World Cup analysis of knowledge of the crowd

After group stage of the World Cup, we have a lot of data we can evaluate and find out how good Steemit's collective guess was. The median number of points is 24 - half of users is under it and half is over - while average guess constructed by choosing most likely results according to the users gained 29 points and the best guess had 3 more points.
If you choose your results according to Steemit's average, you 94,75 % of users are behind you which is well above 50 % you should get randomly.

Distribution of points after group stage.

Number of users ahead of Steemit's average changing after each match.

The worst result for Steemit's average was Poland's loss against Colombia which 39 % expected to win, 27 % to lose and 34 % tie which was also our recommendation.

The best result for Steemit's average which helped it move ahead of lot of others was Australia's loss against Peru that was expected by only 42 %, while 34 % expected them to win and 25 % tie.

How to pick my teams to win?

The next stage requires you to choose also a score after regular time rather than only a winner and by guessing score right, you get 3 points and 1 point for guessing right a winner. This means that by guessing score a tie e.g. 2:2, you will get either 3 points or none meanwhile guessing 2:1 will also guarantee you at least 1 point for scores 1:0, 2:0, 3:1 and so on. You should not choose tie so that you increase your chances. The next step to choose a score is to look at Steemit's data. Let's assume that we have results 2:1, 1:0 and 0:8. We could compute average and get 1:3 but much better is to choose median so you get 1:0 (median is the number in the middle when you sort numbers from lowest to highest). If median leads to tie e.g. 2:2 we will look at the second most common score.

Exp score = expected score calculated as median on Steemit.

Score	Exp score	Team	vs	Team	Exp score	Best guess	Points
4:3	2	France	vs	Argentina	1	2:1	1
2:1	1	Uruguay	vs	Portugal	1*	1:2	0
1(3):1(4)	2	Spain	vs	Russia	1	2:1	0
1(3):1(2)	2	Croatia	vs	Denmark	1	2:1	1
2:0	2	Brazil	vs	Mexico	1	2:1	1
3:2	2	Belgium	vs	Japan	0	2:0	1
1:0	1	Sweden	vs	Switzerland	1*	1:0	3
1(3):1(4)	1	Colombia	vs	England	2	1:2	1

*The best guess differs from median so that we avoid ties.

Team's score	0	1	2	3	4	5	6
France	63	438	463	75	2	1
Uruguay	175	535	280	50	0	2
Spain	35	200	502	293	12
Croatia	53	276	537	166	8	2
Brazil	35	203	572	220	10	1	1
Belgium	18	140	431	392	57	4
Sweden	202	562	241	33	2	1	1
Colombia	236	587	189	21	6	3

Team's score	0	1	2	3	4	5	6
Argentina	160	558	285	38	0	0	1
Portugal	151	434	381	72	3	0	1
Russia	421	483	121	16	1
Denmark	560	402	75	1	2	0	1
Mexico	332	553	137	15	3	1	1
Japan	591	368	68	12	2	1
Switzerland	288	507	222	21	2	0	2
England	66	309	507	148	12

Conclusion

Choosing according to Steemit's average increased you chance to 94,75 % in group stage so it is a good idea to keep guessing according to the data.
Most people expect the group winner to also win their match except Uruguay vs Portugal and Colombia vs England and most common result is 2:1.

Tools and Scripts

Data from Steemit were downloaded through great library Ditch. The json parsed data can be found in the repository below.

Repository

https://github.com/petermail/worldcup_steemit