🌱Jayensee
LightNews LightNews
banner
jayensee.bsky.social
🌱Jayensee
@jayensee.bsky.social
35 followers 87 following 290 posts
Hobby league of legends analyst, also talk about whatever topics catch my interest (usually related to data analysis)
Giving the butterfly app a try

Jayensee on YouTube and GitHub.

jayensee.github.io
Posts Replies Media Videos
jayensee.bsky.social
one of these junglers has a bounty (and it's not jax)
July 17, 2025 at 9:48 PM
jayensee.bsky.social
I was looking more at the difference between models 5 and 6

I think there might have been some issue optimizing the alpha_0
June 21, 2025 at 9:49 PM
jayensee.bsky.social
Here's an essay I wrote for the PoE subreddit about crafting in Path of Exile 1 and 2

It was removed for being low effort (go figure) but it seemed a waste to not post it somewhere
After reading a ton of comments online about why crafting in PoE 2 sucks, I think people really misunderstand what makes crafting in PoE 1 interesting. Crafting an item is a series of decisions to shape it in the direction you want. The amount of RNG is mostly irrelevant, the only things that matter are what you gotta do to get to the interesting decisions and how interesting they are. What makes crafting in PoE 1 so engaging is the incredible breadth of viable decisions: at any point there are tons of options that each let you impact the item in different ways. Here's a showcase of interesting crafting in PoE 1: medium-budget int stacking helmet 1. Start with a high-tier non-fractured and non-influenced ES base between ilvl 75 and 79 2. Spam spite essences until you get good affixes, at least 2 prefixes, and an open suffix 3. Bench craft a prefix if there's space for it 4. Slam a hunter's exalted orb 5. Finish by bench crafting something useful and slamming exalts until the item is full (alternatively, bench craft something bad to block and hope you don't get a low tier of something you can bench craft) By choosing to use a base in that level range and filling the prefixes you're guaranteed to hit % int on the hunter slam, but you can't roll some t1 mods. Every time you use an essence you have to choose whether you want to move on to the slam or not, and if you don't have an open suffix or enough prefixes you have the option of exalting / yolo annuling, if the affixes are good enough to justify it So you start off with an interesting decision and then are faced with interesting decisions every few clicks In the end you get an item that you very much shaped into what you wanted, despite not having full control of the outcome ### A showcase of bad PoE 1 crafting: +2 minion level helmet 1. Start with a high tier non-influenced and non-fractured ilvl 75+ base (influences and fractures lower your chances of hitting in the next step) 2. Spam an average of 3000 (THREE THOUSAND!!!) orbs of alteration 3. Aug+regal+whatever, there aren't many good mods for minion builds on a non-influence base so it doesn't matter much This craft is bad because there's only 1 interesting decision, which is to spam alteration orbs. There are no interesting decisions that can come up while spamming because there's no possible outcome of an alteration orb that would be worth paying attention for 3000 clicks. You write "+2 to level" in the search bar and mindlessly spam until you see the item light up. And when it lights up you get an item with 1 useful mod and basically no potential to get better (seriously, you sometimes see minion characters at the top of poe.ninja wearing magic bone helmets with +2 minion levels and nothing else). ### A showcase of terrible crafting in PoE 1: +1 power charge crown of eyes 1. Buy an uncorrupted crown of eyes 2. Vaal orb it 3. Repeat step 1 if you don't hit (~1/90 chance) Similarly to the minion helmet you get 1 interesting decision for the entire "craft". Instead of left clicking 3000 times you have to buy 90 uniques; it's way less RNG, but it's an order of magnitude less fun because trading in PoE sucks. ## What about crafting in PoE 2? You can't reroll or scour so you only get interesting decisions when you either buy or loot an already good item. If you want a +2 minion level helmet in PoE 2 you have to acquire a few dozen "bases" and slam them with transmutes/augs/regals/exalts. Sound familiar? It's pretty much the same as the worst type of "crafting" in PoE 1. And that's for the most part how you craft every item in PoE 2. You sometimes get interesting decisions on whether to use a chaos orb or not and in some cases you can choose the direction of the craft a little bit with essences, but the vast majority of "crafts" are just slam then vendor. The problem is not that its too RNG / not deterministic enough. **The problem is that interesting decisions are distanced from each other by a lifetime on the trade site / buying from vendors / grinding maps.** And for the most part those decisions aren't even that interesting (do I sell as is? or do I gamba the chaos for a low chance of bricking the item and a lower chance of improving it?) Even the runes are a more boring version of the crafting bench: the mods are blander, there are fewer of them, you can't use them on jewellery, and you can't use them to block bad affixes before exalt slamming. Finally, what I find most frustrating is that picking up rares in PoE 2 isn't even more exciting than in PoE 1, why don't we have fractures in PoE 2? Or >20% quality items from drops? Or synth implicits? Those can actually get me excited to pick up a rare item in PoE 1, and they don't make crafting any less interesting (the opposite, really). Fractures and >20 quality don't even require any design effort. If they wanted to make loot worth picking up then they should make the loot more interesting, not the crafting less interesting.
January 20, 2025 at 9:09 AM
jayensee.bsky.social
A month after PoE 2 went into early access, it's still getting more players than PoE 1 did at its peak

This game is stupid good...
And it's only like 30% finished

I'm pretty sure I haven't even launched any other game since it came out
January 8, 2025 at 6:08 PM
jayensee.bsky.social
My arena recap looks a little better
A very information dense graphic with stats from my arena games this year
December 19, 2024 at 9:34 AM
jayensee.bsky.social
Information dense graphic showing stats from my Aram games this year
December 19, 2024 at 9:24 AM
jayensee.bsky.social
I think this is my 3rd year in a row getting the artist clip from Sammy Rae

Toby Fox takes the lead just because the Undertale OST is my default "lock in" music, kinda wish it didn't count so I could see who's number 6 (probably Miniature Tigers)
Spotify recap Top artists: 1. Toby Fox 2. Miki Fiji 3. Sammy Rae & The Friends 4. Amelia Day 5. Wic Whitney Top songs: 1. Clay Pigeons by Michael Cera 2. Chamomile by Wic Whitney 3. Tell It to the Volcano by Miniature Tigers 4. Sink In by Amelia Day and Miki Fiki 5. Coming Home Song by Sammy Rae & The Friends
December 4, 2024 at 7:28 PM
jayensee.bsky.social
At least in The Rating of Chess Players Past & Present the only mention of players who quit early that I can find is this

(screenshot of the quote because it doesn't fit within the character limit)
"(...) people who learn to play chess but show only limited aptitude tend to drop out of competitive activity, while those with more aptitude stay for many years. Thus in any rating pool the population of players at higher levels is comparatively more stable, while at lower levels there is greater turnover. These transients probably have little effect on the pool as a whole. It is the new and improving players who have the most significant effect on all ratings."
November 29, 2024 at 6:53 PM
jayensee.bsky.social
Sneak peek of some of the results I got

The numbers are the information content predicted per game minus the information content predicted by just the side advantage and draw probability in hundredths of bits

(confusing metric but should be good for comparing across datasets)
A table with prediction ability of rating systems across several datasets The results are very inconsistent, no rating system performs very well or very poorly across datasets
November 28, 2024 at 1:18 AM
jayensee.bsky.social
YouTube algo is so weird, randomly started showing the short I posted yesterday to a bunch of people and then abruptly stopped completely

The retention wasn't even going down as it showed up in more feeds
A graph showing what portion of the viewers swiped away from the short Starts at nearly 100% then drops to 33% after around 7 hours before slowly climbing back up to around 50% A graph showing how many times the short was shown in viewers' feeds Starts very close to 0, then suddenly starts increasing by around 120 per hour after 7 hours, then jumps by nearly 400 at hour 14 before stopping almost entirely
November 27, 2024 at 9:58 PM
jayensee.bsky.social
Yeah, bsky's ToS is pretty lax, if it's not permissive enough to allow training with it, then I don't think any license is

From my understanding they extend this license to developers who use their tools with almost no extra restrictions (basically just require moderation if you develop an app)
A screenshot of the Bluesky terms of service: Bluesky’s Permission to Use Your User Content. You keep your ownership of User Content, subject to the license below. Bluesky does not own rights to your User Content except as provided in that license. By sharing User Content through Bluesky Social, you grant us permission to: i. Use User Content to develop, provide, and improve Bluesky Social, the AT Protocol, and any of our future offerings. For example, we can store and present User Content to other users in Bluesky Social. This allows us to show your posts in the Bluesky app to other users; ii. Modify or otherwise utilize User Content in any media. This includes reproducing, preparing derivative works, distributing, performing, and displaying your User Content. For example, we can resize your posts to fit the Bluesky mobile or desktop app, or feature examples of User Content for promotional purposes; or iii. Grant others the right to take the actions above. For example, we can grant content moderation tools access to User Content in order to monitor Bluesky Social; iv. Remove or modify User Content for any reason, including User Content that we believe violates these Terms, the Bluesky Social Community Guidelines, or other policies.
November 27, 2024 at 11:03 AM
jayensee.bsky.social
The family back together 🤗
November 22, 2024 at 6:46 PM
jayensee.bsky.social
These sorts of images always make for a neat rabbit hole to fall into

The previous image is from Hubble, and captures only visible-frequency light

This one uses combined data from GALEX (UV light) and the Spritzer Space Telescope (IR light), and I personally think it looks even cooler
Another picture of the Helix Nebula, this in this one the "white of the eye" is blue instead of red and the "iris" is black like the background but with a shining pink center
November 22, 2024 at 10:19 AM
jayensee.bsky.social
Here's the picture used in its Wikipedia page to save ppl some googling
A picture of the Helix Nebula taken by the Hubble Space Telescope It is looks like an eye with red instead of white and a blue iris in the center
November 22, 2024 at 9:35 AM
jayensee.bsky.social
Post mettigel
A photo of mettigel, a German dish consisting of raw pork meat sculpted to look like a porcupine using raw onions for the spikes and olives for the nose and eyes
November 21, 2024 at 4:20 PM
jayensee.bsky.social
Totally forgot that the patch would drop today, so I guess this is the end of the series

She did end up reaching 50% winrate in top, how much of that was because of the changes is unknown
November 20, 2024 at 6:00 PM
jayensee.bsky.social
Over on the item stats, nothing interesting is happening, but I'm still posting them for posterity 🤷‍♂️
A screenshot with the item stats of Topbessa at all ranks A screenshot with the item stats of Jungbessa at all ranks
November 20, 2024 at 1:42 AM
jayensee.bsky.social
We have now reached an important point in this series of threads: we finally got a graph that looks like an animal 🦆

This resemblance doesn't really mean anything, tho, and the curve is just trending toward what you'd expect from a standard high-rank-skewed champ
A plot of Winrate v Rank that kinda looks like a duck
November 20, 2024 at 1:35 AM
jayensee.bsky.social
Day 13

Diamond+ in top lane has the highest long-term predicted winrate and it's sitting at 49.4% as of right now

That number will probably go up over time, but we'll see how much
The core stats of Topbessa on day 13 Winrates: All ranks: 47.17% Diamond+: 48.01% Iron: 47.33% The core stats of Jungbessa on day 13 Winrates: All ranks: 44.67% Diamond+: 41.75% Iron: 46.98% A plot of pickrate and winrate over time for Topbessa A plot of pickrate and winrate over time for Jungbessa
November 20, 2024 at 1:27 AM
jayensee.bsky.social
No
A kitty with blushing cheeks and angry brows drawn on its face
November 19, 2024 at 1:42 PM
jayensee.bsky.social
say pizza to drugs say no to yes
November 19, 2024 at 1:10 PM
jayensee.bsky.social
> exist in hopeless world > look closer > hope catsmellingcamera.jpeg
November 18, 2024 at 12:16 PM
jayensee.bsky.social
Also, I'm just covering top and jungle because they're her most popular roles, but in mid she's already crossed 50% and with a good build (aka cyclosword) you could potentially be looking at a winrate as high as 54%

So if you're a mid lane player, you might want to give her a shot
November 17, 2024 at 6:48 PM
jayensee.bsky.social
Cyclosword's popularity climb has slowed to a crawl and it's shaping up to be kind of a balance nightmare scenario for Riot

When an alt build has a much higher winrate than the main build, they always have to consider the risk that it gets picked up and Ambessa's winrate suddenly spikes
A screenshot with the item stats of Topbessa at all ranks A screenshot with the item stats of Jungbessa at all ranks
November 17, 2024 at 6:36 PM
jayensee.bsky.social
The rank curve keeps flattening, although I am surprised that it really seems like the middle ranks are the ones climbing most

(remember not to trust the winrates at the highest ranks tho, they have very few games)
November 17, 2024 at 6:31 PM