Different Approaches to Rank

Rank is one of the most bothersome things to deal with in Street Fighter.  In order to get past rank four or five, you have to consistently win almost all the time.  One loss can mean the end of the road for a character with an eye on becoming a World Warrior.  Furthermore, taking a look at the records of typical NPC's, it seems that everybody wins more than they lose.  This seems to imply that there are a nigh-infinite number of "jobbers" on the circuit who just lose time and time again, eventually slipping into obscurity and disgrace.  It would also mean that some of these jobbers managed to get to rank six, seven and eight before they started losing constantly, for the sole purpose of giving the World Warriors enough wins to get to nine and ten.

In addition, this also places an unnecessary burden on the players to win all the time.  It makes for an unsatisfactory series of chronicles if your players keep getting dropped by their managers for losing.  This leads to placing all available experience into combat statistics, especially if the GM is particularly stingy *coughDerekcough*.  It also leads to doing unscrupulous things like picking easy fights to maintain one's record, something which is boring to play.  How can we get around this?

Lying

First of all, keep in mind that a fighter's record in the circuit is not usually kept by a regulating body.  It's quite conceivable that many fighters simply lie about their records in order to get some respect and a shot at a decent match once in a while.  In real life analogues like Vale Tudo and other combat sports, fighters tend to exaggerate their records, especially if they come from somewhere far away where the truth of their claims cannot be verified.  Nobody really knows a fighter's true record except maybe the fighter.  Thus, even though you might have a real record of 20-15-2, a slick manager/promoter might be able to pass you off as 25-3-1.  It's all about perception.

In a system like this, fighters can also just not count certain losses.  People tend to talk more about their victories than their losses, and if nobody was around to see the match, who will know?  The GM may rule that it is an accepted practice amongst fighters to "forget" about the losers of fights which are relatively low-key, such as a spontaneous back alley match between two fighters.  It can also be used as a source of blackmail:  the winner may agree to not tell tales about the loser getting his ass handed to him, but naturally he will count the win, and well people might get curious about who the loser was in this particular battle...

Changing Identities

Fighters also tend to change identities, like professional wrestlers, in order to start clean.  A fighter who dons a mask and uses a different name who is obviously very skilled will get attention as he cleans his way back up through the lower ranks, and may someday be forced to come clean about his earlier, lackluster career.  It also means that other fighters will be trying the same tricks, and may prove to be a frustrating problem for players of low rank.  Think of Dragon Rising's Kallista... a rank 1 with Ice Blast, fantastic maneuvers, tons of health and super high stats?  Yeah whatever.

Changing Divisions or Taking Sabbaticals

A fighter who is wholly disgusted with his record can try the old "change your division" trick to start over.  Usually these fighters will switch out between Freestyle and Traditionalist divisions.  A GM may also rule that a fighter who abandons the circuit for an extended period of time (say, a long stretch with no sanctioned fights) may try to start over with a clean record.  The period of inactivity would have to be substantial, like several years, in order for the fighter's old record to fade from the minds of the public.  Managers won't like this, as they can't be carrying dead weight around.

One possible solution to the hosing that fighters with a bad start get is to time-weight their records, much in the way standings for professional tennis players is done.  In this system, fights lose importance as time passes, allowing a fighter to bounce back from an early losing streak.  It requires a little extra bookeeping and careful time tracking.

Since fights will probably happen less often than tennis matches, the sample system presented here decays one's record more slowly than the tennis system.  You may adjust this system as you see fit.

Keep the fighter's record, wins, losses and draws, as is for the purposes of... whatever people use those numbers for.  However, in regards to rank calculation, calculate the win/loss record as follows, and determine rank based on total number of fights and the adjusted winning percentage:

• Fights within the past two years count full value (draw counts as 1/2 win).
• Fights between four and two years ago count half value.
• Fights between four and six years ago count one-quarter value.
• Fights more than six years ago count one-eighth value.

To keep things simple for the GM and players, it is suggested that all fights be adjusted for time at one annual point, such as the new year or the time of the World Warrior Grand Tournament.

Example:  A fighter has been on the circuit for 6 years.  His record is as follows:

Jan. 1 2004 rolls around, and his record is adjusted as follows:

Year	Actual Record (Total Fights)	Divisor	Adjusted Record (Adjusted Fights)
1998	2-3-0 (5)	8	.25-.375-0 (.625)
1999	2-4-1 (7)	4	.5-1-.25 (1.75)
2000	4-7-0 (11)	4	1-1.75-0 (2.75)
2001	8-3-0 (11)	2	4-1.5-0 (5.5)
2002	8-2-0 (10)	2	4-1-0 (5)
2003	9-0-1 (10)	1	9-0-1 (10)
2004	0-0-0 (0)	1	0-0-0 (0)

An Alternative Ranking System (Points System)

One could abolish the ranking system as is, and replace it with a more logical system based on challenges for rank and promotion in the circuit.  Here is a proposal for such a system:

All new fighters start at Rank One, as usual.  In order to progress in rank, fighters must meet certain challenges, which entail winning matches and gaining a certain score.  The fighter's "rank score" is increased by beating those of his own rank-3 or higher, and lost by losing to fighters of his own rank+3 or lower.  The system for this score is as follows:

For a win:  Gain [ 2 + ( difference in ranks / 2 ) ] points, never less than zero.  A fighter cannot gain points for winning over an opponent who is more than 3 ranks lower than he.

For a loss:  Lose [ 2 - ( difference in ranks / 2 ) ] points, never less than zero.  A fighter cannot lose points for a loss to an opponent who is more than 3 ranks higher than he.

For a draw:  The lower ranked fighter gains ( difference in ranks / 4 ) points, and the higher ranked fighter loses the same amount.  No score for fighters of equal rank.

Under this system, fighters will not actively seek out fights against lower ranks, as they stand to lose more.  Almost nobody will consider a fight versus someone 4 ranks lower than himself, as he stands to win nothing and lose much.

To advance one rank, a fighter must accumulate a number of points in this fashion equal to [ 5 + ( Current Rank x 2 ], at which point he advances one rank and his score is reset to zero.  Thus a rank one fighter could get to rank two by winning 3 matches and drawing 1 against other rank one fighters, but a rank 8 fighter would have to win 10 matches and draw 1 against other rank 8's to advance to rank 9.  A rank 10 fighter can never go above rank 10, but may gain up to 25 points from wins in order to safeguard his position against future losses.

If a fighter reaches -5 points, he is lowered by one rank and must start over with a score of zero.  Thus, a rank 5 fighter who is unlucky enough to get to a score of -5 points is now a rank 4 fighter with a score of 0.  A rank 1 fighter can never go below rank 1, but may have a score of down to -5 from losses which may count against his ambitions to go to rank 2.

Under this system, a fighter will advance more slowly to higher and higher ranks.  He will also find that he can lose his rank fairly quickly through a string of losses, but he will not be endangered by every single loss as he would under the original system.  This keeps the higher ranks of fighters progressively smaller than the lower ranks, while not unfairly penalizing those of high rank for an understandable loss.

The number of points required to increase in rank are as follows:

FOR GROUP FIGHTS, EVEN TEAM SIZES:  If the size of both teams are equal, the effective difference in ranks is calculated using the average ranks of each team.  For example, team A consists of a rank 2, two rank 4's, and a rank 5.  Team B has two rank 3's, a rank 5, and a rank 6.  Team A's effective rank is 3.75, while team B's effective rank is 4.25.  If team A wins, each member of team A will gain 2.25 points, and each member of team B loses a like amount.  If team B wins, team B only wins 1.75 points, and team A loses the same.  In the event of a draw, team A as the lower ranked team would gain .125 points, and team B would lose .125.

FOR GROUP FIGHTS, UNEVEN TEAM SIZES:  If for some reason the size of the teams are unequal, there is a strong advantage to the larger team that supercedes the simple sum of ranks (i.e. 4 rank 3 fighters can usually beat 3 rank 4 fighters, and one rank 5 fighter will have trouble dealing with the multiple attacks of 5 rank 1 fighters).  Unequal fights are avoided by promoters for this reason, but they can happen.  In these cases, effective team ranks are calculated as follows:

"Difference in size" is expressed as a negative number for the smaller team and a positive number for the larger team.

For example, team A's rank 2 guy is AWOL, but the fight must continue.  Team A is the smaller team, with an effective rank of (4+4+5)/3 - 0.25 = 4.083, and team B has a rank of (3+3+5+6)/3 + 0.25 = 5.917, for a difference of about 1.833 ranks.  If team A somehow manages to get the win, they stand to gain 2.917 points each, while team B would lose the same.  A win for team B is worth 1.083 points to them and a similar loss to team A.  A draw would result in points of  0.458 and -0.458 for teams A and B.

With this formula, it is possible to come up with some interesting uneven matches ("Balrog takes on all of Team Raven plus a mystery Rank 2 fighter, tonight at the Strip!").  Here are some possible matchups, assuming teams that have members of equal rank (Team B is the smaller team in all cases):
 

Team A	Team B
2 fighters rank 1 each, effective team rank 2.25	1 fighter rank 3, effective team rank 2.75
5 fighters rank 2 each, effective team rank 3.83	3 fighters rank 4 each, effective team rank 3.5
4 fighters rank 3 each, effective team rank 6.5	2 fighters rank 6 each, effective team rank 6.5
3 fighters rank 4 each, effective team rank 6.25	2 fighters rank 7 each, effective team rank 6.75
7 fighters rank 1 each, effective team rank 8.5	1 fighter rank 10, effective team rank 8.5

Another Alternative Ranking System (Elo System)

This is a far more complicated system, similar to the Elo rating system used for competitive chess and go.  It may be more work for the GM, but since this work is only done occasionally, the rewards of a more accurate and fair system may be worth the hassle.  One would have to subsume that there is a monitoring body that has information about matches in order to make this work, like a network of underground fighting journals, a central governing body, or even a federation of street fighters.  A GM may wish to use a seperate Elo system for each division, or subdivision, or only apply it to a highly regulated division such as a traditionalist circuit.

A basic explanation of Arpad Elo's system may be found here.  The basic idea is that a fighter's rating is used to predict the likelihood of his winning over another fighter, based on past performance.  The win expectancy is approximated by using the difference between the fighter's ratings.  Using this system, a fighter who wins over a much stonger opponent stands to gain a lot of rating points, while a fighter who beats a much weaker opponent doesn't gain too much, thus encouraging fighters to seek matches around their own level.

The winning chances for the stronger fighter are calculated with the following formula:

where D is the difference in ratings (posisitve if the fighter is higher rated, negative if lower rated).  Sample chances to win by rating differences:
 

Difference Win -1000 -900 -800 -700 -600 -500 -450 -400 -375 -350 -325 -300 -290 -280 -270 -260 -250 -240 -230 -220 -210 0.003 0.006 0.010 0.017 0.031 0.053 0.070 0.091 0.104 0.118 0.133 0.151 0.159 0.166 0.174 0.183 0.192 0.201 0.210 0.220 0.230

Difference Win -200 -190 -180 -170 -160 -150 -140 -130 -120 -110 -100 -90 -80 -70 -60 -50 -40 -30 -20 -10 0 (even) .240 .251 .262 .273 .285 .297 .309 .321 .334 .347 .360 .373 .387 .401 .415 .429 .443 .457 .471 .486 .500

Difference Win 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 0.514 0.529 0.543 0.557 0.571 0.585 0.599 0.613 0.627 0.640 0.653 0.666 0.679 0.691 0.703 0.715 0.727 0.738 0.749 0.760 0.770

Difference Win 220 230 240 250 260 270 280 290 300 325 350 375 400 450 500 600 700 800 900 1000 0.780 0.790 0.799 0.808 0.817 0.826 0.834 0.841 0.849 0.867 0.882 0.896 0.909 0.930 0.947 0.969 0.983 0.990 0.994 0.997

A fighter's rating changes based on his results in a match, and the rating of his opponent, according to the Elo formula:

where Rn=new rating, Ro=old rating, C=a constant modifier, S=score (1 for win, 0 for loss, .5 for draw), and Se=expected score, which is the chance to win.

The value of C is traditionally changed as the competitor's rating changes.  For fighters of under 2000, C=75.  For fighters over 2400, C=25.    From 2000 to 2400, C=325-(Rating/8).  A higher C value allows more fluctuation in the fighter's rating while he is young, while a C of 10 at the top levels prevents such wild fluctuations.  Note that the values for C here are significantly higher than those used in chess (from 30 to 10), as street fighters have far fewer matches than chess players, and a great fighter may in fact rise to 2500+ with fewer than 100 matches, as opposed to the thousands required by the standard Elo system.  Depending on the frequency of fights in your campaign, you may wish to adjust the values for C up or down.

Example 1:  A fighter with a rating of 1700 defeats an opponent with a rating of 1750.  He records the win, and his new rating is [1700 + 75 x (1-0.429)], or 1743.  The opponent's rating drops to [1750 - 75 x (0-0.571)], or 1707.  If the 1750 fighter won, his new rating would be [1750 + 150 x (1-0.571)], or 1782, while the loser would drop to [1700 + 30 x (0-0.429)], or 1668.  Since the odds were in favor of the 1750 fighter, he gains less rating than the 1700 fighter does if he wins.

Example 2:  A fighter with a rating of 2100 defeats an opponent with a rating of 1980.  The winner's new rating is [2100 + 62.5 x (1-0.666)], or 2121, and the loser drops to [1980 - 75 x (0-0.334)], or 1955.  In this case the 2100 fighter will gain fewer rating points than the 1980 fighter loses, since at 2000 and up the fighter's rating becomes more stable to reflect his establishment.  Still, if the 1980 fighter manages to win, his new rating leaps to [1980 + 75 x (1-0.334)], or 2029, and the loser drops to [2100 - 62.5 x (0-.666)], or 2058.

Example 3:  A foolhardy fighter with a rating of 2290 decides to challenge a fighter with a rating of 2520; the odds are 79% in favor of the higher ranked fighter.  A win for the 2290 fighter would give him a new rating of 2321, whereas the higher ranked fighter would drop to 2500.  A win for the 2520 fighter would yield new ratings of 2525 and 2282.  The high status of these fighters means that their ratings are more stable than most other fighters, and it will take a lot of wins for them to advance in rating, just as it would take many many humiliating losses for them to be dislodged from their legendary positions.

Example 4:  A new fighter enters a 4-fight round robin tournament in order to get an initial rating.  Everyone starts at 1200.  He wins his first match, bringing his rating to 1237.5.  In the second round, he is paired with another 1-0 fighter with the same rating, and wins again, bringing his rating to 1275.  In the third round, he is paired against another 2-0 fighter, but draws this round, making no difference in his rating (as the change in ratings for a draw vs. the same rating is zero).  He enters the last round as 2-0-1 and gets paired against someone who is 3-0 with a rating of 1212.5.  He wins the match, and his rating after 4 matches with a result of 3-0-1 is 1316.53, which is rounded up in next's week's fight sheet as 1317.  He is more than halfway to rank 2 after his first four matches.  A writeup of his performance in a manager's report might look like this:

All beginning fighters start with a rating of 1200.

A fighter's rank is determined more or less by his rating  A sample chart follows showing the relationship between rating and rank.
 

Rating Rank 0-1399 1 1400-1549 2 1550-1699 3 1700-1849 4 1850-1999 5

Rating Rank 2000-2149 6 2150-2299 7 2300-2599 8 2600-2899 9 2900+ 10

The GM may impose limitations on increasing rank, such as having had a certain number of total fights, etc.  Players may also lose rank if they do not periodically defend their status in the ring every once in a while.  Rank is not the same as rating, and may carry with it additional requirements.  A white belt may be as good a match fighter as a black belt, but that doesn't make him a black belt as soon as this is established (with exceptions like Kimo becoming an instant black belt in Joe Son's school).

A major change in the way new Street Fighters are established under the rating system is that a fighter's initial rating can be much higher than one would expect if he is naturally gifted.  A new fighter may achieve a rating near 2000 in just a few fights if he wins consistently against strong opponents early on.  This could help to explain the meteoric rise of many of the World Warriors as described in the rulebook, especially those who first become Street Fighters already in possession of master-level techniques like Fei Long, Blanka, Ryu, Zangief, etc.

Ratings can provide a quickly digestible "fight card summary" after the fight is done for publication in seedy underground journals, adding atmosphere to your campaign.  For a sample summary sheet, click here.

FOR GROUP FIGHTS, EVEN TEAM SIZES:  Any rating changes are applied individually for every fighter versus the average rating of the enemy team.  For example, Team A has fighters with ratings of 1400, 1520, 1550, and 1700.  They are Team B with ratings of 1130, 1510, 1600, and 1780.  Team A wins.  Every member of Team A calculates his new rating by counting a win against Team B's average rating of 1505.  Likewise, everyone on Team B calculates their new ratings based on a loss against Team A's average rating of  1542.5.  After the local eggheads who run the underground fight sheets have done their calculations, the fighters' new ratings are published in their latest edition:  team A has ratings of  1449, 1556, 1583, and 1718, while team B now has ratings of 1124, 1476, 1556 and 1720.  This system tends to "average out" disparate ratings of team members over time as they fight together.

FOR GROUP FIGHTS, UNEVEN TEAM SIZES:  A sanctioned event with different numbers of fighters per side is generally discouraged, and is a big headache for the ratings people (you), but they can happen.  In this case, calculate the effective team rating as follows:

This is different from the previous system, since ratings and rank are not the same.  A 2000 rated fighter should cream two 1000 rated fighters.  Once again, calculate rating changes after the fight for each fighter based on the effective team rating of the enemy.

Example:  Team A's 1400 rated fighter is unable to compete.  They have the match anyway.  Team A's effective team rating is [ ( 1520 + 1550 + 1700 ) / 3 ] + ( -1 x 125), or 1465.  Team B's effective rating is [ ( 1130 + 1510 + 1600 + 1780 ) / 4 ] + (1 x 125), or 1630.  The odds are about 3 to 1 in favor of team B.  If team A should win anyway, their new ratings will be 1569, 1596 and 1733, while the humiliated team B would drop to 1120, 1468, 1549, and 1716.

Some comparative notes for scheduling uneven group matches:

These values are not necessarily coincidental with the values given for uneven matches in the previous section... hey, I just made it up.  You can get away with this too... if it is obvious that the system is flawed (this would become apparent only if your campaign is rife with lopsided team matches), simply say that the governing body is "working out the bugs" and change the multiplier per additional fighter on the larger team.