Recommend
4 
 Thumb up
 Hide
24 Posts

Terra Mystica» Forums » General

Subject: Some Tournament Stats rss

Your Tags: Add tags
Popular Tags: [View All]
Silly Words
Australia
flag msg tools
mbmb
Just made some stats based currently in div 1 and div2. As usual it highlights the brilliance of the game designers to achieve a game, where despite the complexity and the variations, that ten of the fourteen factions have practically identical chances. And of the other four, Alchemists, Giants and Auren are all viable, only Fakirs got screwed.

Here is the data.


Row Labels SumGame W-Score W-Pos Ave AvePos RankGam RankScr ScrDiff RankPos RankDiff
A Alchemists 24 3135,00 59,99 130,6 2,50 10 12 -2 13 -3
A Auren 7 913,00 16 130,4 2,29 13 13 0 11 2
C Chaos magicians 61 8404,97 124,03 137,8 2,03 3 7 -4 8 -5
C Cultists 19 2639,01 36,02 138,9 1,90 11 6 5 6 5
D Darklings 120 16830,12 221,1 140,3 1,84 1 2 -1 5 -4
D Dwarves 27 3768,01 46,99 139,6 1,74 9 4 5 1 8
E Engineers 57 8040,98 102,96 141,1 1,81 4 1 3 3 1
F Fakirs 2 228,00 8 114,0 4,00 14 14 0 14 0
G Giants 10 1313,01 23,01 131,3 2,30 12 11 1 12 0
H Halflings 42 5840,00 77,01 139,0 1,83 7 5 2 4 3
M Mermaids 50 6979,02 87,09 139,6 1,74 6 3 3 2 4
N Nomads 78 10649,98 151,1 136,5 1,94 2 10 -8 7 -5
S Swarmlings 34 4654,00 72,99 136,9 2,15 8 9 -1 10 -2
W Witches 57 7832,08 116,05 137,4 2,04 4 8 -4 9 -5
(blank)
Grand Total 588 81227,18 1142,33 138,1



Here are some points.

1) Unsurprisingly, Darklings are the number one pick. And with an average score rank of 2, this doesn't seem like a bad plan. However, if you look at the average position stats, they are only 5th. So maybe they are being over-picked a bit.

2) As stated in the intro. Ten factions are practically identical. The difference from the first placed score of Engineers 141,1 to the 10th of Nomads 136,5 is just 4,6 or around 3%.

3) Auren, Giants and Alchemists, form a group of the not-so-good factions, averaging around 130, whereas the top 10 average around 139.

4) The sample size is small, but Fakirs seem to suck mightily. No prize for re-stating that.

5) I did a small stat, that compares the scores (scrDiff) and the rank (rankDiff) to the number of times that they get played. A negative number in this column indicates that the faction was picked more than it's results would imply and a positive would indicate that the faction was picked not enough. On first viewing, it would appear that the dwarves and the cultists are the most underpicked factions. Although they have the highest position rank, the dwarves are only picked 9th overall. That could mean one of two things. a) they are truly underpicked or b) you don't pick them unless the board is right, and if you pick them at the right time, they score well.

6) Just for explanation purposes. The w- fields stand for weighted i.e. where the number of games e.g. 8 is multipled by the average of the player.

In addition to these points from the data above.

7) Of the 21 players, there are only four rated under 1300 and there are eight 1400+ players amongst them. If you want to get to this sort of level, then you have to be averaging 138,1 in the tournament format. (I, personally, am way under that, at 126 or so).

8) The highest averages are unsurprisingly the top two ranked players on the site and the only ones with ratings above 1500, namely Xevoc at 148.6 and mikaejlt at 150.4. The next best average is the site creator Juho at 143.4 followed by a bunch of other players around that number. And although Xevoc is *only* second highest, his score is all the more remarkable, given that he has played ALL of his matches in division 1, against the toughest opponents, who compete the hardest for cult and network endpoints.

cheers

SW.

p.s. If I get another procrastination attack I might add in the div3 players.

10 
 Thumb up
0.25
 tip
 Hide
  • [+] Dice rolls
Robin Zigmond
United Kingdom
Durham
flag msg tools
mbmbmbmbmb
Thanks for the analysis, SW

SillyWords wrote:
p.s. If I get another procrastination attack I might add in the div3 players.


I wouldn't do this, as I am currently polluting those stats, and I don't feel like the weakest in div3, by any means (even though I've had a terrible season and am going to go down, mind you, I wasn't that far off going up to div 2 after last season! surprise). Stick to the top 2 divisions, and you've got a decent chance of most or all of the games actually involving 4 players who know what they're doing
2 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Robert
Germany
Bocholt
flag msg tools
badge
I paid 100 Geek Gold so that you can read this! :-)
Avatar
mbmbmbmbmb
As somebody who's done some similar analysis before (*click*), I have to point out that the average score as a measure of success is inferior to the margin (= delta between the score in a game and the average of all scores in that game). If a faction does well in low-score games (few of BON6/7/9/10, no late TOWN>>5 scoring etc.), and thus is picked more often in such games, then its average will look worse than what is the faction's actual performance. Better to win with 135 VP than to come in last with 138.
5 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Space Trucker
Germany
flag msg tools
badge
Avatar
mbmbmbmbmb
Thanks for the stats!

One thing that I maybe get wrong: In a four player game the average position should be 2.50. In your data all factions except fakirs have an average position that is 2.5 or better. Shouldn't there be a good number of factions in the >2.50 area (two fakir games can hardly balance 120 darkling games)?
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Matthias Reitberger
Germany
Nürnberg
flag msg tools
Avatar
mbmbmbmbmb
DocCool wrote:
As somebody who's done some similar analysis before (*click*), I have to point out that the average score as a measure of success is inferior to the margin (= delta between the score in a game and the average of all scores in that game). If a faction does well in low-score games (few of BON6/7/9/10, no late TOWN>>5 scoring etc.), and thus is picked more often in such games, then its average will look worse than what is the faction's actual performance. Better to win with 135 VP than to come in last with 138.


I don't agree, it's also better to win a game with 135 VP with a margin of +1, than to come in second to last with a margin of +10.
You assume that there are factions that are particularly good in low scoring games, there is no proof for that.
One could as well assume that there are disruptive factions that will render an opponent dead last while often not being able to win against the others.
4 
 Thumb up
0.05
 tip
 Hide
  • [+] Dice rolls
Chris Sandman
Germany
flag msg tools
I compared various measures in this thread
Why Darklings are not OP and Giants, Alchemists and especially Fakirs still seem a bit weaker than the rest
Maybe i will update the thread with new data after this season is finished.
I would rank the averages of the following measures in this order from "worst" to "best" approximation of relative strength subject to large sample size which is ubfortunately not given for some factionswhistle

Score < margin < position < league points
2 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Space Trucker
Germany
flag msg tools
badge
Avatar
mbmbmbmbmb
You're right, the strength in tournament play is probably best measured with the average league points per game.

In normal rated games being 2nd and 3rd in two games is considered as good as being 1st and 4th in two other games (both results in 3 "1v1 games" lost and 3 "1v1 games" won).
In tournament play the points are non-linear - 2nd and 3rd is 4 points, 1st and 4th is 6 points, while the average position is the same.

So compared to normal ranked games in tournament games the rank of volatile factions might increase a bit.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Robert
Germany
Bocholt
flag msg tools
badge
I paid 100 Geek Gold so that you can read this! :-)
Avatar
mbmbmbmbmb
1869 wrote:
DocCool wrote:
As somebody who's done some similar analysis before (*click*), I have to point out that the average score as a measure of success is inferior to the margin (= delta between the score in a game and the average of all scores in that game). If a faction does well in low-score games (few of BON6/7/9/10, no late TOWN>>5 scoring etc.), and thus is picked more often in such games, then its average will look worse than what is the faction's actual performance. Better to win with 135 VP than to come in last with 138.


I don't agree, it's also better to win a game with 135 VP with a margin of +1, than to come in second to last with a margin of +10.
You don't agree that it's better to win with 135VP than to com in last with 138? That's surprising.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Silly Words
Australia
flag msg tools
mbmb
DocCool wrote:
As somebody who's done some similar analysis before (*click*), I have to point out that the average score as a measure of success is inferior to the margin (= delta between the score in a game and the average of all scores in that game). If a faction does well in low-score games (few of BON6/7/9/10, no late TOWN>>5 scoring etc.), and thus is picked more often in such games, then its average will look worse than what is the faction's actual performance. Better to win with 135 VP than to come in last with 138.


Yeah I agree.

Margin is a better comparison. But I grabbed the numbers from the tournament site as a quick and dirty. I couldn't be arsed collecting all the game data. However, if you have a large enough sample set (and 50+ games is almost large enough) then you have a good indication of what score you need to be aiming at to be viable in div1 or div2. Rank is a better indicator for comparing the factions than score. And tournament points is only useful for optimizing your tournament points.

Anyhow, the point i had was to see what sort of numbers you need to get to to start looking like a div1/div2 player. High 130s seems to be the answer.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Robert
Germany
Bocholt
flag msg tools
badge
I paid 100 Geek Gold so that you can read this! :-)
Avatar
mbmbmbmbmb
I agree with this
DieKrake wrote:
Score < margin < position < league
but am doubtful about this
DieKrake wrote:
position < league points
outside of the tournament context.

Of course you can argue that the data stems from tournament games and people are thus taking it into consideration. But do players even in the top divisions have the predictive power to identify two factions as offering the same positional average, but one being more on the all-or-nothing side (which is what the league point system favors). It would be interesting to learn whether Petri, Juho, Daniel and the top bunch actually pick faction X over faction Y because they think X is an all-or-nothing faction (e.g. 60:40 to be #1 or #4 = avg. position 2.2 = avg. league points 3.6) while faction Y is a don't-come-in-last faction (20:40:40 to be #1, #2 resp. #3 = avg. position 2.2 = avg. league points 2.8).
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Space Trucker
Germany
flag msg tools
badge
Avatar
mbmbmbmbmb
DocCool wrote:
I agree with this
DieKrake wrote:
Score < margin < position < league
but am doubtful about this
DieKrake wrote:
position < league points
outside of the tournament context.

In ranked games you basicly get point for [3,1,-1,-3] wins, the 'point distribution' is linear to the position. So here it's 'position = points' instead.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Matthias Reitberger
Germany
Nürnberg
flag msg tools
Avatar
mbmbmbmbmb
I play to win, not just in Tournament. Even if I pick factions outside of tournament that I often struggle with the aim is not to achieve a second place with Auren or Fakirs.
All or nothing is rather dependent on setup than specific for a certain faction in all setups, but especially in non tournament setups I would pick the most promising faction even with the risk of total failure than a safe one.
2 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Petri Savola
Finland
Espoo
Unspecified
flag msg tools
designer
mbmbmbmbmb
Picking Engineers from seat 3 is a typical high risk move. If player 4 picks green (or maybe red) it'll be really difficult to pull off a good game, but else Engineers are maybe the favorite to win the game in most setups.

Same holds for starting positions also. For example you can try to pick Chaos Magicians and hope that Nomads will not take the critical yellow tiles. If they do, you'll struggle, but otherwise you're in great shape. That's what I did during this season and the gamble badly failed.
2 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Chris Sandman
Germany
flag msg tools
DocCool wrote:
I agree with this
DieKrake wrote:
Score < margin < position < league
but am doubtful about this
DieKrake wrote:
position < league points
outside of the tournament context.

Of course you can argue that the data stems from tournament games and people are thus taking it into consideration. But do players even in the top divisions have the predictive power to identify two factions as offering the same positional average, but one being more on the all-or-nothing side (which is what the league point system favors). It would be interesting to learn whether Petri, Juho, Daniel and the top bunch actually pick faction X over faction Y because they think X is an all-or-nothing faction (e.g. 60:40 to be #1 or #4 = avg. position 2.2 = avg. league points 3.6) while faction Y is a don't-come-in-last faction (20:40:40 to be #1, #2 resp. #3 = avg. position 2.2 = avg. league points 2.8).



If you play competitively you play to win and pick the faction that you are most confident with to be able to win that particular game.
By disproportionately rewarding higher positions with "additional" points you pay attention to that.
Pure winning percentage takes differing from a linear distribution to the extreme but goes way too far (eg. {1st,4th,4th,4ht,4th,4th} is worse than (2nd,2nd,2nd,2nd,2nd,2nd}).
So tournament points are used for a reason to value players relative strengths as well as they fit best (or at least in a similar (disproportionate) way) to rank a factions strengths.

EDIT: Have not read what Matthias said
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Gambia
Brest
msg tools
Avatar
Petri wrote:
Picking Engineers from seat 3 is a typical high risk move. If player 4 picks green (or maybe red) it'll be really difficult to pull off a good game, but else Engineers are maybe the favorite to win the game in most setups.


Does it implies that if the setup is good for engineers and you are 3rd pick, if green and grey are open, your best option is often to pick green ?
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Matthias Reitberger
Germany
Nürnberg
flag msg tools
Avatar
mbmbmbmbmb
If there is a blue faction, especially Mermaids, I wouldn't even consider it.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Gambia
Brest
msg tools
Avatar
1869 wrote:
If there is a blue faction, especially Mermaids, I wouldn't even consider it.


I'm not sure. Witches and mermaids can sometimes cohabit if there is Nomads for example (Mermaids might stay in the West to take all the leech from Nomads, and Witches can expand in the East).

But if you don't consider green, would you pick grey then? (As last pick is less likely to pick green because of the presence of blue).

Or would you pick another colour letting grey open for last pick?
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Petri Savola
Finland
Espoo
Unspecified
flag msg tools
designer
mbmbmbmbmb
I think black-blue-yellow-gray is a fair setup where all players have roughly equal chance to win or lose. However, if player 1 takes black, player 2 takes yellow and 3rd player then takes green, last player is in horrible position, because all options are very bad. If he picks brown or blue, he's between two colors in the color wheel, which is often rough. If he picks gray, black will have a huge advantage and both 3rd and 4th player are in bad situation. I think red should be taken in that situation, but if setup doesn't happen to be suitable for Giants, it's always risky to pick Chaos Magicians against the Nomads. And playing Giants with unsuitable setup is not too good either.
4 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Gambia
Brest
msg tools
Avatar
Petri wrote:
I think black-blue-yellow-gray is a fair setup where all players have roughly equal chance to win or lose. However, if player 1 takes black, player 2 takes yellow and 3rd player then takes green, last player is in horrible position, because all options are very bad. If he picks brown or blue, he's between two colors in the color wheel, which is often rough. If he picks gray, black will have a huge advantage and both 3rd and 4th player are in bad situation. I think red should be taken in that situation, but if setup doesn't happen to be suitable for Giants, it's always risky to pick Chaos Magicians against the Nomads. And playing Giants with unsuitable setup is not too good either.


Thanks for the analysis.

If expansion maps and scoring tiles were allowed, Giants would become much easier to pick in this situation. (Again, that shows that expansion did overall a great job at balancing factions).
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Space Trucker
Germany
flag msg tools
badge
Avatar
mbmbmbmbmb
Petri wrote:
However, if player 1 takes black, player 2 takes yellow and 3rd player then takes green, last player is in horrible position, because all options are very bad. If he picks brown or blue, he's between two colors in the color wheel, which is often rough. If he picks gray, black will have a huge advantage and both 3rd and 4th player are in bad situation. I think red should be taken in that situation, but if setup doesn't happen to be suitable for Giants, it's always risky to pick Chaos Magicians against the Nomads. And playing Giants with unsuitable setup is not too good either.
I'm a bit surprised about that, but think I have to disagree here. Picking Chaosmages against Nomads may be tricky if he places his starting dwellings in a nasty way, but picking CMs against Nomads and green sounds very risky - that may become a game you've lost after starting dwellings have been placed. I (sadly) experienced (twice, there hopefully won't be a third time) that green tends to fill all the gaps Nomads leave in the centre, so you might end contained in one corner of the map escaping only at high cost if at all (as it also was the case in your league game that went wrong) - or you are left with an empty north which is bad, too.
I think it's way safer to take e.g. Swarmlings and focus on the western side (both green and black are not very likely to build much there and with three 7/5pw faction in game you are likely to get a good start with power actions, also there is no power heavy facion in game).
If we consider it risky for black to place two dwellings in the east in this setup, there may be enough space for one brown starting position in the east, so it may also be safer to pick brown (especially halflings). Or are there some other aspects I don't see?
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Petri Savola
Finland
Espoo
Unspecified
flag msg tools
designer
mbmbmbmbmb
SpaceTrucker wrote:
Petri wrote:
However, if player 1 takes black, player 2 takes yellow and 3rd player then takes green, last player is in horrible position, because all options are very bad. If he picks brown or blue, he's between two colors in the color wheel, which is often rough. If he picks gray, black will have a huge advantage and both 3rd and 4th player are in bad situation. I think red should be taken in that situation, but if setup doesn't happen to be suitable for Giants, it's always risky to pick Chaos Magicians against the Nomads. And playing Giants with unsuitable setup is not too good either.
I'm a bit surprised about that, but think I have to disagree here. Picking Chaosmages against Nomads may be tricky if he places his starting dwellings in a nasty way, but picking CMs against Nomads and green sounds very risky - that may become a game you've lost after starting dwellings have been placed. I (sadly) experienced (twice, there hopefully won't be a third time) that green tends to fill all the gaps Nomads leave in the centre, so you might end contained in one corner of the map escaping only at high cost if at all (as it also was the case in your league game that went wrong) - or you are left with an empty north which is bad, too.
I think it's way safer to take e.g. Swarmlings and focus on the western side (both green and black are not very likely to build much there and with three 7/5pw faction in game you are likely to get a good start with power actions, also there is no power heavy facion in game).
If we consider it risky for black to place two dwellings in the east in this setup, there may be enough space for one brown starting position in the east, so it may also be safer to pick brown (especially halflings). Or are there some other aspects I don't see?

I agree with you, I was trying to say that you should pick red based on the color wheel, but one of the red factions is very fragile (Chaos Magicians) and one of them depends quite much on setup (Giants), so it's not easy to pick either of them. Chaos Magicians can die in the setup, but so can Swarmlings and Halflings if black decides to be hostile with his initial dwellings.

Taking blue would probably mean giving the game to Nomads. Taking brown would mean giving the game to Witches and taking gray would mean giving the game to Darklings. At least in theory, I'm glad games don't always play out as you initially expected.

Picking brown before black is much better than picking brown after black, because then black has to place both dwellings before your 2nd dwelling and you can decide whether you go to east or not. If black places 2 dwellings there you can still go if yellow is not blocking your way to south-west, but otherwise you need to ignore the east -- or just place a leech dwelling there without any expectations to form a town.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Matthias Reitberger
Germany
Nürnberg
flag msg tools
Avatar
mbmbmbmbmb
Skyswooper wrote:
1869 wrote:
If there is a blue faction, especially Mermaids, I wouldn't even consider it.


I'm not sure. Witches and mermaids can sometimes cohabit if there is Nomads for example (Mermaids might stay in the West to take all the leech from Nomads, and Witches can expand in the East).

But if you don't consider green, would you pick grey then? (As last pick is less likely to pick green because of the presence of blue).

Or would you pick another colour letting grey open for last pick?


I would take engineers and hope that the last player is sensible enough to take Nomads. Unlike Petri I'm not very fond of playing Nomads, they often look good on the board, build three towns and still come in third.
In a situation with black, green and yellow I would most often pick Swarmlings. And yes that might give Nomads a good position as I just encountered in http://terra.snellman.net/game/QuickEuro1810 especially if black thinks he should start in C1 which is often not good for them.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Space Trucker
Germany
flag msg tools
badge
Avatar
mbmbmbmbmb
1869 wrote:
And yes that might give Nomads a good position as I just encountered in http://terra.snellman.net/game/QuickEuro1810 especially if black thinks he should start in C1 which is often not good for them.
I didn't expect to ever see someone build 4 town and win cluster scoring....wow And 3 cults.
6 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Per Olander
Denmark
flag msg tools
mbmb
SpaceTrucker wrote:
1869 wrote:
And yes that might give Nomads a good position as I just encountered in http://terra.snellman.net/game/QuickEuro1810 especially if black thinks he should start in C1 which is often not good for them.
I didn't expect to ever see someone build 4 town and win cluster scoring....wow And 3 cults.


what makes it even more impressive, is that he got fav10+11+12, instead of income/2-3 step favors...
4 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Front Page | Welcome | Contact | Privacy Policy | Terms of Service | Advertise | Support BGG | Feeds RSS
Geekdo, BoardGameGeek, the Geekdo logo, and the BoardGameGeek logo are trademarks of BoardGameGeek, LLC.