Slitherine

Thanks! Its our first macht in ELO ranking.

rbodleyscott wrote: ↑Mon Dec 04, 2023 5:50 pm It is fairly clear that the PBEM code for determining Elo rank changes is not working as intended. We are assisting the PBEM programmer to fix the issue(s).

(It was tested in the open beta test, but if it wasn't adjusting the scores correctly, nobody pointed it out)

Once the code is working as intended, we will probably need to reset all of the scores to 1000. (Because correcting all the errors might be prone to error).

This is probably slightly late but I have noticed the same issue.
I completed my first two ranked games in FoGII between 17 and 27 December.
It was a mirror game so both games started at the same ELO point for me, Triarii, and my opponent.
He commenced at 1015 and I at 1000.

Both games were normal victory conditions and wins for me, Triarii.
The first game finished on 26th December result 44%-19% in my favour with me playing Greek 1179-681BC
The mirror finished on 27th December result was 54%-26% in my favour with me playing Cypriot 800-681BC

The ELO scores on 28th December show Triarii at 962 - down from 1000 and Deano2 at 1044 up from 1015.
This would not be zero sum for our game (-38) vs (+29) but I understand my opponent also had another game going on (that was conceded to him) between our start and finish. My opponent told me that the game conceded to him did not seem to have increased his ELO score.

When I was involved in the Beta Test for Medieval the ELO scoring and direction of movement seemed broadly as expected from the formula shared.

Looking at the formula provided in the Beta testing the result in my case seems to be equal and opposite to the value expected. Could the issue be as simple as incorrect +/- assignment in the code?

Triarii wrote: ↑Thu Dec 28, 2023 12:54 am

rbodleyscott wrote: ↑Mon Dec 04, 2023 5:50 pm It is fairly clear that the PBEM code for determining Elo rank changes is not working as intended. We are assisting the PBEM programmer to fix the issue(s).

(It was tested in the open beta test, but if it wasn't adjusting the scores correctly, nobody pointed it out)

Once the code is working as intended, we will probably need to reset all of the scores to 1000. (Because correcting all the errors might be prone to error).

This is probably slightly late but I have noticed the same issue.
I completed my first two ranked games in FoGII between 17 and 27 December.
It was a mirror game so both games started at the same ELO point for me, Triarii, and my opponent.
He commenced at 1015 and I at 1000.

Both games were normal victory conditions and wins for me, Triarii.
The first game finished on 26th December result 44%-19% in my favour with me playing Greek 1179-681BC
The mirror finished on 27th December result was 54%-26% in my favour with me playing Cypriot 800-681BC

The ELO scores on 28th December show Triarii at 962 - down from 1000 and Deano2 at 1044 up from 1015.
This would not be zero sum for our game (-38) vs (+29) but I understand my opponent also had another game going on (that was conceded to him) between our start and finish. My opponent told me that the game conceded to him did not seem to have increased his ELO score.

When I was involved in the Beta Test for Medieval the ELO scoring and direction of movement seemed broadly as expected from the formula shared.

Looking at the formula provided in the Beta testing the result in my case seems to be equal and opposite to the value expected.

The Medieval versions seem to work correctly. (At least for games that have not been kyboshed by the "can't upload" bug)

Could the issue be as simple as incorrect +/- assignment in the code?

No, but the fix may be such a hack, to avoid breaking anything else.

FWIW, the ELO in Ancients has been working correctly for me so far. I won a game and went up and lost and went back down to starting rating.

Karvon

Karvon wrote: ↑Sun Dec 31, 2023 1:20 am FWIW, the ELO in Ancients has been working correctly for me so far. I won a game and went up and lost and went back down to starting rating.

Karvon

What sort of games were these? Individual custom battle? Mirror custom battle? etc.

When did you finish them?

rbodleyscott wrote: ↑Mon Jan 01, 2024 9:04 am

Karvon wrote: ↑Sun Dec 31, 2023 1:20 am FWIW, the ELO in Ancients has been working correctly for me so far. I won a game and went up and lost and went back down to starting rating.

Karvon

What sort of games were these? Individual custom battle? Mirror custom battle? etc.

When did you finish them?

They were both custom mirror battles; each being one of two. The remaining ones should both be finished shortly. One ended the 28th, the other the 30th.

Karvon

See: https://www.slitherine.com/forum/viewtopic.php?t=114137

Please report any further Ancients Elo scoring anomalies there.

FYI, I finished another 2 mirror match custom battles and won both games. My ELO was adjusted correctly.

Karvon

The FOG2 Ancients Elo ranking have now all been reset to 1000.

Nothing will show in the Elo Rankings page until some new results come in. In the meantime it will display "The leaderboard is on maintenance, please wait."

This looks to be broken.
I have a ranked game that was aborted on turn 1 and it has credited me with a win.
No ranked games completed beyond that stage.

When we tested this for Medieval it didn't credit wins or losses until games had gone 3 turns.

The issue with as it is now is that TDC and HOML both allow for map rerolls so as it stands the Slitherine system is incompatible with this mass of MP games.

tyronec wrote: ↑Wed Jan 10, 2024 8:19 am This looks to be broken.
I have a ranked game that was aborted on turn 1 and it has credited me with a win.
No ranked games completed beyond that stage.

When we tested this for Medieval it didn't credit wins or losses until games had gone 3 turns.

The issue with as it is now is that TDC and HOML both allow for map rerolls so as it stands the Slitherine system is incompatible with this mass of MP games.

How exactly was it aborted on Turn 1?

When I know that I will bring this to the attention of those concerned.

As far as I know the only way to abort a game on T1 is to complete your deployment and then concede defeat, have asked my opponent what he did.

rbodleyscott wrote: ↑Wed Jan 10, 2024 8:35 am When we tested this for Medieval it didn't credit wins or losses until games had gone 3 turns.

What exactly do you mean by 3 turns? 3 turns each by both players, or 2 by player A and 1 by playerB?

What exactly do you mean by 3 turns? 3 turns each by both players, or 2 by player A and 1 by playerB?

Three turns by both players.

Medieval looks to be working fine for aborted games.

I auto selected, auto deployed and then surrendered. ELO dropped from 1000 to 985.

markleslie wrote: ↑Wed Jan 10, 2024 8:16 pm I auto selected, auto deployed and then surrendered. ELO dropped from 1000 to 985.

Ancients has a different PBEM codebase from Medieval. I expect this is an oversight in the Ancient coding. I have notified those concerned.

Question re 'draws'. I finished a match down on points but with no outright victor. I know the match screen says we lost the field, but is there no such thing as a draw regard ELO? Unless scores are exact? I dropped by 15 ELO which I think is the same result as if opponent achieved a 25+ margin or routed 60%.

Kroche wrote: ↑Mon Jan 22, 2024 6:38 pm Question re 'draws'. I finished a match down on points but with no outright victor. I know the match screen says we lost the field, but is there no such thing as a draw regard ELO? Unless scores are exact? I dropped by 15 ELO which I think is the same result as if opponent achieved a 25+ margin or routed 60%.

The game engine does not recognise the existence of draws. Hence the Elo system doesn't either.

Also the Elo system does not take into account degrees of victory. You get the same ranking change for a marginal victory as for a decisive victory.

rbodleyscott wrote: ↑Tue Jan 23, 2024 6:58 am

Kroche wrote: ↑Mon Jan 22, 2024 6:38 pm Question re 'draws'. I finished a match down on points but with no outright victor. I know the match screen says we lost the field, but is there no such thing as a draw regard ELO? Unless scores are exact? I dropped by 15 ELO which I think is the same result as if opponent achieved a 25+ margin or routed 60%.

The game engine does not recognise the existence of draws. Hence the Elo system doesn't either.

Also the Elo system does not take into account degrees of victory. You get the same ranking change for a marginal victory as for a decisive victory.

Does this mean that games we consider to be draws, neither player breaking, are counted as a loss for both players or for the lower scoring player in terms of Slith Elo calculation?

Karvon

Karvon wrote: ↑Tue Jan 23, 2024 7:28 am

rbodleyscott wrote: ↑Tue Jan 23, 2024 6:58 am

Kroche wrote: ↑Mon Jan 22, 2024 6:38 pm Question re 'draws'. I finished a match down on points but with no outright victor. I know the match screen says we lost the field, but is there no such thing as a draw regard ELO? Unless scores are exact? I dropped by 15 ELO which I think is the same result as if opponent achieved a 25+ margin or routed 60%.

The game engine does not recognise the existence of draws. Hence the Elo system doesn't either.

Also the Elo system does not take into account degrees of victory. You get the same ranking change for a marginal victory as for a decisive victory.

Does this mean that games we consider to be draws, neither player breaking, are counted as a loss for both players or for the lower scoring player in terms of Slith Elo calculation?

Karvon

Think it counts as a win for the higher scoring player.