Six slots that come online: giving item builds real weight
Since launch my draft simulator ended and the items just happened invisibly. This is the story of giving every hero a real six-slot build you pick and order yourself, one that comes online during the match and actually swings fights.
Here is something that quietly nagged at me for weeks. You would finish a draft in Dota Captain, hit simulate, and the match would play out with net worth climbing and fights breaking out, and nowhere in any of it did the items you would have built actually exist. They were happening, in a vague averaged sense, buried inside the numbers. But you never chose them, you never ordered them, and you never got to feel the moment a hero came online. That bugged me more the longer I sat with it, because if you strip items out of Dota you have stripped out roughly half the game. A draft tool that ignores what you build is only telling you half the story, and it was telling that half with a straight face.
The gap I kept avoiding
The reason I avoided this for so long is that items are hard to do honestly. It is easy to slap a shopping list onto a hero card and call it a build. It is much harder to make that list mean something inside a simulated match, to have it arrive at the right time and change how a fight goes. For a while I told myself the draft was the interesting part and the items could stay abstract. That was a convenient story, and it was wrong.
Real Dota is not just who you pick, it is when your picks become dangerous. A carry with an empty inventory is a liability. The same carry six items deep is a different creature entirely. If my sim could not represent that difference, it could not really represent Dota. So in 0.7.0 I finally stopped avoiding it and built the thing properly, from the strategy phase all the way through the live match.
A hero is a timing, and items are how you buy it
Here is the idea I wanted the whole feature to teach, because it is one of the most useful ways to think about Dota. A hero is not a fixed quantity of power. A hero is a timing. Some heroes are terrifying at ten minutes and fade. Some are helpless early and unbeatable once they finish a key item. What you build, and the order you build it in, is how you buy the timing you want. Rush the item that ends the game early, or stack survivability so you live to your late-game peak. Those are real choices with real tradeoffs, and until now my tool let you make none of them.
So every hero you draft now gets a six-slot item build in the strategy phase. Each one starts with a curated default so you are never staring at an empty grid, but you can fully edit it, pull any item from the catalog, and order it however you like. During the match the build completes in order as net worth climbs, and each item adds genuine power as it lands. What you build and when it comes online are now actual strategic levers, not decoration. And to be clear, this counts in PvP too, not just solo games. The items you and your opponent choose both matter.
Ranking items by how much they actually swing a fight
Once builds mattered, the obvious next question was which items matter most, and I did not want to answer that from vibes. So 0.7.0 also added an Item Impact page. It ranks every item by how much it actually shifts a fight in the sim, and it shows you the heroes that tend to build each one. The goal is to turn a fuzzy instinct (this feels like a strong item) into something you can look at and reason about.
I also wanted the numbers to stay honest over time, because a stale item is worse than no item. Item costs and recipes sync from the live patch, so when Valve reprices something or changes a recipe, the values the tool reasons about move with it. It is a small piece of plumbing that quietly does a lot of work, and it means the impact rankings are anchored to the game as it actually exists, not the game as it was some patch ago.
Rebuilding the room around the builds
Adding builds meant the old strategy and match screens no longer fit, so I rebuilt them around the new reality. Each hero card now carries a visual build picker with a searchable catalog. You drag to reorder, and you can hover an item to see its impact right where you are making the decision, which is where that information actually helps. I also made the hero portraits bigger, partly because they had been cramped for ages and this was a good excuse to fix it.
The live match got the same treatment. The roster now reads like a proper scoreboard, with net worth, last hits, KDA, and the items each hero has completed so far. When you watch a game play out, you can finally see the builds coming online in real time and connect that to what is happening in the fights. That connection, seeing an item land and then seeing its owner start winning skirmishes, is the whole point. It is the thing that was invisible before.
The follow-ups I owed people
No feature ships finished, and 0.7.2 was mostly me paying down what 0.7.0 left rough. The big convenience add was saveable build presets. If you find a six-slot build you love, you can save it and reuse it on any hero, and presets sync to your account so they follow you. I do this constantly myself now, which is usually a good sign a feature was worth building.
I also owed the mobile players a lot. The mobile draft got reworked with a denser hero pool, an always-visible pick and ban button, and coaching moved into a bottom sheet so it stops fighting the pool for space. Clicking your already-selected hero now deselects it, which sounds trivial but was a genuinely annoying dead end before. I fixed a bug where item builds would occasionally show duplicate entries, which looked broken because it was. And I bumped the strategy phase timer to three minutes, because the old timer assumed you were only assigning roles and now you have builds to think about too. If you missed the mobile draft rework in its own writeup, this is where a chunk of it landed.
What this still is not
I want to be straight about the limits, because it would be easy to oversell this. The match engine is an approximation. It does not know every ability interaction, every timing window, or every reason a specific item is broken against a specific lineup this week. When it ranks an item or plays out a fight, it is reasoning from a model I built, and my model gets things wrong. Treat the impact numbers as a thoughtful second opinion, not a verdict.
What I am confident about is the direction. For the first time the tool takes seriously that half of Dota happens in the shop, and that a hero's power is a curve over time rather than a single number. That is a foundation I can keep improving, and I would rather ship an honest approximation of the right idea than a polished version of the wrong one. If you build something clever and the sim disagrees with you, tell me. Most of what is in 0.7.0 exists because someone did exactly that.