What This Site Does Not Publish, and Why That Is the Point
No augment win rates, no item tiers, no Arena module, no region split. A plain explanation of each gap and what it would have taken to fill it.
A data site is defined as much by what it refuses to publish as by what it does. Every gap on this site exists for a reason, and knowing the reasons is how you decide whether to trust the numbers that are here.
No augment win rates
This one gets misunderstood more than any other gap, so here is the whole chain.
The public match API does have augment fields: a participant record carries playerAugment1 through playerAugment6. So if a Mayhem match were returned, "who took which augment and who won" would arrive in the same response, and a win rate would be trivial to compute.
But a Mayhem match is not returned. Requesting one returns 403 Forbidden, and Riot has confirmed that is intended rather than a bug. The data is not restricted by a field being blanked out — the match record is simply not served. That closes the compute-it-yourself route at the first step: there is no raw material, so there is nothing to aggregate.
Independently of that, Riot's developer policy also states that products cannot display win rates for augments or Arena mode items, and that this applies to all websites, applications and overlays.
So there are two separate walls, and the site is behind both:
- Availability. Mayhem matches are not served by the API, so the number cannot be computed from an official source.
- Policy. Even with the same data collected some other way, publishing the win rate is the prohibited act. The rule governs display, not provenance.
What exists instead is one of two substitutes. English-language trackers ship a companion app that reads finished matches from the player's own client and uploads them — real Mayhem games, but tens of thousands of them, and the upload is the practice the policy addresses. Sites aimed at the China server draw on that region's far larger published statistics, which reach millions of matches but describe a different player base.
Neither is usable here: the first is the prohibited act plus a client this project does not ship, and the second is another region's data presented as if it were global. The methodology page lays out all five routes and why each one is closed.
What exists instead is client-side collection, which is the practice the policy addresses. Sites that publish Mayhem augment win rates are publishing a number their own data source is not supposed to be distributing.
This site publishes pick rates instead. It is a weaker metric for answering "what is strong" and an honest one for answering "what do players choose".
No champion win rates
The same 403 explains why. Champion win rate would come out of the same match record as augment win rate, so it is blocked by the same wall. Note the asymmetry worth knowing: Riot's policy text names augment and Arena item win rates specifically, and does not name champion win rates for this mode. The barrier for champion win rate is availability, not an explicit prohibition.
The source dataset does not publish champion win rates either: it excludes "win rates, tiers and win-rate-based ranks" from anything intended for reuse. So there is no champion win rate column here, and the tier labels are pick-rate percentiles — which every table showing them states.
No item tiers
Item tiers require per-champion item performance. The source's item file carries names, categories and costs but no tier or coverage fields, so an item ranking would have to be invented. The item pages therefore show gold, category and stats, and nothing else.
No Arena module
Arena augments and Arena items fall under the same policy clause as Mayhem augments. On top of that, none of the available sources expose Arena fields — the Mayhem dataset has no Arena dimension, and there is no separate Arena endpoint. Building the module would have meant both violating the policy and fabricating the data, so it is out.
No region selector
There is no reliable region dimension in the data. The source describes its sample as limited and explicitly not worldwide. Offering a region selector would imply a split that does not exist, so the site labels every page "limited sample" instead and lets you weigh that.
No combined score
A single number that blends pick rate, change and coverage would be easier to sort. It would also hide which input drove the result, and readers would treat it as a strength score within a week. The site shows the components and lets you read them.
No historical backfill
The pipeline builds from the current snapshot. Historical patch files exist at the source, but backfilling every entity page for every patch would multiply the page count without adding information — the older data would be about a patch nobody is playing. Only the current patch has a page.
What this buys you
Everything on this site can be traced to a source labelled in the footer, with a metric definition on the methodology page and a data date on every page. When a number is missing, the site says so with a dash or a zero rather than filling the gap with an estimate.
That is a smaller product than a site that publishes everything. It is a product where every displayed number means what it says, which is the only reason a data site is worth reading.
How to tell whether a data site is trustworthy
Five checks, in order of how often they expose a problem:
Does it show a data date and a patch?
A number without a date is not a measurement, it is an assertion. Every page here carries both in the header. If a site shows a tier list without a patch, assume it is stale until proven otherwise.
Does it name its sample?
"Worldwide" and "millions of games" are claims that require evidence. If a site cannot tell you where its sample comes from or how large it is, the precision of its percentages is decoration.
Does it publish metrics it is allowed to publish?
For this game mode specifically, a site publishing augment win rates is publishing a number Riot has asked developers not to distribute. That is a compliance signal, and it tells you something about how the rest of the data was obtained.
Does it separate popularity from strength?
Pick rate and win rate answer different questions. A site that blurs them — usually by calling a pick-rate ranking a tier list — will lead you to confident wrong conclusions.
Does it disclose what it cannot do?
The least trustworthy sites are the ones with no gaps. Every real dataset has holes; a site with none has either a much better source than everyone else or is filling them in.
What would change this site
Two things, both outside this project's control. If Riot opened a compliant data channel for Mayhem match data, champion and augment performance metrics would become publishable and the site would carry them. If Riot revised the augment win-rate clause, the augment pages would gain the column they are missing.
Until then the honest version is pick rates with the gaps labelled, rather than estimates with the gaps hidden. Corrections and challenges are welcome through the feedback page.