Analyst Predictions during Worlds 2015: Did they win you money?

20 November 2015By: admin

analyst-desk-worlds-2015
Leading up to the League of Legends World Championship in 2015 I were thinking about the analysts and the job they are doing during the tournament. As a veteran noob when it comes to League of Legends I really appreciate their work and I think the casters and analysts are among the best out of all the esports out there.

This lead me to an idea I had been pondering for some time: given that these experts have a great understanding of not only the games, but the players and the dynamic between the pro teams and their strength and weaknesses and so on, shouldn't they be able to pick out the winners of most matches with ease? I certainly would think so, and probably most of you would agree this to be true. Sure there are probably people out there that are even better at this than they are, they aren't hired only for their skill at the game.

What I wanted to find out were how these predictions they make before each match during Worlds would fare if they actually were bet on. So I jotted down every prediction that were made during the tournament, as well as the odds offered on the teams that were to play. From this I calculated the return (or loss) from said predictions.

Method and Rules to this experiment

In calculating these bets we had to make a slew of different assumptions in order to make this possible:

We used the predictions offered up during the analyst desks pre-talk of the match in question.

If this were not applicable, we added them only if it was a clear favorite (odds < 1.30) and the bet won, to give the analysts the benefit of the doubt.

We place a 1 unit bet, and for discussion purposes we assume that 1 unit = $100 in this example.

The odds were acquired from Bet365, usually a day before the event.

We are also gonna try a couple of different betting strategies that makes more sense than simply wagering on every prediction that is made. Betting $100 on IG and then $200 on C9 does not make a lot of sense. So we will also use a consensus method where we bet $100 on whichever team "wins" the vote from the analyst desk. We will also try a progressive bet system where we bet more if there is a unanimous vote (when all 3 vote for the same team), thus betting $200 on such an event.

The Results

I will not bore you with all the numbers, stats and data, I will link to the sheets I used so you can look over and use the data yourself if you are interested, but here I will cut straight to the findings:

Straight Betting

If you would have followed all of the predictions given by the analysts and bet on them at Bet365, you would have lost money over the course of the tournament. If you bet $100 on every game, this would come out to a net loss of $1,014 to be exact. Not a very lucrative method of making money, that is for sure.

The return on investment (ROI) is naturally quite bad sitting at -6.18%

Results: -$1,014 (ROI: -6.18%)

Lean Betting

By using a lean, where we only place one bet at a time based on which team gets the most votes, one would think this would lead to a greater return, or at least a lesser loss in our example. By moving to a single bet per match we reduce the loss to $646 overall, which is still quite bad, but better than losing over a grand. It should be noted that the ROI in this instance is much higher, at a whopping -11.96%, as we reduce the amounts of bets by a third and still can't reduce the loss by a large enough margin.

Results: -$646 (ROI: -11.96%)

Lean and Progressive Betting

We also try our hand at upping our bet size once there were full consensus among the analysts. So if all are predicting the same winner, which should indicate a strong team that should take it more often than not, we double our bet size. The end result here is that our net loss is bigger than in the case of our lean betting, which makes sense given the results we have seen so far. In this case our bankroll suffered a set back of $775 over the tournament, but the ROI was a bit lower this time around with -8.71%.

Results: -$775 (ROI: -8.71%)

Individual Analyst Results

	Monte.	YamatoC.	Crumbz	Zirene	Spawn	Tabe	TOTAL
Number of bets	25	34	26	32	27	12	156
Net result	-$368	$351	-$210	-$179	-$423	$68	-$760
ROI	-14.71%	10.33%	-8.08%	-5.59%	-15.67%	5.67%	-4.87%

So the results here are nothing crazy. The losses are somewhat evenly spread out, with Montecristo and Spawn having the biggest losses, closely followed by Crumbz and Zirene. The former of these had a very good run to start out the first 4 days, but then had a horrible streak at the end of the tournament.

yamatocannon-analyst Tabe ends up as a winner in his predictions, which is quite solid, but he has a very small sample size with merely 12 bets to go after, and he sat during the the last day with Group C, where the predictions where kinda straight forward.

But what ends up being a pleasant surprise is seeing YamatoCannon not only being a winner, but a solid one at that. And he had by far the biggest sample out of them all. He had a bad stretch after the first day, but caught up big with two big predictions during the knockout stages.

Discussion

The results we have found here might be a bit disheartening for some: if none of the professionals, who get paid to analyze these matches, are able to come out ahead against the bookmakers, then surely there are no chances for me to do any better?

Luckily, that is not the conclusion you should draw from this article. Although the chances of you beating the bookmakers are slim anyways. One of the main flaws with our "analysis" is that the analysts are simply being asked, on the spot, who they believe are going to win a specific match and they are not presented with any betting odds to go with that, nor do they have any money on the line. If we asked them to predict given a specific set of odds, then this might have ended up differently.

Obviously the analysts will almost always go with the favorites leading into a match, and that is exactly what we saw happened. If you should take anything away from this, it is that blindly betting into favorites is not a winning strategy. Bookmakers are aware of this phenomenon where many punters like to bet on the big favorites as they see the outcome as a "sure thing", and thus shade the odds even further on those teams and players.

Another thing to take from this is that the sportsbooks are pretty good at setting the odds correctly. Usually how the bookmakers operate is by coming up with the probabilities on the outcomes of a sporting event, then convert that to betting odds and then add a theoretical hold, i.e. they lower the odds about 5% on each side, which is the cut they take for offering these bets to their customers.

If the odds they had set were approximately close to the true probabilities, then the expected value on each bet would be about -$5, and with 164 bets placed this would turn into -$820, a little more than what we ended up with in our case during Worlds 2015. This can probably be ascribed to most of the bets going on favorites, which usually have attached a premium on them by the bookmakers.

There are other problems like our exclusion of Pinnacle Sports in this instance. If we had used the odds from Pinnacle instead, we could probably have increased our ROI from -6% to something like -2%. That is a huge difference! Think about it. If you had an ROI of around -2% during Worlds, then moving to Pinnacle could move you from a small loser to a small winner!

Alos, and this is a big one, the amount of data in this close to laughably tiny. Sports betting operates with very tiny limits and one can have long streaks of either winning or losing. Simply concluding from about 50+ matches is not very wise, and we will refrain from doing it, but we will use it to illustrate some basics about esports betting.

Conclusion

So our conclusion should end with a reiteration of what we have already have said a couple of times: no, the analysts wouldn't have made us money. Yes, they are still very good at what they do. And yes, it is hard (type: very) beating the bookmakers at their own game. But there are ways of doing it, and this little article shows a couple of the hows: