I made a sheet with player statistics. Warning, it's fairly large and may be a bit laggy: https://docs.google.com/spreadsheets/d/1NsM3cCofvnlpUeZcrluS8q7fUMHAjFb0-susGnI9mM4/edit?usp=sharing

Here are the stats I gathered for everyone and used to rank with:

• Run Count • Full Game Run Count • Unique Game Count • Unique Leaderboard Count • Unique Full Game Leaderboard Count • Total Time • Total Full Game Time • Run Activity • Full Game Run Activity • Time Activity • Full Game Time Activity • Combined Activity • Full Game Combined Activity • World Records Held • Full Game World Record Held • Top 3 Records Held • Full Game Top 3 Records Held • Sum of Weighted Records • Sum of Weighted Full Game Records • Best Weighted Record

To the right of the sheet are rankings, depending on how you value things. • MasterOfOne ranks a player's best rank, from any one stat • AverageRanking does exactly that • JackOfAllTrades ranks a player's worst rank, from any one stat • MasterOfMasters compares everyone's best rank, across all ranks • AverageAverage works similar to the above and below • JackOfJacks compares everyone's worst rank, across all ranks • And Weakest Link is what the player's lowest rank is. Most peoples' are Best Weighted Record.

Notes/Comments:

• Separate Full Game was used since ILs can inflate positions more than they should. That, and people generally like full game runs more than individual levels, except for best of compilations, which I enjoy (doing).
• Combined Activity is the same as Activity on the site for games. It's Run Activity + Time Activity.
• Weighted records are [NumberOfPlayers]^0.5/[Position]. Ocarina of Time Any% has 753 records, so the WR holder would have 27.44, 2nd place 13.72, 3rd 9.15, etc. It's slightly skewed as it includes emulator runs regardless of settings. I may end up changing how it's calculated.
• Proper separation of Full Game and Level is dependent on the games' moderators. Halo Co-op is an example where level categories are classified as full game. I can't do much about it unless I manually adjust them when calculating.
• It does not know when a leaderboard is sorted descending, such as Metroid Prime's Frigate Escape. The API currently does not supply that information. Even if it did, I'm not entirely sure how to go about implementing it.
• -- Found a work-around. Using Excel to help with this for now. Shaved off 11% of the total time, most of which were from fake runs. Next time I make stats it'll exclude those runs' times.

Separate post for reading sake.

For individuals with a #1 position (followed by #2):

• Run Count ------------------------------------- Nordicboa @ 543 ------------------------ Racingmonster @ 542 • Full Game Run Count ----------------------- Jumpyluff @ 302 ------------------------ Cyberdemon531 @ 187 • Unique Game Count ------------------------ Jumpyluff @ 159 ------------------------- Dugongue @ 141 • Unique Leaderboard Count ---------------- Racingmonster @ 480 ----------------- Jumpyluff @ 355 • Unique Full Game Leaderboard Count - Jumpyluff @ 236 ------------------------ Dugongue @ 185 • Total Time ------------------------------------- Baffan @ 1476:09:38 ------------------- 360Chrism @ 645:40:35 • Total Full Game Time ----------------------- Baffan @ 1476:09:38 ------------------- 360Chrism @ 645:40:35 • Run Activity ------------------------------------ portal @ 118.07 -------------------------- Racingmonster @ 101.81 • Full Game Run Activity --------------------- portal @ 55.07 ---------------------------- Twyn_o @ 42.20 • Time Activity ----------------------------------- Baffan @ 84.15 -------------------------- portal @ 58.70 • Full Game Time Activity -------------------- Baffan @ 84.15 -------------------------- portal @ 57.29 • Combined Activity ---------------------------- portal @ 176.77 -------------------------- Racingmonster @ 117.53 • Full Game Combined Activity ------------- portal @ 112.36 -------------------------- Baffan @ 85.69 • World Records Held ------------------------- Racingmonster @ 224 ----------------- Fe @ 183 • Full Game World Record Held ------------ werster @ 59 ----------------------------- Dugongue @ 56 • Top 3 Records Held ------------------------- Racingmonster @ 387 ----------------- Dugongue @ 242 • Full Game Top 3 Records Held ---------- Dugongue @ 121 ------------------------ Jumpyluff @ 113 • Sum of Weighted Records ----------------- Racingmonster @ 389.95 ------------- TheWinslinator @ 297.96 • Sum of Weighted Full Game Records -- Jumpyluff @ 172.40 --------------------- Dugongue @ 135.45 • Best Weighted Record ---------------------- skater82297 @ 27.44 (OoT Any%) - bounceyboy @ 24.00 (SMS Any%)

The best Jack of All Trades (comparing everyone's worst rank) are:

1. Mhmd_FVC (#157)
2. TheFuncannon (#304)
3. Vallu (#324)
4. Racingmonster (#332)
5. Jumpyluff (#398)
6. Kobral (#469)
7. Saiyanz (#474) 8) Rubberduckyassassin3 (#481)
8. pikachulikespie (#544)
9. ThePackle (#575)

And for average:

1. Jumpyluff
2. Racingmonster
3. Cyberdemon531
4. portal
5. Rubberduckyassassin3
6. Mhmd_FVC
7. Redlof 8) 360Chrism
8. TheFuncannon
9. crrool

Notable mentions: • RPGodfather - The gap between his worst and 2nd worst rank is the largest (2988 vs 281). He could the 2nd best Jack of All Trades if he can improve his Best Rank, though to get there he'd need a WR run that beats about 27 other players. • Cyberdemon531 - All his other ranks would put him above Mhmd_FVC in Jack of All Trades, but like with the above, he'd need a better Best Rank. He'd need a WR run that beats at least 43 other players to get there, ouch. • Removing Activity from the calculations doesn't make a huge change in positions, but portal takes a huge fall since his top stats are all for activity, while Dugongue gets the largest boost since he's been gone for a little while. Kobral would be the #1 Jack if it wasn't for activity. • werster, Pull, Jcool114, and mike89 would be in the top 10 Jack of All Trades, if activity was excluded. • Opting to use standard deviation over ranking results in mostly the same. Best positions remain identical; Demolition14 and Nord enter the top 10 jack of all trades; and Dugongue, werster, PrinceRallis, and Deln enter the top 10 average.

Bonus -- The best median users:

1. Jumpyluff
2. Racingmonster
3. Cyberdemon531
4. Dugongue
5. portal
6. crrool
7. werster 8) PrinceRallis
8. Deln
9. sun100

Could you get "Unique Run Count"? I'm just assuming that you currently count obsolete times too.

[quote=andypanther]Could you get "Unique Run Count"? I'm just assuming that you currently count obsolete times too.[/quote]

Unique Leaderboard Count/CategoryCountAll is exactly that.

Thanks for explaining :)

Added a means to exclude most absurdly long runs from being added into Total Time. Runs with times higher than the upper limit of the interquartile range of the leaderboard aren't included in time related calculations. Runs that are too extreme are removed entirely. ZooKetra and Stingrayy should be ashamed of themselves D:<

It's not fool proof, mods who want to skew their stats up can just make enough fake runs, though if I catch anyone doing that intentionally I'll permanently remove them from the stats. If they try to use someone else, I'll notice by submitter/verifier.

There are a few run that really should be included, but aren't, so I'll adjusted it until it looks right.

currently adding in hundreds of my Doom IL runs so that I can be #1.

:)

I'm trying to come up with a better formula for 'Weighted Records' that doesn't harshly punish those with really close 2nd place or lower. Like for Super Mario Sunshine's Any%, nindiddeh shouldn't have his rank cut in half for being very closely behind an extremely competitive leaderboard. Standard deviation should be able to help remedy this, but I'm not quite sure how to use it.

The same goes with the absurdly high times as I've mentioned before. It's difficult to create a catch-all formula to remove the let's play runs and not include the runs that aren't. Removing times in the upper limit of the interquartile range removes this run when it shouldn't, though it's using all runs. How do I prevent 2nd place in this game from being added, since it's practically a let's play by comparison? Distance from 1st place gives really weird results.