Home Articles Thrones W.A.R. UK 2019 Draft – Player Evaluation

Thrones W.A.R. UK 2019 Draft – Player Evaluation

by AGOT.Admin

The 2019 UK Thrones W.A.R. recently had its draft. Previously, I evaluated how good a job the regents did in buying their factions; this time, we’re going to look at players.

(Warning: if you like reliable data, this is significantly less reliable than the previous article was, and is probably worthless to you for a number of reasons I’ll get into.)

The way the below figures were calculated were as follows:

  • For each player on a team, figure out their win-record on Jousting Pavilion with the faction they’ve been drafted to.
  • If the player has zero history with the faction they’ve been drafted to, instead take their overall win-record, but deduct 5% as a “playing an unusual faction” penalty.
  • Take the win record playing the faction, multiply it by 7 (number of rounds in the tournament), then apply the following two modifiers to it:
    • Multiply by the faction’s win ratio. On average, each faction will get 42 wins (as each faction plays 84 games). So, using the projected wins values in the first article I wrote, I could apply a Faction Win Modifier. For example, Greyjoy is projected to get 52 wins in my “raw data” analysis before, so now any Greyjoy player’s win record is multiplied by (52/42). Likewise, Lannister’s projected to get 35 wins in the “raw data”, so all Lanni win records are multiplied by (35/42).
    • Initially when I did this, the total of wins equalled 348. However, there are only 336 swiss games, so that clearly wasn’t right. To fix this I multiplied everyone’s scores by (336/348), and that resolved the issue.
  • With each player’s projected win rate established, it’s then a simple matter of adding them up for each faction to figure out how many wins they’re expected to get.

Now, those of you so-minded will have noticed a number of flaws with this system. They are as follows:

  • Firstly, all data is inextricably linked to the faction analysis data I already provided, which had its own lengthy list of caveats. So if you feel that, for example, Lannister and/or The Night’s Watch were hard done by with that data, it will carry over into this article too.
  • Additionally on the faction data, there’s an element of “double jeopardy” to the calculations. Not only is an assumption being made about the win-record of the faction, but also the players were already using that faction when accumulating their wins and losses with it. This means that, for example, a Greyjoy player who played Greyjoy a lot this year would already be having their win-record flattered by how good the faction is, and is now having it flattered again; equally, a Lannister player who played Lannister a lot this year would be suffering twice over.
  • Some of the players are seriously lacking in data on their chosen faction. This can result in wildly unlikely scenarios. I’m not going to name names at any point throughout this article, partly because on a player-to-player level the data is rubbish and partly because it kinda feels almost like doxxing or something, but to give one example, one of the Tyrell team’s players has played Tyrell a single time. It was a jank deck at a small GNK, they didn’t win a single game at the event, and as a result have a 0% win record with the faction and are projected to go 0-7, despite being a good player.
  • For other players, it’s not that they’re lacking in data so much as that they’re lacking in data for when they take things seriously. For example, going 0-5 with a deck taken to the Brighton Charity Joust with the aim of getting achievements probably shouldn’t count the same as the results taking the same faction to Stahleck. Yet, out of a desire on my part to remain impartial in collecting data, both count the same.
  • This takes any-and-all JP data, regardless of when it is from. Results from the last 12 months with a faction count the same as results from 2016. One of the Baratheon players has only ever played the faction once on JP, back in 2016. But they went 5-2 with it then, so they’re projected to go 5-2 here!
  • It assumes each player on a faction has an equally good deck for this event. This doesn’t properly represent factions that have one great deck, maybe two, but then less impressive decks to fill their slots out with. As a result the number of people projected to cut from each faction is particularly worthless data.
  • Lastly, unlike the other article, I didn’t write an API macro for this, I did it by counting wins and losses myself, so there’s a chance for human error in the actual colleciton itself.

Before regents projected to do poorly get upset, I beseech thee to take into account those caveats! I’m not presenting the below values as my personal predictions for how everyone will do – merely what the data, as it can realistically best be objectively interpreted, shows.

Again I cannot stress enough the layers of “lacking in validity” this data has, but if we think the evaluations on how much factions are worth hold water, then it looks like the following is true:

  • Lannister’s immense talent pool doesn’t consist of anyone so unbelievably good that they can get to 5 wins with the faction in the format. They have a lot of good players predicted to 4-3 or 3-4, but breaking into the cut is tough.
  • Baratheon, Martell, Night’s Watch and Tyrell had pretty average drafts, and will be much of a muchness at the end of swiss.
  • Stark had the best draft. By focusing on players who are good at the faction, a lot of people are projected to finish 5-2. Amusingly, one of those people (I’m committed to not naming names but I suspect you’ll guess who if you know the field) has never competitively played Stark before, but due to the “5% lower than their overall win percent” stat used in that instance coupled with their very healthy win-rate with the faction they’re actually loyal to, still are projected to go 5-2 and make the cut.
  • Targaryen drafted a lot of average players who are brought down to 3-4 by the “Targ are slightly below average” projection, but also a few people who can really make the faction work, resulting in them having more people projected to make the cut than Greyjoy despite Greyjoy projected to get more wins in the swiss.
  • As can be seen on the below table, Targ were the only faction to significantly gain on their projected wins compared to the raw value of the faction. Partly that is down to the way I’ve analysed this data.(see second caveat listed previously), and partly it suggests that even with different amounts of money dedicated to it, it’s very difficult to buy significantly more wins from the player pool – it’s much better instead to spend money on the factions. Targaryen here very much are the exception that proves the rule.
  • Greyjoy are a better faction than Stark, but despite paying close to the same amount as Greyjoy, Stark retook the projected lead in the draft by drafting players at fairer prices. This results in Stark losing only 1 projected win from the raw value, while Greyjoy loses 5.
  • The top 3 projected factions are separated by only a single win after swiss, and combined have 12 of the 19 players projected to go 5-2 or better. This means that the winning team could be decided late in the day – potentially even the final itself!

This concludes my statistical analysis on the draft. Now the only thing left to do is wait a month and find out just how drastically wrong I was with all this nonsense.

You may also like

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More