It's more of a displaying issue because it's hard to show timings on items as well as win percentages
Most matches you view can show items and item timings but if you're collecting data, showing when people got their items as well as winrates means you'd end up with HUGE fucking grids that 99% of the time would show information we already know
Most the item winrates aren't very useful anyway
dotabuff just uses the data shown on the end of the match.
that's why items like brown boots have 32% winrate, as if you only had brown boots by the end of the game, it pretty much means you've stomped
and aegis, obvioulsy having high winrate as dotabuff only tracks the end screen data.
so when you end up going rosh, and then not needing the aegis to win the game.
so don't use that item data, it's incredibly useless and biased
The problem is because ability to buy a late game item is often correlated with how much resource advantage a Player has. Usually the player that is able to buy a 5th/6th slot item, is usually the team that is ahead.
Of course it is entirely possible to improve item statistics by removing resource advantage from the equation. For exAmple, only include stats for games where the item was bought when there is no net worth advantage skewed towards one team. In other words just look at games where both teams are equally resourced. the item must be bought at the point in time when both teams are equal.
idk
i almost feel bad for winning a lot of desolator on the nyx assassin
now noobs will think its a good item on him just coz of the 57%wr xD
ye thats why i thought of buying it the first time
however it feels like the item is a major reason why we win the fights
ofc not even close to the impact of an eul or an aghanim, sheepstick, but still good
Будь ласка, увійдіть до системи для написання коментарів.
Dotabuff currently appears to display the win rate for items that are in player inventories at the end of the game, rather than the win rate of items bought at any point during a game and then possibly sold. This causes the win rate of early game items to be skewed downward, while that of late game items is skewed upward (I think for different reasons).
An example is that tangos have a 26.55% win rate despite being purchased by most players in most games, while Aegis of the Immortal has a 98.09 win rate, presumably since it is highly unlikely for a losing hero to be alive with Aegis at the end of a game.
Question: Does Dotabuff use end-of-match data because it is less resource intensive than storing data from the item progression available in Dota 2 match summaries, or because they are not able to collect this data as a third party service?
If the data is unavailable or unfeasible to obtain and collate then that's understandable, however if it were able to be obtained I feel it would make the Items section of this website far more informative, especially for early/mid-game and transitional items. At the moment the win rates seem less to indicate the value of buying a particular item than how little of a slot detriment they are in the late game. Of course late-game items would stay skewed towards a high win rate, since they will continue to be less often purchased at all by losing teams in normal-length games.
This is the part where you tell me what I've overlooked any why I'm stupid. Have a swell afternoon everybody!