GLI is No Good

[Skippy9Baller]

Is it just me, or do you think this GLI (Game Loss Index) tie-breaker business is NOT the best way to go. For GLI, you ONLY look at your losses. Your number of wins is totally irrelevant. That means, for the matches that you lose, it doesn't matter if you win 7 games or 0 games...it's all the same. Let's give an example...

Wins - Losses

Player A
8 - 6
8 - 7
8 - 7
0 - 8
0 - 8
-----------------
24 - 36 (Total) GLI (7.2)

Player B
8 - 7
8 - 7
8 - 7
7 - 8
7 - 8
-----------------
38 - 37 (Total)GLI (7.4)

Let's take another look at your reasoning and this example. You say that the number of wins a player has when he losses a match is irrelevant. If he loses a match 0-8, it lowers the other players GLI pretty significantly, especially when there are only 5 players per bracket.

In your example above, it is obvious that when player A played player B, that Player A won 8-7. If player A had won 8-0, then it would look like this:

Player A
8 - 6
8 - 7
8 - 0
0 - 8
0 - 8
-----------------
24 - 28 (Total)GLI (5.6)

Player B
8 - 7
8 - 7
8 - 7
0 - 8
7 - 8
-----------------
31 - 37 (Total) GLI (7.4)

The GLI is greatly affected. The fact that player A did not win any games in matches 4 & 5, aslo means that his opponent had NO losses, which would greatly help their GLI. If a player loses 7-8 instead of 0-8, it will help him if those two players end up tied. If he loses 7-8, then he significantly added to the other players GLI.

I'm not saying that GLI is better than a pure winning % calculations (I'm still undecided on that), but it is more impactfull than what your original statement implies IMHO.

 

[sjm]

I'm happy with GLI. Any method leaves the door open for abuse of the system.

GLI rewards the blowout win more than the near miss loss, but I'm OK with that. I think this adds tension to close matches, which become all-or-nothing, and also helps motivate the best to go for the throat even with the lead in a match.

Despite it's undeniable limitations, I think GLI will help the cream rise to the top.

 

[Colin Colenso]

I don't see any advantage that GLI has over win percentage...am I missing something?

And win percentage include the data of our wins accumulated, certainly data that should be included. GLI basically treats a 0-8 loss the same as a 7-8 loss.

Also Win % is much more easily understood by the newly initiated.

 

[MasterRacker]

GLI would be a good tiebreaker if the tournament were a true round robin. In that case, everyone has played everyone else. GLI is essentially sorting players with the same number of match wins, by whose wins were the most dominant. It's a good tie breaker in a full round robin.

The weakness here is that this tourney is running in flights and players can be helped by landing in a "weak" flight or hurt by landing in a "strong" flight. In this situation total wins or winning percentage of total games played would probably be a better tie-breaker.

When it's all over, it will be interesting to re-work the score cards with different tie-breakers to see how the results might change.

 

[jsp]

This quote is from Celtic from another thread...

...As for the Owen robbery. I am with you on not liking the progression mechanism. After round 2 I am not sure how in the heck Williams did not advance to the third round.

Williams had the same 2-2 record of matches won but what should be the second deciding factor is the games won and Williams had 11 wins in his loosing matches. Gabe had 9 wins in his loosing matches, and Cory had 5 total wins in his loosing matches. So in total wins over the course of that round Williams had 27 wins, Gabe had 25 wins, and Cory had 21, and the guy with the least number of wins advanced? Plus Williams had 12 B&R's in that round, second only to Mika. He also gave up a total of 5 games combined in his losses, Gabe and Cory both gave up more.

I dont have any idea how Cory advanced, he was basically the lowest player of all 3 and in virtually ever fashion I can see Williams was the top player of those 3. And I am NOT a Williams fan at all and would much rather see Cory there, but not through some moronic system like we have with the IPT. Williams deserved to advance.

Does anyone think it was fair that Cory advanced over Charlie and Gabe? Does anyone still think GLI is a good idea? I knew it was a matter of time before we got a situation like this. I didn't expect that it would come as early as the 2nd day.

 

[MasterRacker]

I have to amend my post. For some reason (temporary braindeath maybe?) I was thinking GLI was only calculated off matches won. Looking at today's cards, it total losses. Under those circumstances, I'm definitely leaning to simple winning percentage.

 

[SlimShafty]

This quote is from Celtic from another thread...



Does anyone think it was fair that Cory advanced over Charlie and Gabe? Does anyone still think GLI is a good idea? I knew it was a matter of time before we got a situation like this. I didn't expect that it would come as early as the 2nd day.

There is no question in my mind that Charlie was the one (in that round) that should have advanced.

I too favor a winning %, but the thing I dislike the most is the carry over effect, it should be how well you perform for that round only, you should not be judged on how you played against past players, but the current players in that round, after all your deciding a tie breaker for that round.


Charlie Williams....match GLI 5.25 and a winning % of .563
Gabe Owen.........match GLI 6.00 and a winning % of .510
Corey Duel..........match GLI 6.75 and a winning % of .438


So if you judge performance in the round (even with GLI) then Williams was clearly the best player of the 3 (in that round)

No matter what you use, shouldn't it be based on your performance against the players in that round??

Poor Charlie