A couple weeks ago, I decided to make a spreadsheet to keep track of my own daily challenges competition with someone else. I also included a bit of "statistics" for fun. Seeing the twitter bot ( https://twitter.com/KlepekVsRemo ) I tough that some of you might be interested in some Remo vs Patrick data! Tonight I decided to swap the data in my spreadsheet and replace it with Patrick and Remo runs.
These are more "statistics" than statistics. I'm not a statistician and I chose to bend a lot of rules. Hence the stuff here must be taken with a grain of salt, especially anything related to standard deviation (t test), the data right now doesn't seem to adhere to a bell curve. This means that it may be horribly wrong.
(The link is further down if you want to skip my brief summary of what data I have now.)
As of January 31, 2014:
- Remo has 138 consecutive runs.
- He had an average run of 61 965 $, reached on average level 2-1 and has an average rank of 1612.
- His best run made him 571 975 $ while his worst run only made him 625 $.
- He has amassed 8 551 175 $ over the course of all his runs.
- His median, 30 838 $, point toward his runs having extreme values (occasional runs that are way better than his average run)
- His standard deviation, 95 643 $, point towards great variance between runs (his first run is really great, then the next one can be really bad).
- Patrick has 5 consecutive runs and 19 total runs.
- His average run is 79 588 $, just a bit better than Remo, reached on average level 2-2 and has an average rank of 1877.
- His best run is 328 875 $, 243 100 $ less than the best run of Remo.
- His median and standard deviation, 43 800 $ and 88 845 $ respectively, are similar to those of Remo. The same thing could be said about them (extreme values and great variance).
- Combined, they have raided 10 063 350 $ out of mines, temples and other special places.
- Patrick has won 58 % of the runs while Remo has won 42 %.
- If we compare with a student t test both averages, there is no significant proof at a 95 % level of confidence to distinguish them. Hence, we cannot at this time differentiate who has the best average run.
I would suggest to you, even if you are scared of the data and statistics to go look at the 3 Money vs time graphs. They do a pretty good job at showcasing visually just how much a run can go really good or really bad. The "Av + Std" line could be considered as: anything over is an exceptionally good run.
The limitations of the analysis are as follows. It's still unclear if the score in each run represents a bell curve. For Remo, it sorta does seem that yes up too ~75 000 $. The weird dips may be cause by the values that can be collect and might disappear if a better "slices" of money where chosen, for example 10000 instead of the 7000 currently used. I'm not sure if I understand statistics enough to figure out if yes or no. As mentioned earlier, this would have as effect that the standard deviation, or anything related, would have a great chance of not really representing the data. Another problem is the "average level". The way it's calculated is: ( level x c ) / n = number (rounded), where level is a numerical value given based on the level from 1 to 20, c is the number of times the furthest level has been reached in a run and n is the number of total runs. It can kinda figure out what is the average level, but I don't really trust. Maybe a simple graph of count vs level would be better choice. Lastly, a significant issue is the low number of data points for Patrick and the number of win/loss. 19 points may not really evaluate what is going on or be fully representative.
For now, it does seem that Patrick has a slight advantage over Remo in term of wins while their average score cannot be differentiated with certainty It's also quite obvious that it is really hard to keep a good track record from day to day. If this is the case, it may lead to an interesting and close battle of wins ans losses for both of them! As Patrick increases his number of runs, the statistics may change and reveal some new things.
It would be interesting to push the statistical analysis to find out if both Remo and Klepek have improved over their play time and by how much. It would also be interesting to see if this new challenge, competing against Patrick, has affected in any way the performance level of Chris. Maybe some new contenders will add themselves to the fight (Jeff)?
I hope you guys like this post! If you have any comments, suggestions, fixes, ideas, feel free to leave a comment! (Or if you find an error in the data set.) I will try to keep it updated regularly if you guys are interested.
Have a nice day :D