EPT
But ESPORTS betting is easy. Just look at the front page of Dota2lounge. Right after Merlini joined Zephyr, the odds for Pokerface vs Zephyr were 29%/71%. That's far too skewed in Zephyr's favor given those two teams' recent performance, and it's unclear whether Zephyr was even the favorite. People got caught up in the Merlini hype and bet accordingly.
This blog is about Dota2lounge's TI4 prediction event and how I made my decision on who to bet on. My goal for this blog is to help bring Dota 2 betting to a level of sophistication closer to that of sports betting. The only way this will happen is if more people exploit bad odds.
http://dota2lounge.com/predict
The above link goes to a page where you were once able to bet on which team you thought would get top 3 at TI4. Take a look at the odds for EG: what those odds are saying is that if EG had placed 1st, someone who bet on EG would both keep the items he/she bet and receive 3.06x the value of those items from other people. As it happened, EG placed 3rd, so EG fanboys got their rares back along with 1.02x the value of their bet.
Some fairly high value items could be bet in this way. You could bet arcanas or Rotten Staches (~$60 value). Accounts were limited to two items bet maximum, but dedicated bettors could (and did) make multiple accounts.
Let's take a look at the odds for the teams as they existed shortly before TI4 began (they changed slightly as betting closed):
![[image loading]](http://www.teamliquid.net/staff/motbob/Blogs/rares1.png)
First of all, look at the difference in odds between Newbee and DK. You might remember that, in the time period leading up to TI4, DK and Newbee played an unusual number of tournament games against each other, going almost even in their series. Yet the odds for DK and Newbee are very different. This is an example of "favorite team bias," which is another way of saying that the odds on popular teams are almost always too high, and the odds for teams that aren't in the eye of Western bettors can sometimes be way, way too low (the Zephyr vs Pokerface match is another example of this). If people thought DK was the favorite to win, why was Newbee, a team clearly right up there with DK in skill, so far down in the rankings? This alone meant that a bet on Newbee was probably a +EV bet.
The next thing to look at are the odds for Alliance and Na`Vi vs the other Western teams. Bettors apparently thought that Na`Vi had better odds of winning than Newbee and had more than 5x more of a chance to place than C9 did. Why the huge difference between C9 and Na`Vi? The answer might lie in bias towards CIS teams, which would also explain why Empire had such (comparatively) good odds. It might also lie in the fact that C9 never won any of the major tournaments they attended this year, and that leads to another type of bias: wins stick in the minds of people who watch tournaments. The highs of Alliance's and Na`Vi's 2014 were higher than C9's, but their lows were certainly lower (see: D2L for Alliance). People remember the highs, especially if they're fans of a certain team and want to have faith. In fact, that last statement might be a better explanation than anything else I've written: people wanted to have faith. At any rate, thanks for the rares!
I wanted to find the best team to bet on, and it wasn't easy. The odds for NAR and the SEA teams were ridiculously high, but were they high enough to warrant a bet? Did a Chinese team down on its luck (LGD) really only have a 1-in-69 shot at winning?
So that led me to simulate the whole tournament in Excel. I gave teams an ELO and gave the ELOs the opportunity to fluctuate based on chance, to reflect the fact that it's hard to know the true skill of a team based on results. For example, Fnatic was assigned an ELO of 1300 with a chance to fluctuate up to 125 points, while Newbee was assigned a 1480 ELO with a chance to fluctuate up to 75 points. Fnatic was sort of an unknown quantity due to the whole Era situation, while Newbee had posted very consistent results.
I ran the tournament 10,000 times and kept track of how often they placed first, second, or third. Then I used the odds on Dota2lounge to calculate expected returns on each team, and came up with this:
![[image loading]](http://www.teamliquid.net/staff/motbob/Blogs/rares2.png)
Accordingly, I bet my most expensive items on Newbee. Too bad I only bet Kunkka whale swords and not arcanas...
Obviously, when you use this sort of method, assigning the ELOs is everything. I think there was room for improvement in that regard. VG's performance in TI4 definitely showed that they were better than even the highest ELO my model allowed them to be. I think I need to switch to a model that allows for teams to get much better than their current performance has shown, if the stars align. Also, I fell into the trap of overvaluing C9's recent (bad) performances and gave them a low ELO that they didn't deserve. I think I relied too much on Gosugamers's ELO ranking. That ranking underrated Chinese teams before TI4, so when C9 lost to a bunch of Chinese teams in a row, their ranking tanked.
Anyway, ESPORTS betting is easy as long as you're not an Alliance or Na`Vi fanboy. Or maybe it's easy if you always bet on Chinese teams. Or maybe it's not easy and requires close attention to the scene and thoughtful analysis. Pay attention to the biases that the rest of the betting community has, try to eliminate your own biases, choose your bets wisely, and soon you'll be swimming in rares. I think "choose your bets wisely" is the most important piece of advice here. Just bide your time, and eventually a bet will come along that will make you smile and think "Thanks for the rares" in advance.
.
