Post by Genomico on Oct 25, 2011 18:05:01 GMT 1
How to calculate the exact skill-value of all skills of all cyclist with one decimal accuracy.
2 weeks ago I told that there is a way to calculate all 6 main skills with one number behind the comma accurate. Today I will share this method with you all.
This example will first use cyclists that didn’t have any training at all yet. When a cyclist has got training already, it’s harder to estimate decimal values. I’ll get back to that.
Part 1: Sort your cyclists on the cyclist summary page from lowest to highest value on a skill. Note down the order of the cyclists (e.g. Cyclist1, C2, C3, C4, C5, C6). Now sort the cyclists on the same skill from highest to lowest value (C6, C5, C4, C2, C3, C1). The results won’t be exact the opposite as the previous sorting. Why? Because some skill-values of cyclists are exactly the same!! So in the example C2 and C3 have the same skill-decimal.
So: C6 > C5 > C4 > C3,C2 > C1 at this skill.
At the start of the cyclist’s career, all skills will have a value with 1 decimal and the maximum value of that is .8. So if all 6 cyclist in the example have a skill-value of 1, this means that C1 has the value 1.0, 1.1, 1.2, 1.3 or 1.4. C2 and C3 from 1.1 to 1.5, C4 from 1.2 to 1.6, C5 from 1.3 to 1.7 and C6 from 1.4 to 1.8. With more cyclists the number of possibilities will be lower (i.e. the ranges smaller).
You can do this for all 6 skills and then you will end up with ranges on all skills for all cyclists.
Part 2: For new cyclists (not from a youthscout) the total of all skills will be exactly 12.0. Shuffle the skill-values of your cyclists so that the sum of the skills is exactly 12.0. Keep skill-values from which you know they must be equal (part 1) equal during the shuffling and skill-values that are not equal not equal. I also keep in mind that skills have a higher probability being closer to 2 than to 1 or 3. So in the example, you could start shuffling with C1 having a value of 1.4, C2 and C3 of 1.5, C4 of 1.6, C5 of 1.7 and C6 of 1.8. If they would all have a skill of 2 instead of 1, then I would start with 2.0 for C1, 2.1 for C2 and C3, 2.2 for C4, 2.3 for C5 and 2.4 for C6. If e.g. the sum of skill-values for C5 is lower than 12.0 you could increase his skill-value on this skill to 2.4, but that means that C6 will get a skill of 2.5 or higher.
This puzzling might take some time, but you can end up with “EXACT” skill-values for the large majority of all skills of the cyclists. A few values will still have a little uncertainty, but you will be able to sort them out by buying new ‘untrained’ cyclists.
2 weeks ago I also told that every week waiting with estimating, the margin of error will get bigger. This is because a skill that is trained will get a 2nd (and probably more) decimal(s) and they won’t group up together anymore after sorting them on the cyclist summary page AND you don’t know the exact sum of skill-values anymore. For instance if you buy a cyclist that got flat-training and before buying this cyclist you knew that on flat the skills are like C6 > C5 > C4 > C3,C2 > C1 and after applying the method including the new cyclist are like C6 > NEW > C5 > C4 > C3,C2 > C1 the difference between C5 and C6 can still just be 0.1, because of the training the new cyclist won’t group up with others as it will probably have a value with 2 or more decimals. So if you will do this method after cyclist got their first training already, you will end up with estimations with certain margins. It gets even more complicated if you have cyclists of which you don’t know if they got any training. For instance when you buy a cyclist of a BOT-team of which you don’t know what trainer he had.
Well, because everyone probably got a trainer already for a few weeks, it will be nearly impossible to get all exact values for all skills, but it "was" possible
The method will give you a good guide to get as near as possible or at least give you clues to lower margins of error of you predictions.
2 weeks ago I told that there is a way to calculate all 6 main skills with one number behind the comma accurate. Today I will share this method with you all.
This example will first use cyclists that didn’t have any training at all yet. When a cyclist has got training already, it’s harder to estimate decimal values. I’ll get back to that.
Part 1: Sort your cyclists on the cyclist summary page from lowest to highest value on a skill. Note down the order of the cyclists (e.g. Cyclist1, C2, C3, C4, C5, C6). Now sort the cyclists on the same skill from highest to lowest value (C6, C5, C4, C2, C3, C1). The results won’t be exact the opposite as the previous sorting. Why? Because some skill-values of cyclists are exactly the same!! So in the example C2 and C3 have the same skill-decimal.
So: C6 > C5 > C4 > C3,C2 > C1 at this skill.
At the start of the cyclist’s career, all skills will have a value with 1 decimal and the maximum value of that is .8. So if all 6 cyclist in the example have a skill-value of 1, this means that C1 has the value 1.0, 1.1, 1.2, 1.3 or 1.4. C2 and C3 from 1.1 to 1.5, C4 from 1.2 to 1.6, C5 from 1.3 to 1.7 and C6 from 1.4 to 1.8. With more cyclists the number of possibilities will be lower (i.e. the ranges smaller).
You can do this for all 6 skills and then you will end up with ranges on all skills for all cyclists.
Part 2: For new cyclists (not from a youthscout) the total of all skills will be exactly 12.0. Shuffle the skill-values of your cyclists so that the sum of the skills is exactly 12.0. Keep skill-values from which you know they must be equal (part 1) equal during the shuffling and skill-values that are not equal not equal. I also keep in mind that skills have a higher probability being closer to 2 than to 1 or 3. So in the example, you could start shuffling with C1 having a value of 1.4, C2 and C3 of 1.5, C4 of 1.6, C5 of 1.7 and C6 of 1.8. If they would all have a skill of 2 instead of 1, then I would start with 2.0 for C1, 2.1 for C2 and C3, 2.2 for C4, 2.3 for C5 and 2.4 for C6. If e.g. the sum of skill-values for C5 is lower than 12.0 you could increase his skill-value on this skill to 2.4, but that means that C6 will get a skill of 2.5 or higher.
This puzzling might take some time, but you can end up with “EXACT” skill-values for the large majority of all skills of the cyclists. A few values will still have a little uncertainty, but you will be able to sort them out by buying new ‘untrained’ cyclists.
2 weeks ago I also told that every week waiting with estimating, the margin of error will get bigger. This is because a skill that is trained will get a 2nd (and probably more) decimal(s) and they won’t group up together anymore after sorting them on the cyclist summary page AND you don’t know the exact sum of skill-values anymore. For instance if you buy a cyclist that got flat-training and before buying this cyclist you knew that on flat the skills are like C6 > C5 > C4 > C3,C2 > C1 and after applying the method including the new cyclist are like C6 > NEW > C5 > C4 > C3,C2 > C1 the difference between C5 and C6 can still just be 0.1, because of the training the new cyclist won’t group up with others as it will probably have a value with 2 or more decimals. So if you will do this method after cyclist got their first training already, you will end up with estimations with certain margins. It gets even more complicated if you have cyclists of which you don’t know if they got any training. For instance when you buy a cyclist of a BOT-team of which you don’t know what trainer he had.
Well, because everyone probably got a trainer already for a few weeks, it will be nearly impossible to get all exact values for all skills, but it "was" possible
The method will give you a good guide to get as near as possible or at least give you clues to lower margins of error of you predictions.
Equipe à Grande Vitesse
