0

I am writing a small application to aggregate Historical Fencing matches results. This data is then used to calculate a couple of factors:

  • Effectiveness: simply shows how many of your matches are won:
   won matches 
   ___________
   all matches
  • Fierceness: shows how well you do in lost matches, if you are getting crushed or put up a fight:
  points scored in lost matches / all lost matches 
  ________________________________________________
                           2
  • Integrity: shows if you can consistently win without losing a lot of points, so high integrity means matches won with a big advantage in points, low integrity means you win just barely:
  (won matches * 5) - (points lost in won matches)
  ___________________________________________________
                 (won matches * 5)

The first two are rather straightforward but the last one is a bit more tricky to implement. What is 5 you may ask? That is the maximum amount of points you can score in a match and is the issue here, because it is a constant.

What I would like to have is a formula, that could normalize different values for the maximum amount of points so that I can calculate it for all the matches a user was in. Right now it makes sense only in a tournament setting, where all matches have the same maximum points amount.

I did not come up with these formulas, but was given them. My knowledge of statistics is functional, but not very deep. If you have better ideas on how to calculate them or where I could start some research, please let me know.

Here is a set of examples I have generated. I decided to do them by hand, because it would make a more readable example.

Piotr Szymaszek
  • 1
  • 1
    Would "sparing" mean sparring, perhaps? – whuber May 15 '22 at 18:42
  • Yes, in my language you write it with a single "r", hence my mistake. – Piotr Szymaszek May 16 '22 at 22:17
  • In English, using "sparring" and "fights" is an unusual and obscure way to describe fencing outcomes. More people might understand your situation if you referred to these as the results of fencing matches. Regardless, you appear to be asking about "feature engineering." Whether these formulas are appropriate depends on how well they work. That's not something we can determine without actually analyzing the data. – whuber May 16 '22 at 22:22
  • Alright, I will try to rewrite my question sometime soon and provide examples using the said formulas. They will be generated via a script, but should provide enough information on how well they work. – Piotr Szymaszek May 17 '22 at 10:36
  • @whuber I have updated my question, please let me know if the data set is big enough or should I make a bigger sample. – Piotr Szymaszek May 22 '22 at 14:10
  • I still do not understand what you might mean by "normalize" or how that could be useful. The problem is that you have defined effectiveness, fierceness, and integrity in terms of formulas rather than in terms of any generalizable or meaningful properties, so you offer no basis to extend them to any other situation. – whuber May 22 '22 at 15:06
  • And this is why I have posted this question in the first place. I am not sure where to go from these formulas. I have provided the idea of what these are suppose to display, but I am not sure what you mean by generalizable properties? Effectiveness is just how often you win, which I think has good enough formula as it is. Fierceness means to show how much you can score despite losing. Integrity shows how clean of a victory you can achieve and that means not losing points in won matches. – Piotr Szymaszek May 22 '22 at 21:00
  • Normalization is perhaps not the right word for it. In general, I want to calculate these factors for all the matches a person ever had. The issue is, they will not have the same maximum score - for example, one match can have a maximum score of 5, the other of 3. Therefore the fierceness formula will probably needs adjustment (although I am not sure), but the integrity one is outright impossible to implement in these circumstances. – Piotr Szymaszek May 22 '22 at 21:10

0 Answers0