roos
Joined: Aug 02, 2003
|
 Posted:
Aug 09, 2003 - 09:41 |
 
|
Thank you for releasing the formula. I hope that you hope to get input for making it better. Well, here are my suggestions.
Suggestions:
1. Combination of pass and throw team mate should only add one to the value if there is a player with right stuff that can play next game.
2. The AvaregeAV formula is imperfect, the player multiplier should affect the AV on the player before it is added or else a player with significally different AV from the rest with a low multiplier will get to high impact. It doesn't make very much difference though.
3. The formula doesn't take in account at all that players number 12+ doesnt play the whole game. I suggest you sort the players by value, highest first, before applying the multiplier and from number eleven in that count multiplie the multiplier by 1/2, 1/3, 1/4 and so on until 16. One could find the proper statistics of this for each race by going through the logs. I guess 1/2 of the game played for player 12 might be a little lower than avarege but that 1/4, 1/5 might be slighly the other way. Finding some statistics would be good though. Anyway, the str available for each kickoff cant include the players on the bench without punishing the teams with reserves.
4. Long legs, sprint and sure feet should get an extra CUMU addition if speed is enough for a one-turn TD.
5. A rating value for each player should be shown on the team web page to help those who doesnt understand or bother to figure out how the str system works. Could be shown as % of the team value for each player. The same for rerolls, coaches and so on.
6. Im not sure, but the coaches and cheerleaders might be overrated, of the top 10 teams, none have cheerleaders and some dont even keep the retired players as coaches. That at least means that those winning doesnt think they are worth their value.
7. The stars of the teams seems more important than the rest of the team. That could be fixed the same way as the reserves problem. The highest ranking players could get a higher multiplier. Adding maybe, 0.5, 0.4, 0.3, 0.2 and 0.1 the the five highest might do the trick. Before doing that Big Guys maybe should be put at possitions 11, 10 and so on. Both the reserves and star multiplier shift could be made with some sort of algorithm instead ofcourse.
Questions:
1. The cumulative, is it like?
This value is (added to the player value and increased by 0.5 for every skill) unless otherwise specified. The increase is done after the addition each time.
or
This value is added to the player value and (increased by 0.5 for every skill) unless otherwise specified. The increase is done after the addition each time.
I mean, is the player value increased with the cumulative for every skill? If so, does it add the CUM total after every skill or the CUM current as skill get applied? In that case, in what order does the skills get applied?
Else (if so), why not add the CUM increase to the value directly?
In fact i think i get it, but would like it clearified, the order of the skills is very interesting if it works as I figure it though.
/Roos |
|
|
Christer
Joined: Aug 02, 2003
|
| roos wrote: | | 1. Combination of pass and throw team mate should only add one to the value if there is a player with right stuff that can play next game. |
I honestly doubt this makes a difference as I don't think anyone will bother to choose pass on a big guy if there isn't a gobbo of fling to throw..
Either way, I added this requirement to that particular combination.
| roos wrote: | | 2. The AvaregeAV formula is imperfect, the player multiplier should affect the AV on the player before it is added or else a player with significally different AV from the rest with a low multiplier will get to high impact. It doesn't make very much difference though. |
Well.. I think you misread that part:
AverageAV = Sum(av*Multiplier) / PlayerCount.
If I understand what you're saying, you're requesting it to work just like it does 
| roos wrote: | | 3. The formula doesn't take in account at all that players number 12+ doesnt play the whole game. I suggest you sort the players by value, highest first, before applying the multiplier and from number eleven in that count multiplie the multiplier by 1/2, 1/3, 1/4 and so on until 16. One could find the proper statistics of this for each race by going through the logs. I guess 1/2 of the game played for player 12 might be a little lower than avarege but that 1/4, 1/5 might be slighly the other way. Finding some statistics would be good though. Anyway, the str available for each kickoff cant include the players on the bench without punishing the teams with reserves. |
On the other hand, it also doesn't take into account that if you have 11 players, all of them won't play the whole game. The playercount modifier is there in order to approximate this though. So it's just a matter of finding the "correct" player count modifier.
| roos wrote: | | 4. Long legs, sprint and sure feet should get an extra CUMU addition if speed is enough for a one-turn TD. |
I know I have considered adding a special combination for one-turners. I seem to have forgot to add the effect of sprint though.
I have modified the strength for the one-turners a bit. They were a bit low.
| roos wrote: | | 5. A rating value for each player should be shown on the team web page to help those who doesnt understand or bother to figure out how the str system works. Could be shown as % of the team value for each player. The same for rerolls, coaches and so on. |
Admins have the player value on the page already. I will clean up how it looks and display it for everyone.
| roos wrote: | | 6. Im not sure, but the coaches and cheerleaders might be overrated, of the top 10 teams, none have cheerleaders and some dont even keep the retired players as coaches. That at least means that those winning doesnt think they are worth their value. |
The average game has 3.2 TDs (taken from the fumbbl results). One of these are probably at the end of a half, so let's assume we have 4 kickoffs where the reroll for coaching staff makes a difference.
There is a 4/36 (1/9) chance of rolling the "Brilliant Coaching" result on each kickoff so you are expected to get one roughly one every other game.
Having 6 coaches _more_ than your opponent will guarantee you the extra reroll for the half (or half a reroll for the game). So you end up getting 3 strength every other game (or 1.5 strength per game).
So, it turns out you're right. Coaches are value too high (and with the same reasoning, so it cheerleaders). I did this calculation when I decided on the factors for staff, but miscalculated the number of rerolls per game.
I'll change it to the "proper" 1/4 factor for coaches. I'm going out on a limb and will guess 1/5 for cheerleaders. It's not like it will make much of a difference though lol.
This is most likely way too much thought for a very small benefit
| roos wrote: | | 7. The stars of the teams seems more important than the rest of the team. That could be fixed the same way as the reserves problem. The highest ranking players could get a higher multiplier. Adding maybe, 0.5, 0.4, 0.3, 0.2 and 0.1 the the five highest might do the trick. Before doing that Big Guys maybe should be put at possitions 11, 10 and so on. Both the reserves and star multiplier shift could be made with some sort of algorithm instead ofcourse. |
Interesting point. However, the stars of the teams will have a greater effect in the current system because of the cumulative. Three skills on a single player results in a higher strength effect than three skills divided on three players.
It will also not make sense on "bashy" teams as their stars are likely to be the PO/RSC chaos warrior instead of the 3-4 beastmen who do the scoring.
Before I consider adding this feature (which incidentally is rather resource demanding) I want to have a more extensive stats analysis. The change must increase the accuracy of the _average_ team, which I'm not conviced it would.
| roos wrote: | | The cumulative |
Example:
Human Lineman with block, tackle, guard:
Player Value starts at 5.
We look up the 6 3 3 8 stats and see that this is exactly the "average" stats, so no change here.
Skills:
Block: Add 2+0 to the player value, adding 0.5 to Cumulative after we add
Tackle: Add 2+0.5 to the player value, adding 0.5 to Cumulative
Guard: Add 2+1 to the player value, adding 0.5 to Cumulative (this last add is irrelevant though)
Player Value is now at 12.5 and we check the "slowdown" filter:
The value is above 12, so we halve the effect of everything above 12 ending up with a player value of 12.25 which is the final value. Naturally, the team-wide effects are being applied to it (playercount weight) but this is nothing we can calculate without having the rest of the team specified.
Hope this answers some questions.
-- Christer |
|
|
roos
Joined: Aug 02, 2003
|
| Posted:
Aug 09, 2003 - 19:16 |
|
| Quote: | | I honestly doubt this makes a difference as I don't think anyone will bother to choose pass on a big guy if there isn't a gobbo of fling to throw.. |
The gobbo or whatever could be lost or SI.
| Quote: |
Either way, I added this requirement to that particular combination. |
There are a few other skills I didn't notice with the same situation: always
hungry and throw team mate itself
| Quote: | | If I understand what you're saying, you're requesting it to work just like it does |
You're right ofcourse, i missed the each player part.
| Quote: | | On the other hand, it also doesn't take into account that if you have 11 players, all of them won't play the whole game. The playercount modifier is there in order to approximate this though. So it's just a matter of finding the "correct" player count modifier. |
But they will, being KO or even a casualty is participating in the game. He is there as a result of ops play. It just opens up another spot for a reserve to join too. In a game you can loose players, but thats part of the play. He is activated and takes up a spot in the injurieboot. A player that is in the reserves on the other hand, he isn't active at all. He doesn't affect your teams ability to win until someone gets hurt.
However just as you say, one could roughly approximate this with the Weight table. But then it has to be much more inpact on the values. Player number 16 is only used when there is 5 men out. He is ofcourse the player with the lowest value but he might roughly represent 1/30 of the player value. The drop from 15 to 16 should be according to that. I dont know how big part the FF, rerolls and so on is so i can't compute a value. It might be a little bit lower since one can use the reserves by fouling a lot which is a working strategy.
| Quote: | | Interesting point. However, the stars of the teams will have a greater effect in the current system because of the cumulative. Three skills on a single player results in a higher strength effect than three skills divided on three players. |
The effect from the CUMU, however, is somewhat negeted by that beyond 12 points all the points count only half as much. In fact it might be the 12+ thing that makes the stars better than their value.
| Quote: | | It will also not make sense on "bashy" teams as their stars are likely to be the PO/RSC chaos warrior instead of the 3-4 beastmen who do the scoring. |
On the other hand, their way of winning is by reducing the number of the ops, so that the beastmen or whoever can score. Its still the stars that enables them to win.
The graph looks somethign like this:
The * is at 12 points
.................................................................
............................................................../..
.........................................................._/....
......................................................_/........
................................................__/............
......................................... __/..................
...................................___/.......................
.........................____/...............................
.............*____/.........................................
..........._/...................................................
......__/......................................................
__/............................................................
.................................................................
0...............................................................
| Quote: | | Before I consider adding this feature (which incidentally is rather resource demanding) I want to have a more extensive stats analysis. The change must increase the accuracy of the _average_ team, which I'm not conviced it would. |
Sounds very sensible. It could also be done by modifying the 12+ algorithm. Just change it so it works only from 12-40 or whatever. I still don't have the feel of how high a players value is so 40 just a wild guess.
| Quote: | CUMU
Hope this answers some questions. |
Everything but the order which the skills get applied. I hope it is not in the order the player gets them as one could plan the players by choosing skills in the right order to get a lower value, that wouldn't be good at all. Players with the same skills and stats should always have the same inpact on the str value.
/Roos
[/quote] |
|
|
Christer
Joined: Aug 02, 2003
|
The highest player value in the leauge is (as far as I know.. I can't check for sure) is 31.3 and this is a one-turner gutter runner with all skills and no injuries.
The order of skills should make no difference.
Looking over the list of "exceptions", there seems to be two skills that make make it break: Big Guy and Wild Animal. These two modify the cumulative without adding it to the player value, which will make it differ depending on when it gets processed. I will change these to a static reducement in player value instead so that this problem is resolved.
-- Christer |
|
|
roos
Joined: Aug 02, 2003
|
| Posted:
Aug 09, 2003 - 21:17 |
|
ofcourse the order make a diff, let's say you have three skills, two with low CUMU and one with high.
If the high one is processed first it will make an impact twice, if applied last it never will. |
|
|
Christer
Joined: Aug 02, 2003
|
There is no "high" or "low" Cumulative values. It always updates with the same value. |
|
|
Malthor
Joined: Aug 02, 2003
|
| Posted:
Aug 09, 2003 - 21:48 |
|
Quick comment. From my quick scan of the formula, apart from 0 and 1 TRR teams, 2-8 TRRs gives the cost of the TRR for starting team x number of rerolls for STR.
It seems to me that the higher agility teams (namely Elves) need less rerolls than lower AG teams. Look at how many high TR Dwarf and Chaos Dwarf teams go for 6-8 rerolls whereas high TR Elf teams often have less (partly because they don't have the money due to replacing players!).
Anyway, my comment is that it is obvious that some teams get better use of rerolls 5, 6, 7 and 8 than other teams, and perhaps this could be reflected in the Strength formula. Eg, TRR 7 for team with Average AG or 3.75 is worth .6, TRR 8 is worth .5.
Don't know if this is easy to implement, just a thought I had that may or may not be useful
M |
_________________
ex Monkey (original Team Approvers in 2004)
ex Admin
ex Ranked Tournament Manager
still disliked all round! |
|
fly
Joined: Aug 02, 2003
|
| Posted:
Aug 09, 2003 - 22:48 |
|
i got curious, christer could you post a link to the player with the highest value ? |
_________________
I play for fun. I play to win.
Do you play CPOMB 'cause you can't win otherwise?
No, that's a rhetorical question. |
|
MattiasF
Joined: Aug 02, 2003
|
| Posted:
Aug 12, 2003 - 17:14 |
|
From Christers description, I guess he meant he unfamous 'Swifty'.
http://fumbbl.com/FUMBBL.php?page=team&team_id=6974
Back to the weighting of # of players on the team part....
It just occurred to me that a team with many nigglings have a better use for their reserves. The Strength of the player itself is reduced by the nigglings, but maybe this should be somewhat countered in the weigthing part? Then the reservers maybe should not be discounted. Hope you understand my point...
Brain gone al haywire after a very warm day with lots of coding  Dont have the spirits in me (yet) to actully calculate on this. Maybe its a subliminal effect... |
|
|
|
|
| |
Gnomes are trash
Search
