EPT...wait what?
Yes. A spreadsheet that predicts the future. In probabilities I should add.This is nothing revolutionary. Actually, far from it. The predictions are almost always like one would have felt about a matchup anyway (the favourite has a higher chance to win, right!?)
WTF why are you posting this?
Well, it is one thing to have a feeling about a match that's about to take place, but another to have the win chance probabilities visualized in actual numbers. Actually, the main reason I'm posting this is the fact that a lot of people might enjoy just toying around with a spreadsheet like I am.
So...it's a toy?
Yes, it is a toy. But it's not random. You give input, you get output. It's simple.
How are the match outcomes predicted?
Ofc by statistics only. ELO / Glicko Score of the players and Matchup Win Percentage are somewhat combined and calculated into a final win probability number. For basic bo1 matches, but more important (and more accurate) for bo3 and bo5 matches, too. Even the probabilities for a specific score (like 2-1 or 0-3) are calculated for the lulz.
You are wasting my time, but since I spend energy on clicking this pile of shit thread tell me how I calculate the probable outcome of one fucking match!
Sure
Let's make an example. Let's randomly pick White-Ra vs Sheth.What you need to do first is get the players Glicko / ELO score and Matchup stat. I prefer sc2charts.com to do so, you could also just use TLPD I guess, I just like the Glicko system more and feel the database is bigger. So let's look up Sheth:
![[image loading]](http://img855.imageshack.us/img855/6163/sheth.jpg)
Get his Glicko / ELO and his MU%, this case ZvP.
Do the same for White-Ra and type both in the sheet:
![[image loading]](http://img689.imageshack.us/img689/4127/whvssheingabe.jpg)
Watch as win probabilities are appearing:
![[image loading]](http://img689.imageshack.us/img689/8273/whvssh.jpg)
The sheet gives you the probabilities of both players winning a bo1, bo3 and bo5 against each other. The top lines show you the probabilities if only the ELO / Glicko rank was considered, just for comparison. Important are the bottom lines where all stats are combined for calculation. As you can see even the probabilities for certain scores like 1-2 are calculated.
Was Polt beating HerO so bad a surprise?
NO:
![[image loading]](http://img684.imageshack.us/img684/1521/polthero.jpg)
not at all. But you all knew that anyway. But this is the proof my spreadsheet can actually predict the future
Conclusion
As I said this whole thing is more a toy than anything, but I am sure there are people who like toying around with numbers and they might just appreciate this. I tested this on like 20 matches that have already been played and it was correct in its "prediction" in 19 out of 20. Sure you might say blabla obvious favourite is favourite but consider that most of the time people just compare rank / ELO out of laziness and don't regard matchup percentage. The example picture of the Polt vs. HerO prediction shows it best: If you just go by rank stats are 50.34% to 49.66% in Polt's favor. So basically even. But if MU% is added into the equation we can see: In a bo5 Polt is statistically the favourite by 86.53% vs HerO!!!! Wow.
I wanna toy around with this, too!
No problem, you can find it here. Plz only edit the player + ELO + MU% section, not the formulas behind it
I trust you. If you wanna use different formulas go make your own.I hope some people just have fun with this, I only wanna share the fun. If you don't have fun with this, plz go away from this thread :p
Edit (Read!):
ok, sooner or later some evildoer will screw the formulas up on purpose, so in case that happens, here you can find the same one which can not be edited online, but you can save it or make a copy etc. and use it in private.
From watching the spreadsheet: Guys, it is not enough to just type in the players' names, even if my spreadsheet is close to being a god, it isn't. You must type in Glicko / ELO and MU% of the players manually. And if you do so, plz also update the names, you yourself know ofc whom you compare, but the rest of the watching people isn't
HF!Adjustment: I changed the formulas to give a little more impact on ELO than before as shackes suggested, so don't be confused if you can't reproduce the exact numbers from the examples. Tendencies stayed ofc the same.
