Random just as good ?

[Hollandspeler]

Just trying to figure out if using statistics  and knowledge is better than pure random picking.

I am building a data base of predictions from a site on the internet (not this one) and compare their predictions with pure random picks.

 

For so far i have only a couple of days. (not very accurate yet) .

What i see does not impress me much.

btts and win  

site prediction 

1 yes 39      predictions  correct       11

x yes  24      predictions correct       6

2 yes   30     predictions correct       8

total predictions   93   correct   25    =    26.88%

 

random predictions   93     (same matches)  correct 17   = 18.28%

1x2 predictions

site prediction

1    75 pred     correct         39

x     29 pred    correct          8

2      51 pred   correct         29

total 155 games   correct   76   =  49.03% 

 

random prredictions  1x2   (same matches )   155

total correct (1x2)    53.        34.19%

 

Over 2.5

site prediction 

73 predictions         correct    38    = 52.05%

under 2.5

site prediction   

80 predictions      correct    40  =   50%

random predictions   (same matches )   153   correct  73  =  45.75 %

 

Edited January 8, 2021 by Hollandspeler

[harry_rag]

On 1/7/2021 at 11:04 AM, Hollandspeler said:

Just trying to figure out if using statistics  and knowledge is better than pure random picking.

Of course it is, if using it well. Using it badly would be a different story.

 

On 1/7/2021 at 11:04 AM, Hollandspeler said:

I am building a data base of predictions from a site on the internet (not this one) and compare their predictions with pure random picks.

So what you are doing is comparing the predictions on that site to random picks, not comparing the whole concept of "using statistics and knowledge". The quality of predictions on that site could be good, bad or indifferent. The data could even include some picks that have actually been made at random!

 

On 1/7/2021 at 11:04 AM, Hollandspeler said:

For so far i have only a couple of days. (not very accurate yet) .What i see does not impress me much.

I got a flash of Shania Twain off that last comment!  Tiny sample as you say but, from a quick glance, the predictions substantially outperformed the random picks in 2 categories and lagged slightly in the other. So that doesn't really reflect the spin you've put on it.

The whole thing seems a bit pointless. It might be less so if you factored in the betting returns; are the predictions showing a profit to level stakes or a loss? How much more would you win or lose with the random selections? I'd be mildly interested in an update once you've got a sample size of at least 1000 predictions.