results in circumstances where you can make large gains in one area (eg. Constructed) that does not similarly affect other areas (eg. Total ranking, if you were already highly ranked due to a high Limited ranking).

It should be noted that the ELO system is under constant scrutiny by the Chess community, and there have been several attempts to improve it, with few tournaments using the straight ELO system, which is currently widely regarded as flawed. Right now there is a competition to find a completely new system. The K-Value is constantly being tinkered with, with both larger (32+) and smaller (4) figures being trialed for high-level players to see what brings the most consistency to the game.

So, back to my situation.

Over the past few years, I’ve concentrated largely on my limited rating. My big break in my rating was when I managed to come second at a Sealed PTQ. This boosted my rank from 1634 to 1757 overnight. For the next two years my Limited ranking bounced between 1720 to 1800 in the Australian Magic scene, and my Constructed rating managed to dip around 1550 – largely because I insisted on playing terrible decks – and then back up to 1599.

Since then I’ve moved to Hong Kong. Hong Kong has a much smaller player base than Australia, or even New South Wales. In Australia I was rated within the top 300 players. Simply by moving location my ranking has moved into the top 100.
One thing I decided to do was ensure that, next year, I’d be qualified for Hong Kong Nationals. As the Top 75 players get auto-invited to Nationals, I felt a Total rating of 1800 would be a pretty safe cut-off point here.

The first step in my plan to achieving this was to secure at +1800 Limited rating, followed by a +1800 Constructed rating. As such I dutifully turned up at every FNM I could, and at rated Sealed events, in the hope of achieving a +1800 Limited rating.
And what a drag it’s been.

I started on a ranking of 1780 off the plane from Sydney. Due to some settling in issues (basically, learning to play Limited Magic with cards in Traditional Chinese!), I lost the first few games. However, I forced myself to memorise the pictures of every card in M11 and turned it around. My total matches in M11 in Hong Kong are 21-7-4, essentially running a 3:1 win ratio, really 4:1 if you excuse the first dismal draft when learning how to play in a foreign language.

So, across those months of playing and, objectively, having a pretty good track record, how many points have I managed to pick up? Most of these were 8k events, so you’d assume – at an average gain/loss of 4 points per match, (21*4) – (7*4) – (4*1), or about 52 points.

Right now my Limited DCI rating is 1799, a grand total of a 19 point gain, or less than 1 point per win.

This looks even worse when you consider that one of those 4:1 win ratios was during a 24k event, which I won – and gained a mere 3 DCI points in total.

Here are some reasons why this has occurred.

• I’m now playing players well below my ranking: With the limited number of players in Hong Kong, there are more players with low DCI ratings. This means that game that I lose due to variance, rather than play errors, have a much greater impact on my ranking. For instance, at the aforementioned GPT, I lost a single Match. In Game 1 I had to mull to 4. In Game 2 I had to mull to 5. Meanwhile, my opponent managed to drop his entire pool on the table – Grave Titan, Hoarding Dragon, and other large beats. Unfortunately my opponent’s rating was 1528. This meant I lost 20 rating points, which I barely made up pushing through match wins against better players. The 1528 rated opponent didn’t make Top 4, but that didn’t stop my rating dropping through the floor due to that single loss.
• Those with high ratings protect them and don’t participate: One thing I’ve noticed is that those with the highest ranking tend to avoid playing in FNM in order to protect their ranking. They’ve clearly already recognized the issue I’m up against – the chance of gains in this broken system are not worth the risks of losses. This means that those trying to improve their rating – such as myself – are stuck in a cycle of beating and occasionally losing to low rated players, thereby endlessly cycling around the same rating.
• There are fewer high-level events and more FNMs: This means access to games with a larger point swing are limited. With variance in consideration at the smaller FNMs – often three rounds – it means that unless I can guarantee a 3-0 win each round, a logical impossibility, I will most often lose rating for each FNM or draft I participate in.
• As my rating improves, the DCI makes it harder and harder to progress: Partially due to the small player pool, partially due to the low ranking of those who participate in FNM, the stranglehold the DCI formula places on my progression gets worse and worse the higher I climb, thereby exacerbating an already difficult situation.

Clearly the DCI rating system is broken. But it’s not just broken for me; it’s broken for everyone.

It’s broken for the top level players who ID and lose rating as a result.

It’s broken for the small playgroups who bounce around the same ratings, unable to advance with any significance.

And it’s broken for Wizards of the Coast, as it discourages involvement at FNMs and 8k Drafts by higher-level players as it offers a substantial threat to their rating, thus hurting their income.

Personally, I would like to see WotC abandon the Elo system, much like Blizzard did when it revamped its Arena system for World of Warcraft. I would like to see a system that holds to the following fundamental tenants:

• You will always gain points at an event where you win more games than you lose
• You can never gain points for an event where you lose more games than you win
• You can neither gain nor lose points when you draw (intentionally or otherwise)

Now, I’m no mathematician, so I’m probably not the best person to determine the formula that redefines the DCI. But there are a lot of smart people at WotC who are capable of finding a better way of ranking players, and I hope that they chose to divert some resources to that effort.