Second step, is your name spelled correctly on the BikeReg results? Did you enter as "Dave" but your profile says "David?" Did they spell your name as "Daivd" because they were in a hurry?
Third step, are the results for that race formatted in the standard BikeReg way? Do they have crystal-clear columns? If not, then the whole race is probably being held up because I have to manually format the data.
1) Put your real name in the first box
2) Click your real name when it comes up
3) Put the misspelling of your name in the 2nd box
4) Click your misspelled name when it comes up
5) Click "Merge Racers"
The graph on the left is wacky, I'll give you that. What it's showing is which "fifth" of the field you finish in, with what frequency. So there are 5 sections of the field you could finish in: top 20%, 20%-40%, 40%-60%, 60%-80%, bottom 20%. If you finished in the top 20% 5 times you'll see a value of 5 on the vertical axis above the 1 on the horizontal axis. 4 times in the 40%-60% range? That will give you a value of 4 above the 0.6 on the horizontal axis. Get it? Basically, if you're good there's a big spike on the right edge. If you're bad there's a big spike on the left. If you're average, the graph spikes in the middle -- and if you're highly variable (one day you win, next day you lose), the graph will be flat.
Are the results poorly formatted, inconsistent text? Probably not.
Are they excel or nicely formatted text? Now we're talking! Email me a url and we'll see what we can do.
We also take the average of your last 10 races to use as your "carried points" -- these are used for generating new points when results get added.
The real trick is how you score points in a given race. Really to rock some math? Ok, open up a race you did with points in another browser window and click the "Show Points" link. Above each category will magically appear a table listing five names, and "quality" and "median" scores.
For each of those five names, the high and low score is crossed out. They don't affected the scoring. The other three are averaged to get quality. Now, multiply that by 0.85 This is the number of points you get for winning the race.
Now get out your graph paper. Make a 2D plot with a horizontal axis that starts at 1 and ends at the number of racers in the field. Now make a vertical axis that starts at "Quality" and goes to infinity (hope you have big paper).
In the middle of the horizontal axis, make a point that is equal to "Median". This is the median point value of all the racers in the race. Now draw a line from the lower left corner straight through the median point, out to infinity. The points each place scores are equal to the value of that line above wherever they placed on the horizontal axis. Linear interpolation!
I'll add a picture here eventually.
Updated 9/3/2010! Look, I finally added a picture!
* - Ok, so when the site started in 2006 the points where just for fun. Now they get used for staging and rankings across the country... SO STOP DNF'ING!
First off -- data is limited to the last 12 months. I don't care whose butt you were whooping back in 2001 and you shouldn't either. Time heals all wounds.
Next, we introduce a stat called "defeats." A defeat is when Racer A beats Racer B by ten or fewer places. This makes Racer B angry, and makes Racer A boastful. Why only ten places? Well, we had to draw the line somewhere -- when you're getting beat by 20 places, it's hard to call that person a rival, hmmm? Each defeat has an attribute called the margin, which is the number of places between the racers -- the best margin for Racer A is 1, meaning he beat Racer B by one place.
So when we show Nemeses and Victims on a racer history page, we're showing who victimized you the most often/whom you victimized the most, and we break ties by lower average margin. Beating someone three times by 7 places might get you a note on their wall -- but do it three times by one place each time and they'll be having nightmares.
Just kidding. We actually have code to take care of this, but unlike the "duplicate racer name" function we don't trust you to do it yourself. Instead, kick us some feedback with your name, along with (1) what state you're from and (2) what team name we can expect to see on your results, and we'll take care of it.
- The race was uploaded within 60 minutes of you looking at it.
- Your category had fewer than six finishers
- The category name was "new" to the database and requires a manual check before we run points on it.
It's probably that last thing. PROMISE we will take care of your points before the next list is run. But you can still email us about it if you want.
So what we do is, we take all your results. Ever. Then we use some nonlinear regression to create a best-fit curve for your career's trajectory, based on the points scored in each race.
The we look at each race result and see where it fell relative to that curve. This tells us if you had a "good" or a "bad" race relative to your normal performance.
Then, we look to see how much the various course conditions (Technical, Hilly, Accelerate-y, Wet) correlate to you having good or bad results. Technically we do this by computing the Pearson product-moment correlation coefficient between the two datasets.
Lastly, we multiply the correlation coefficient ("R") by 20 so that we aren't showing you some boring number that is mostly decimals.
In the vast hierarchy of things that are made up around here, this might be the most made-up thing. But it's still interesting.