This past Tuesday, the UK Community held the draft for its first ever W.A.R. event. As a non-regent but a huge stats-nerd, I’ve spent the past couple of months tracking the value of the factions in a bid to ascertain their value, so now I present to you my working, and we can analyse: which regents did the best job in getting a faction?
To figure this out, I used the API data Jousting Pavilion provides (more info here – and don’t worry, it looks incredibly complicated I know but I’ve literally never used API data before and didn’t find it too complicated!), and then wrote a short Excel macro to convert the data you get from there into a simple table, looking at every game that’s been uploaded to Jousting Pavilion, which faction won and which faction lost, devoid of all other context:
This goes on for over 3,000 rows
From there, we can get an idea not just of how much a faction wins its games, but much a faction wins and loses games vs another faction in particular.
This data has a few important flaws in it, which it’s worth evaluating now before we get any further:
- It ignores pack releases. I have exclusively taken data from since the last restricted list, as I feel that restricted lists impact the data far more than individual products do. However, the earliest data points are a CP2 meta. Since then we have had: the Fury of the Storm deluxe box; the Pit of Snakes chapter pack; and the Beneath the Red Keep chapter pack. These were released at varying points throughout the collection of the data, so while most (though not all) include the Bara box, very few include Beneath the Red Keep. Additionally, the actual event will have the Blackwater chapter pack as well, which this data doesn’t factor in at all. The meta has evolved, and this may short-change some of the factions that have recent upswings, but the hope is that it’s not enough to fundamentally de-value the data. Ultimately I valued having more datapoints above having only the current meta.
- It looks at every single game involving a faction, rather than ‘best of the best’, i.e. the only ones really relevant for the W.A.R. format. This means that, for example, Night’s Watch’s best 6 decks may well have a much higher win ratio than the faction as a whole. Hopefully this is equally distorting for all 8 factions, although I suspect NW in particular get the short end of the stick here. On the flipside, any faction that has a stronger win-record but that relies on only one or two decks to produce it will be overestimated.
- It doesn’t take into account the format. Decks in a standard tournament will make more efforts to be well-rounded, because you don’t know what you’ll face. Decks in the W.A.R. format will face each faction exactly once (plus extra cut games), and may choose to, for example, not bother teching against Greyjoy so much, just accepting that loss and trying to go 6-1.
- The data comes from all around the world. While the W.A.R. tourney does have a reasonably strong amount of international representation, it is still predominantly British. The UK meta may not match the global meta in terms of which matchups are weak and strong. Broad trends should of course still apply though.
- There are over 3000 datapoints, which should be enough to make the data quite reliable, but certain matchups are more reliably recorded than others – for instance, we have 262 instances of Stark Vs Greyjoy, but only 36 of Lanni Vs Night’s Watch. More results could change those win ratios pretty sharpish.
Despite all this, I think there’s a lot of value in this data, so let’s by golly gosh go and have a look at it!
How this worked is I made a table of all the wins and losses each faction had against each other, and made them into a handy table, like so:
Once these values are known, you can figure out a win percentage for each matchup.
Once you have the win percentage, and you know each faction is going to play against each other faction 12 times, you can calculate the number of games you think each faction will win.
This gives us the factions being projected the following numbers of wins:
1st) Greyjoy – 52
2nd) Stark – 48
3rd) Night’s Watch, The – 42
4th=) Baratheon – 40
4th=) Targaryen – 40
6th=) Martell – 39
6th=) Tyrell – 39
8th) Lannister – 35
Now, while some of those results might surprise you, it’s fairly standard – Greyjoy are top followed by Stark, with the rest of the pack bunched up and Lannister slightly trailing, big whoop. However, what’s interesting is that we can then use that information to determine not just what order of priority the factions should be drafted in but also how much should be spent on them.
How can we figure this out? Well, while from an emotional, “functioning like a human being” perspective the regent is buying a team of people, what they’re actually doing from a pure problem-solving logical standpoint is buying wins, and trying to buy more wins than each other regent. And wins have a value.
In the tournament, we have 7 rounds, with 4 pairs of teams having 12 games against each other in each round. That means there are 336 games in the swiss, plus a further 15 in the cut for a total of 351. Now, not all of those wins are up for grabs, because the regents don’t have to buy themselves. To that end, let’s assume that, on average, each regent wins 4.5 of their 7 games – some might win more, some might win fewer. I’m going to go out on a limb and say none will exactly that number, but that’s what averages are for innit. That means approximately 36 of the wins aren’t available to purchase in the draft, leaving 315 that are. Each of the 8 regents has 120 gold to spend on their team, meaning that between them they have 960 gold to buy 315 wins. That means each win is worth:
…I’ll use the exact figure when calculating, but let’s call it “a little over 3 gold per win”.
Now, as the cheapest faction, Lannister should go for 1 gold. A regent might specifically want it and spend more to get it (spoiler: that’s what happened), but by default, we can price the number of wins the weakest faction gets at 1 gold. That means that to calculate the raw value of a faction, we’ve got a rough formula of 1 + 3.04… * (number of wins more than Lannister they’re projected to get). This gives us the following table:
Now, there are two main things it’s important to remember with this data. Firstly, these projected wins don’t include the 15 cut games, because there’s no reliable way of being able to estimate pre-draft how many cut games each faction would be expected to win – if you think a faction will get a few wins in the cut, add a little extra onto their price. Secondly, these values are only the values of the factions if each team is made up of the same level of ability of player, and pretty much by definition of the fact that Greyjoy costs more than Lannister, the Lannister regent should then be able to build a stronger team of players with the gold saved on faction choice – so the raw value is not actually how much you should pay for the faction in the draft format, and in fact paying it probably means you’ll lose. Oh, and thirdly, all the caveats we already mentioned, so I guess there are three main things.
My personal opinion is that you should be looking to pay somewhere in the region of 70-80% of a faction’s raw value to secure it. If you can pay less than that, obviously that’s great. But moving from the hypothetical theory to the post-draft reality, how much did the regents actually pay?
From this, we can see that some of the factions went for about what was predicted; a couple for slightly over, and some were real bargains. The average price a regent paid was 80.86% of the raw value of the faction, slightly more than I was expecting. Of course, this is massively inflated by Night’s Watch, which as previously mentioned is particularly short-changed by the caveats of the data.
So let’s look at who got the best deals:
Greyjoy, as the first to go, went for a fair price. Stark up next went for a slightly less fair price, but still broadly in the realm of being a good deal. Night’s Watch… I mean the data says Kostas overpaid massively, but as previously mentioned the Watch are the one faction where I think you can broadly disregard the data.
If we believe the heavily-caveated data, Matt overpaid for Martell. They should have gone for the same price as Tyrell, not close to twice as much. Sandy, conversely, got about the correct deal for Tyrell. Martin MASSIVELY overpaid for Lannister according to the percentages, but this is one of those cases where we can say “the difference between 1 gold and 2 is almost nothing” and move on.
Of course, the real bargains came from those who didn’t care what faction they got – understandably so given the fact that Josh and Wedge had to draft all-rounder teams as a result due to the uncertainty. Nonetheless, even though they didn’t necessarily expect Targaryen or Baratheon, getting them for 18.48% and 12.32% of their raw values respectively represents tremendous value.
Overall the regents were savvy with the draft. The only one I personally feel got a truly bad deal on factions, Matt, had in my opinion some of the best player-drafting of anyone. How these numbers will play out in practice, of course, remains to be seen, but I for one am looking forward to finding out.
Note: if anyone is interested in the document I used to calculate this, contact me. I’m not precious about sharing and it could be useful for other tournaments going forward!