From the start, one of the most central mantras of All Elite Wrestling has been that "wins & losses matter!" The presentation of their fictional world is intended to be representative of the sort of thing you might see on a real-world live sports broadcast. Although some matches have some personal grudge involved, the primary motivator is almost always to win and therefore rise in the rankings and get closer to championship opportunities.
On November 8, 2019, less than six months after the first AEW show, they introduced their top 5 ranking system. In each division—Women's Singles, Men's Singles, and Men's Tags— the top 5 ranked competitors (not currently holding a title) would be updated and published (approximately) once per week, and the top-ranked contender would be next in line for a shot at the World title of their division.
From the start, the rankings caused some confusion and contention. Some didn't agree that a fictional sport needed or would benefit from such a thing, and claimed it would never last. Others did not understand "strength of schedule" and would be confused if (for example) a perfect record ranked beneath one with some losses.
Personally, I loved the concept and would defend it regularly against detractors, and tried early on to create my own independent version of a ranking system based on Elo ratings used in chess. But as I went along, I came to see some of the fundamental weaknesses of an Elo system when competitors are not forced to take on all comers. Beating a weak competitor doesn't move your rating much, but doing so repeatedly never stops moving one up slightly, despite the fact that it proves nothing new about one's competitive level.
Below is an example of a chart from my old abandoned Elo system:
Another thing that became clear, much as I hated to admit it, was that strength of schedule could not explain everything that seemed wonky about the official AEW rankings. On Feb 9 of this year, Austin & Colten Gunn, the "Ass Boys," were ranked #1 in the tag division and that night were granted a title shot at the reigning champions.
They had been an undefeated team—six and oh, all time—but in terms of schedule strength their ranking could not be justified: four of those wins were against completely winless teams, and the other two against teams just barely a level above that—Chaos Project and Lee & Brock. That was the moment I decided that I needed to take another swing at creating a rating system for AEW competitors that doesn't reward repeatedly whaling on tin cans and properly accounts for strength of schedule.
After several weeks of ludicrously long hours and hard brain-work for the sake of a completely unpaid hobby project, I finally completed the AEW Tier-based Ranking system. For this system, what I do is put each competitor on a grid:
If one competitor has only wins over the other, that grid space gets a 1, if only losses then -1. If they had a draw or a mix of wins and losses between them, a 0 goes in the grid. Then the spreadsheet looks at the opponent's level number (between 1 and 10) and if the opponent had a higher level, and they have a 1 in the grid, they are raised to a level 1 higher than the opponent. If there is a 0 in the grid space they are raised to a level somewhat under the opponent. A -1 against a higher level opponent, or ANY result against a lower level opponent has no effect on a competitor's level. This continues recursively through all matchups and finally each competitor reaches their current level.
The levels are then linked to tier labels based on distribution. The lowest level, 1, indicates a competitor who has never won, and is assigned to Tier E. Levels 2 & 3 are the C tier, levels 4 - 6 comprise the B tier, and levels 7 & 8 make up the A tier. Levels 9 & 10 make up the highest tier, S.
Once the letter tiers are established, a + or - subtier is added based on each competitor's performance within their tier. The "+" indicates that a competitor has more wins than non-wins against competitors of the same tier or any wins against a higher tier. The "-" indicates otherwise.
'
Once all of this was set up, it allowed me to create a very revealing visual representation of a competitor's wins, losses, and draws in the context of their strength of schedule, by making a stacked bar chart of results sorted into the tiers of the opponents. For example, below is such a chart for Daniel Bryan:
You see here a pattern you will see with almost every competitor, just at differing levels: matches against one or two (MAYBE 3) tiers have mixed results, if the opponent is below that level, the competitor always wins, if above, the competitor always loses.
This is the most useful way to get a really complete picture of a competitor's strength of schedule. However, to do a ranking, it has to be boiled down to one number. To try to distill this information, I associated a base rating to each subtier in increments of 10, so 10 points for a match against an E tier opponent, all the way to 90 points for S+. Then I take a weighted average between the ratings associated with all of the tiers with relevant information: from the highest tier the competitor has wins or draws in, down to the lowest tier that they have any draws or losses in.
A small bonus or penalty is then applied based on if recent matches support the rating. From this I arrive at a single-number rating which is an attempt to boil the complexity of strength of schedule to one number which could then be used to rank AEW competitors. Below is an example of the sort of rankings arrived at and how they compare to official AEW rankings, specifically the current rankings for the AEW Women's division:
The column headed "HTNL" is the highest tier opponent that competitor has any wins or draws against. After the forward slash is the number of non-losses against that tier. "RY W/D/L" stands for "Running Year Wins Draws and Losses". The column headed "AEW" is the official AEW rankings, and "Ch" is the current championships held. As you can see the current and recent champions hold high ratings, but the official AEW rankings have little to no relationship to the ranks based on tiers and strength of schedule as my system figures it.
This set-up seemed to be working fine for assigning competitors appropriately into tiers and I was entering the AEW results into the database until I reached August 18, and suddenly my distribution of tiers went crazy. Below is August 18 on the right and the previous week on the left:
I did a little digging into the data and was able to find the issue: on 8/13, Daniel Garcia was ranked #57 and had in his record a defeat versus #45 Lee Johnson. But then he scored a win over #20 ranked Matt Sydal. This extreme upset had a cascading effect on everyone either of them had had matches with and who they had in turn matched with, leading to suddenly about half of the roster being S ranked! Which seems to take much of the meaning away from being "S".
This led me to add toggles to turn something I called "Fluke Filters" on or off. If the Fluke Filter for a tier is on, it takes two wins, rather than one, over the level below to reach the level for that tier. If the tiers are too top-heavy, I turn on the Fluke Filter for S-tier, then if necessary, A-tier,etc. until the tiers seem reasonably balanced. Below is the rankings for the same date, 8/18, but with the S & A filters on:
As you can see, this fixes the problem of tier inflation, but it also changes some of the rankings slightly, moving Christian Cage into the first position, ahead of champion Kenny Omega, and moving MJF behind Miro and Darby Allin. Changes to rank wrought by seemingly irrelevant changes in parameters is one of the reasons not to treat these rankings as any kind of exact science. It is very reasonable to assume, for instance that there were perfectly justifiable differences of parameters behind AEW ranking Orange Cassidy ahead of Mox, Max, and the Murderhawk. AEW ranking Perry and Hobbs 4 & 5 ahead of many others with significantly better strength of schedule, on the other hand....
To return to the misestimate of strength that originally prompted me to start this journey, here is my tier system's rankings of the tag division the week before Billy's kids were ranked #1 in their division and received a title shot:
Below are the charts of two teams with the exact same record, but one has had a significantly more difficult schedule:
Five of Darby & Sting's 6 wins were against B-tier opponents and their lowest opponents were C-tier, while 4 of the Gunn boys' 6 were against E-tier job teams and their highest opponents were C-tier.
Yes, all of this was so that when I say that the Gunn Club sucks and didn't rate a shot at the titles, I could back it up with math and statistics.
Suck on that, Austin.
Last week, AEW evidently updated their rankings, based on the ranks given in the competitors' lower thirds, but they seem to have forgotten to publish them. The awesome @AEWmetrics put out his best estimation of what they should have been on Twitter, which is what I'm going by until and unless AEW puts out retroactive rankings for last week:
Hopefully, AEW gets back on the ball and publishes the new rankings this Wednesday. Presuming that they do, my plan is to publish the first regular edition of "All-Elite Wins & Losses Matter" ASAP once the rankings are out, discussing the differences and similarities between the official ranks and my unofficial ones and discussing the meaning behind all of it. I hope to continue doing so regularly (weekly? monthly?) when rankings are released, as well as irregularly focusing on a particular reign or feud. (Hoping to put one out about Hangman's reign, leading up to his defense against Punk.)
If all of this has prompted any questions or confusion, or ideas or thoughts, I would love it for this article to be a jumping off point for dialogue rather than an end-point. If you are curious about my spreadsheet, the two (color-coded green) user-friendly tabs allow you to change parameters with drop-down tabs to look at rankings as of different dates or with changes to other parameters, and look at the bar charts of strength of schedule of different competitors. Links to each divisions' spreadsheet are located at the bottom of the spreadsheet section of this website. Please reach out to me if you need any assistance!
Comments