Who is the real winner of Dota 2's world championship?

Once a year, Valve invites the 16 strongest Dota 2 teams in the world to one arena and lets them play for a prize pool and a trophy called the Aegis of Champions. The tournament is called The International; it has been running since 2011, and for most professional players, it is the only result that matters.

This year's tournament concluded on Sunday, August 23, in Shanghai, where Team Spirit beat Team Vision 3-2 in the grand final. Team Yandex took third place and BoomBoys fourth. Between them, all 16 teams divided about $3.3 million. Meanwhile, prediction markets paid out $204,199,859 to people who had bet on all matches throughout the event.

Predictbook has already covered some of these traders, including the Polymarket account that has traded $892 million and the top-150 trader who made $477,000 from a single Dota 2 bet. This report identifies the rest of them and shows how a market of that size was distributed.

The majority of contracts traded on Polymarket. Its 282.2 million contracts represent $164,195,037 of trading against $18,216,613 on Kalshi, so about 90% of the money was concentrated in one venue.

It's quite contradictory that Kalshi ranks first among prediction market venues on notional volume, open interest, and platform transactions by a significant margin, and on the biggest esports event of the year it handled a ninth of what Polymarket did.

At the same time, the two venues did almost the identical number of trades here, 622,475 against 586,845. But the average Polymarket trade was $263.78 while the average Kalshi trade was $31.04.

Part of the gap is just how the two venues list things. Polymarket ran one page for the Team Spirit against Team Vision game on August 23, with both teams on it, and you clicked whichever one you wanted. On the other hand, Kalshi ran two pages for the same game, one asking whether Team Spirit would win and one asking whether Team Vision would win, each keeping its own separate books. But they are the same match. It happened 319 times. Of the 703 listings on Kalshi, 638 were basically one market.

The same asymmetry can be seen in trader activity during the qualifiers and main event. Qualifiers ran from June 9 to 28 and drew $27,888,488 across 111 series, while the main event drew $152,713,551 across 58 series on both Polymarket and Kalshi, averaging $251,248 per series compared to $2,632,992. The remaining $1,809,611 came from the outright winner markets.

Peak activity day was August 21 at $27,450,824, with 20 August behind it at $23,589,766. Finals day did $21,064,043 and was only the third-largest day of the tournament. On 17 and 18 August, during the scheduled break between the group stage and the playoffs, trading fell to $136,752 and $110,348, down about 99% from the day before. So, across both venues, 95.5% of the dollars in match and qualifier markets traded on the day of the match itself.

The tournament drew 21,936 Polymarket wallets, of which 11,450 finished in profit and 10,195 at a loss. The median winner made only about $4.84, while the median loser lost $23.15. 10,635 wallets finished within $10 of exactly even, and only 734 in the entire field made more than $1,000, while 906 lost more than $1,000.

Despite how many accounts turned up, only 1,640 out of 21,936 put serious money on the line. Those 1,640 deposited $157,553,967 between them, which is 86.6% of everything that went into these markets on Polymarket. The typical account outside that group put in just $19.98. And half of them still lost. The largest 220 accounts, almost exactly the top 1%, accounted for 72.8% of total value. Ten accounts placed 26.3%.

If we split every account into 10 equal bands of about 2,193 wallets, then only the largest band finished the tournament in profit. The largest group put in $177,944,356, which is 97.8% of the $181,925,882 that Polymarket accounts deposited altogether, and finished $473,415 ahead. That $181,925,882 is Polymarket on its own, which publishes account identities. The $204,199,859 paid out across the tournament adds Kalshi's $22,205,799 to that.

Eight of the nine groups below them lost money, and the losses widen the further down you go, until the bottom group, which deposited $4,074 between 2,194 accounts and gave back $891 of it, 21.9% of everything they put up. The sixth group is the only exception, up $20,019 on $64,728 deposited.

When splitting the wallets by behavior, for 5,004 accounts the money was made and lost buying and selling, and as a group they finished $7,582,812 up. For the other 16,932, the outcome was decided at settlement instead, and they finished $7,611,971 down.

So 77% of everybody who touched this tournament had their result decided by who won the match, and as a group they funded the 23% whose result was decided by trading the market. Being right about Dota turned out to be worth considerably less than trading it.

The other split worth making is between the accounts posting prices and the accounts hitting them.

There are two ways to buy a contract. You can pay the price already showing on the screen and own it instantly, or you can post your own lower price and wait for somebody to come to you. Waiting is normally the profitable choice, and Polymarket even pays a small rebate for doing it.

Not on Dota 2's The International markets though. The 5,609 accounts that mostly left offers up lost $2,351,035, while the 14,122 that mostly clicked on them made $838,149, on almost identical volume. The 2,205 that did a mix of both, never committing to either, made $1,483,726, which is more than the other two groups managed between them and on less volume than either of them traded.

The reason is in how financial markets work. You put up two prices at once, a buy at 58 and a sell at 62, and if both get taken, you have made 4 cents and own nothing. However, if, in the case of Dota 2, a teamfight ends and the contract is suddenly worth 80, your sell at 62 gets taken instantly, because it is now the cheapest thing on the screen, while your buy at 58 sits there untouched. You are now down 18. Which is exactly why the 2,205 wallets won.

The 10 most profitable accounts made $5,302,924 between them, more than the actual prize that the tournament paid to all 16 teams. The 10 worst lost $2,673,536.

0xaabec5…e9158, the third most profitable TI tournament trader, traded only six markets, made 17 trades in total, was in and out within 13 hours, and profited $406,692. It never left a price on the screen once. It only ever took prices other people had left there.

0x2ba6c2…e644c did the same thing, but smaller. He traded three markets, made 14 trades in under 12 hours, and won $156,149.

The other end of the table doesn't look this good, though. Every one of the 10 largest losses was decided at settlement rather than in trading.

0x03448f…0af4db traded just three markets, made 148 trades, under nine hours from first trade to last, $693,522 deposited, and lost every single penny. That is the largest single loss of the tournament.

0x5dab5e…d322e1 made $262,480 trading, which on its own would have ranked it inside the top 20 accounts of the tournament. But then lost $624,120 at settlement across 17 markets and finished down $361,640.

Two others also gave back every cent they deposited. 0x9100cc…637ade across 39 markets and 798 trades, and 0xbb5fbe…ca2e21, which came within $10,980 of it, losing $199,585 of the $210,565 it put into six markets.

With all these wallets covered, there's one important question unanswered: what did the house earn?

Both Polymarket and Kalshi, as it turns out, did rather well.

Kalshi held its fee rate at 7% for the whole tournament. Polymarket ran 3% across every qualifier market opened between 9 and 29 June, then raised it to 5% for every main-event market opened from 6 August onwards.

Across all 1,209,320 combined trades, Kalshi collected $450,054 in fees, while Polymarket collected $2,311,306. So, together, the venues made $2,761,360, almost two times more than the team winning the tournament and 83% of the total prize pool.

Valve spent a decade building The International into the event that made professional Dota worth playing at all, and the prize pool peaked at $40,018,195 in 2021 before the crowdfunding model behind it was retired. Valve announced the end of the Battle Pass on 20 June 2023, and not because it had stopped selling. It had grown into the biggest update Dota shipped each year, which left the other 11 months barren and pushed real game patches back by months, so Valve cut it and spread the content across the calendar instead.

The prize pool went with it. TI11 paid $18,930,775 in 2022, TI12 paid $3,380,455 in 2023, and TI13 paid $2,370,725 in 2024. This year's $3,324,933 is 8% of what the same trophy was worth five years earlier.

It would be easy to read the shrinking prize pool as declining interest. But people are obsessed with Dota 2. It's a global game, and the audience never went anywhere.

The International 2026 grand final drew 1.79 million peak viewers, the third-highest peak in Dota 2 history and slightly ahead of TI 2025's 1.78 million. Hours watched rose about 1.9% year over year, and day one of the main event pulled 804,800 peak viewers, nearly 2.5 times the Esports World Cup 2026 Dota 2 final.

The money, however, moved. With the Battle Pass gone, the six-figure bounties that once went to pro teams now flow to prediction-market traders and the venues that host them. Polymarket and Kalshi collected $2.76 million in fees on TI 2026 markets, almost twice the winner's share and 83% of the entire prize pool. The top 10 Polymarket wallets alone made $5.3 million, more than the $3.3 million split by all 16 participating teams.

So who is the real winner of Dota 2's world championship?