8 World records today..

Torsten · Post by **Torsten** » 23 Apr 2009, 21:25

Just a thought:

If you have a slightly statistical look at the scores then there is a chance of 1 by 100 for every ending beetween xxx,00 and xxx,99 meters. So on 40 hills there are normally 4 hills were your score is between xx,00 and 0,09 and around 1 or 2 hills were you will cracked an full meter by 1 or 2 centimeters, for instance 139,01m or something like that:

Waldemar has got 6(!) of those pretty cool records:

Austria   HS197  2009-03-15  216.02m
Sweden    HS121  2009-04-12  130.01m 	
Estonia   HS185  2009-04-12  205.02m 	
Denmark   HS93   2009-04-12  100.02m 	
China     HS252  2009-03-22  280.02m
Bosnia    HS111  2009-04-12  117.01m

Another thing is the probability of getting a length of only some centimeters over a 10m-mark, for instance 120,02 cm ... the probability of jumping less then 10cm over a full 10m-mark is 1 by 100. So normally you only got one of this records in your list, if at all

(there are just 40 hills ... not even the half of a hundred)
Hmmm ... Waldemar did this on 5 of the 40 hills :

Sweden    HS121  2009-04-12  130.01m 	
Denmark   HS93   2009-04-12  100.02m 	
China     HS252  2009-03-22  280.02m
Belgium   HS126  2009-04-12  140.03m
Turkey    HS177  2009-04-12  190.07m

I've compared the records of Michał Skalik, geir mikael reijners and me (as probably not cheating jumpers) with WALDEMARs, Josefs and Stepans scores. Waldemar is the only one with such a statistical error.

So to me it looks like he is setting the scores manually or he can jump longer than the actual records but is resetting them before an upload for getting "prettier" scores. That's just a thought ... maybe it's a coincidence .. but why does this happen to him of all the people?

Another thing:

Germany   HS261  2009-04-12  289.32m
Bulgaria  HS258  2009-03-29  289.32m

Funny, huh?

Severin · Post by **Severin** » 23 Apr 2009, 22:58

Thanks for that piece of mathematics

Of course coincidence can play a part, but that are some heavy things that would occur at one time...

Nice to have you back, no one yet looked t it with such a view

Torsten · Post by **Torsten** » 24 Apr 2009, 10:27

Oh, i can go more detailed, if oyu want to

Probability for doing records which are only 0-9cm over a full 10m-Mark (40 hills):

Number of    |             |
such records | Probability | Sum (all numbers up to here)
-------------+-------------+---------------------------------------
     0       | 66,8971759% |  66,897%
     1       | 27,0291620% |  93,926%
     2       |  5,3239326% |  99,250%
     3       |  0,6811757% |  99,931%
     4       |  0,0636425% |  99,995%
     5       |  0,0046287% |  99,9997%
     6       |  0,0002727% |  99,999986%
   7..40     |  0,0000140% | 100,0%

So, the probability of doing 5 of these jumps on the 40 hill is 0,0046287% (a chance of 1 by 21600).
And the probability of doing 5 or more of these jumps is 100%-99,995 what means 0,00491544% (1 by 20345)

This means:

The chance, Waldemar did his records by jumping and without any manipulation (resetting too long hillrecords before upload or editing the files) is
1 : 20000

But all these calculations are basing on a random distribution of the hill sizes and no psychological factors like "wow, i did the 150 .. just by some centimeters, but thats enough, so let's stop jumping on this hill"

@Severin: DU warst doch den ganzen Winter nicht "da". Freu mich um so mehr, dass DU wieder dabei bist. (und hoffentlich hab ich mich nicht verrechnet *g*)

Severin · Post by **Severin** » 24 Apr 2009, 12:52

What now would be very nice, would be to see what were the old world records on each hill beforce Waldemar did his. Unfortunately it's not possible to see the result of more than the current longest jump of a jumper.

If he is really able to just type in the results perhaps the world be a system visible, like he just makes his result about the next full meter higher, or whatsoever...

To get your table:
If you would do 20345 jumps, statistically there would be 5 jumps like Waldemars?

The average hillsize in DSJ is 176,6 m, so Turkey is next to average.
To make a jump in Turkey over hillsize you need about 15 seconds.

20345 jumps x 15 seconds = 84 hours and 46 minutes of constant gaming ( without searching wind, launching the game, watching replays, etc. ).

So just jumping.

Waldemar did the 5 jumps from march, 22th to april, 12th, so 22 days
that means jumping 3 hours and 51 minutes every single day.

Ok, that seems possible, however he did 4 of these jumps on the same day, the famous april, the 12th ...
I don't want to do the mathematics right now again, because I am really not good in it, and I am definitely not sure, if the things above are correct

but 4 of these jumps in one day seem unbelieveable...

But on the other side, we don't even know if he did all these jumps on the same day, as he also could've just set the date of his computer to that day everytime he jumped again. Then DSJ would use that date everytime

Torsten · Post by **Torsten** » 24 Apr 2009, 13:16

Severin wrote:...
To get your table:
If you would do 20345 jumps, statistically there would be 5 jumps like Waldemars? ...

nope ...
you can do such "10meter-mark-breaking-by-less-than-10cm"-jumps every 100th jump ... the chance of such a jump is 1:100

There are 100 of 10cm-corridors on every 10m
(Example for 250,00..259,99)
1 250,00 ... 250,09
2 250,10 ... 250,19
.
.
11 251,00 .. 251.09
.
.
50 254,90 ... 254,99
51 255,00 ... 255,09
52 255,10 ... 255,19
.
.
99 259,80 ... 259,89
100 259,90 ... 259,99
1 260,00 ... 260,09
2 260,10 ... 260,19.
.
.
and so on

So normally ONE of 100 jumps is in corridor 1 ... the corridor i spoke about, because it is the "prettiest, cooles, best looking" corridor IMHO

So, if you have 40 records and only every 100th jump is in corridor 1, then the probabilty of having 1 record, that is a jump just a few (0..9) cm over a 10m Mark, is 27%

The probabilty of having not even 1 of such record is 67% (this means: all of your 40 personal records were jumps, into corridor 2-100 ... das ist also bei 2/3 aller Spieler so, dass sie keinen einziger rekord haben, der so aussieht, dass sie nur ein paar cm über eine 10m-Marke gesprungen sind .. kannste ja in der Rekordliste mal nachschauen ... sollte aber hinkommen)

But the probability of having 5 or more of such records at the same time (as waldemar has) is: 0,0049% and this means only 1 of 20000 jumpers will have this situation in his recordlist.

Grufugl · Post by **Grufugl** » 24 Apr 2009, 13:39

Torsten wrote:But the probability of having 5 or more of such records at the same time (as waldemar has) is: 0,0049% and this means only 1 of 20000 jumpers will have this situation in his recordlist.

Interesting. This must mean that statisticaly one can expect to find maybe one single dsj-player with such records, and certainly not very likely several. So what then would be the chance that exactly this player at the same time happens to be the one with allmost all the world records by coicidence?

Torsten · Post by **Torsten** » 24 Apr 2009, 13:58

Grufugl wrote:So what then would be the chance that exactly this player at the same time happens to be the one with allmost all the world records by coicidence?

THAT's the question, as i already mentioned above:

Torsten wrote:maybe it's a coincidence .. but why does this happen to him of all the people?

But the chance for "waldemar is the one with 5 or more of these records" is still 1:20000 ...

Severin · Post by **Severin** » 24 Apr 2009, 13:59

Torsten wrote: But the probability of having 5 or more of such records at the same time (as waldemar has) is: 0,0049% and this means only 1 of 20000 jumpers will have this situation in his recordlist.

ah ok, then forget my statement...

Torsten · Post by **Torsten** » 24 Apr 2009, 15:00

And there is another statistical "problem" with his scores:

Norway   	HS138 	146.59m 
Finland   	HS218 	233.22m 
Czech Republic 	HS192 	208.16m 
Japan   	HS248 	282.32m 
Italy   	HS146 	154.28m 
Russia   	HS223 	240.52m 
Slovakia 	HS180 	197.54m 
Austria 	HS197 	216.02m 
Canada       HS107 	117.18m 
Poland   	HS269 	296.38m 
Switzerland 	HS157 	168.71m 
Sweden   	HS121 	130.01m 
Germany 	HS261 	289.32m 
USA         HS171       182.97m 
France      HS235       255.32m 
Slovenia 	HS281 	327.78m 
Estonia 	HS185 	205.02m 
Romania 	HS228 	250.43m 
Denmark 	HS93 	100.02m 
Ukraine 	HS203 	220.41m 
Belgium 	HS126 	140.03m 
United Kingdom 	HS85 	92.54m 
Kazakhstan 	HS164 	178.34m 
Bulgaria 	HS258 	289.32m 
Korea   	HS174 	189.13m 
The Netherlands HS100 	108.12m 
Spain   	HS242 	265.59m 
Australia 	HS149 	160.57m 
China   	HS252 	280.02m 
Belarus 	HS214 	236.14m 
Hungary 	HS117 	124.73m 
Croatia 	HS278 	315.66m 
Lithuania 	HS153 	164.51m 
Serbia 	        HS69 	73.48m 
Turkey   	HS177 	190.07m 
Argentina 	HS240 	263.26m 
Georgia 	HS207 	230.97m 
Latvia   	HS80 	86.14m 
Bosnia    	HS111 	117.01m 
Iceland 	HS275 	314.72m

Do you see it? Have a look at the last number in each length!

You will not find any 0 or 5!

Because every 5th jump will have a length ending with 0 or 5 the probability of not having one single record with this last number is: 0,013% or 1 : 7500

Finding a player with 40 records, including ZERO records with 0 or 5 at the end PLUS including 5 records with less then 10 cm over a 10m-Mark should normally be impossible,
cause the probability for that (both probabilitys are independent!) is:
1 : 150 Million

So it looks like the lengths are done manually, 0s and 5s are skipped (maybe because he thinks it looks too artifically

or he cannot generate correct checksums for such numbers ... what the completely same lengths in GER an BUL are indicating, too) and too many lenghts were generated for showing new milestones (like 190 in Turkey and so on)

BTW: The google cache of the worldrecords from april 06 is saying the same: no records with 0 or 5 ...

Schnitzel · Post by **Schnitzel** » 24 Apr 2009, 15:37

BUSTED!

onkelsmerte · Post by **onkelsmerte** » 24 Apr 2009, 16:18

This proves that he cheats ? Math was never my strongest subject

Severin · Post by **Severin** » 24 Apr 2009, 16:39

It does not really prove it, but it's quite a strong indication that there's something wrong with his results...

Torsten · Post by **Torsten** » 24 Apr 2009, 16:44

onkelsmerte wrote:This proves that he cheats ? Math was never my strongest subject

This says: if you have a look at the hillrecs of all players at this moment then NORMALLY 1 of 150 Million players with 40 records on the 40 hills will have scores, that will show this pattern:
- zero records of this 40 records have a 0 or 5 at the last position
- 5 or more records will look like 130.02, 140.01, 280.01 (0...9 cm over a 10m-Mark)

For Waldemar's records does this mean:
a) he doesn't jump normal with random result (he is manipulating the scores manually)
b) it's just a coincidence and everyting is ok

The chance for b) is 1:150 Million, so it's the same chance as if i would say to you "think of random number between 1 and 150000000" and then i would tell you exactly THIS number (5986307 for istance).
Another example: it's easier throwing a dice 10 times and always getting a 6 (with only ONE attempt).

Grufugl · Post by **Grufugl** » 26 Apr 2009, 11:42

Torsten wrote:Finding a player with 40 records, including ZERO records with 0 or 5 at the end PLUS including 5 records with less then 10 cm over a 10m-Mark should normally be impossible, cause the probability for that (both probabilitys are independent!) is:
1 : 150 Million

Wow

I guess now he is thinking hard what to write in his next rec

If he still avoids 0 or 5 at the end, the statistics will keep looking extremely bad for him, and if he chooses 0 or 5 it will look as if he read your post and is manipulating to mend the problem

Torsten · Post by **Torsten** » 26 Apr 2009, 19:04

Grufugl wrote:Wow I guess now he is thinking hard what to write in his next rec If he still avoids 0 or 5 at the end, the statistics will keep looking extremely bad for him, and if he chooses 0 or 5 it will look as if he read your post and is manipulating to mend the problem

It doesn't matter if future records will be more random or if they are showing 0s and 5s. If i have a look at the scores on april 24th then the chances of seeing score like Waldemars with the described pattern is 1:150000000 ... so if he is reading this thread and is changing his ARTIFICIAL style of scoring then this wouldn't change the chance of seeing such scores at the day i was looking at them (april 24/23)