Strength Point Replacement

  1. 2 weeks ago

    Dear Vex,
    I believe that strength points should be replaced. Although there is a good intention behind them, to measure the difficulty of a team's schedule, the current strength point system is not the most efficient way to do so. For example, if two robots were to have an identical schedule (theoretically), the better robot will always end up with fewer strength points because their talent will cause the opponent's score to decrease. This is an example of how the dominant teams often win with very large margins and limited strength points, but the team's that may have received a lucky schedule will often time barely win matches, resulting in large amounts of strength points. Because strength points do turn out to determine a tiebreak very often, I believe that Vex should experiment with a new system to measure the strength of schedule. One way they could do this is to measure the combined number of wins of a team's opposition. Using this method, a schedule's difficulty is determined by the actual results of how well a team's opponents did against the rest of the tournament field, and teams that win against a tough field of opponents will be rewarded more than teams that got lucky with facing weaker robots all day. I believe that Vex should take this method into consideration for next year's tournament setup, and possibly even experiment with it to see if they can find a better method for ranking teams than the current strength point system.

  2. CyberHawks7983

    Jan 7 Event Partner, V5 Beta Tester Bakersfield, CA 7983 (A,J,K,M,R,U,W,Z)

    I've always thought opponents win percent minus partners win percent or something similar would make more sense as strength of schedule ranking.

  3. lacsap

    Jan 7 Event Partner, V5 Beta Tester Massachusetts 9791[a-z]

    Read the manual, the next tie breaker after SP is highest match scores...

    No need to devise another ranking system... yet...

  4. @lacsap Read the manual, the next tie breaker after SP is highest match scores...

    No need to devise another ranking system... yet...

    I think you may have misunderstood the point. He is saying that strength points are so often a tie breaker and believes they are an unfair way to do tiebreakers in a game that has swing points. If you’re suggesting skipping strength points entirely and to go with highest score instead that would be interesting.

    I also think the opposite of strength points would show strength points better this year. What if SP were the points of the winning alliance?

  5. lacsap

    Jan 7 Event Partner, V5 Beta Tester Massachusetts 9791[a-z]

    @Hagan_60883A I think you may have misunderstood the point. He is saying that strength points are so often a tie breaker and believes they are an unfair way to do tiebreakers in a game that has swing points. If you’re suggesting skipping strength points entirely and to go with highest score instead that would be interesting.

    I also think the opposite of strength points would show strength points better this year. What if SP were the points of the winning alliance?

    Probably misread OP :) The silver lining on this is that APs are even - hopefully indicating a lot more programming out there! I guess this will have to be a future game change. Unless the GDC wants to "beta-test" the idea at Worlds :)

  6. Got a Screw Loose

    Jan 7 Ohio 44833B

    @lacsap Probably misread OP :) The silver lining on this is that APs are even - hopefully indicating a lot more programming out there! I guess this will have to be a future game change. Unless the GDC wants to "beta-test" the idea at Worlds :)

    That's what they did for BO1, so I wouldn't think it beyond their limits to do so. But we all know how BO1 went...

  7. JuiceBox

    Jan 7 Ohio 6741E

    Let’s not forget the fact that the SP system can be easily manipulated where if a team is winning by a lot they just score for their opponent completely defeating the system that is in place.

  8. @JuiceBox Let’s not forget the fact that the SP system can be easily manipulated where if a team is winning by a lot they just score for their opponent completely defeating the system that is in place.

    Very true, that’s a part of the game I dislike. In our first year (star struck) we were a pretty awful team. I remember we were really happy that our last match at a competition was with one of the top teams that day. We were winning that match until the end when our partner told us to quit for SP. I know our team will never forget that frustrating loss and we don’t like to play that way ourselves.

  9. When a game allows an alliance to score and de-score points for the opposing alliance, SPs make sense. A very good team will track where they are in the standings and the match to adjust the SP as needed. The best teams do take advantage of this.

    The biggest problem with SPs is when the partner team only wants to score max points or games like ITZ where it is challenging to score/de-score points for the opposing alliance. There not much you can do about the later. The former you combat over multiple matches.

    I don't think max score or average score will work b/c that is more schedule dependent than SPs. They really depend on luck of the draw, how good is your partner and how hard is your schedule.

  10. @rpm When a game allows an alliance to score and de-score points for the opposing alliance, SPs make sense. A very good team will track where they are in the standings and the match to adjust the SP as needed. The best teams do take advantage of this.

    The biggest problem with SPs is when the partner team only wants to score max points or games like ITZ where it is challenging to score/de-score points for the opposing alliance. There not much you can do about the later. The former you combat over multiple matches.

    I don't think max score or average score will work b/c that is more schedule dependent than SPs. They really depend on luck of the draw, how good is your partner and how hard is your schedule.

    The same could be said about WP or AP. Even if you have the best autonomous, if your partner has none you are at risk of loosing auton if you are against 2 mediocre autonomous. Schedule will always effect any method of ranking.

  11. JuiceBox

    Jan 8 Ohio 6741E

    @Hagan_60883A The same could be said about WP or AP. Even if you have the best autonomous, if your partner has none you are at risk of loosing auton if you are against 2 mediocre autonomous. Schedule will always effect any method of ranking.

    I’m not sure how well it would work but maybe instead of using the losing alliances score you could do the difference between the two scores. It could be manipulated in the same way the current SP system is but it would be a little harder because if you make it a close game you could lose.

  12. @Hagan_60883A Very true, that’s a part of the game I dislike. In our first year (star struck) we were a pretty awful team. I remember we were really happy that our last match at a competition was with one of the top teams that day. We were winning that match until the end when our partner told us to quit for SP. I know our team will never forget that frustrating loss and we don’t like to play that way ourselves.

    118 suffered their only qualification loss in Skyrise at Worlds doing the same thing :)

    But I totally agree with the OP. Combined match score, median match score, rank of opponents, etc. are all much better tie breakers than SP.

  13. Rick TYler

    Jan 8 Teachers/Coaches, Event Partner, V5 Beta Moderator Redmond, Washington Founder of Exothermic Robotics

    @Anomaly (...)Combined match score, median match score, rank of opponents, etc. are all much better tie breakers than SP.

    This would be the end of defensive strategies. No use winning 10-5 if it kills your tie-breaker score.

  14. @Rick TYler This would be the end of defensive strategies. No use winning 10-5 if it kills your tie-breaker score.

    Remember this is for the 3rd tiebreaker to begin with. But either way I don’t think this would harm defense as the good defensive teams will still win games and be recognized by other teams. They will get to execute their strategy in eliminations as well.

  15. @Hagan_60883A The same could be said about WP or AP. Even if you have the best autonomous, if your partner has none you are at risk of loosing auton if you are against 2 mediocre autonomous. Schedule will always effect any method of ranking.

    Can't really disagree. Though there is no good replacement for WP. At least an EP can make it more accurate with more matches. (Though I'm guilty of not doing so to make it easier for teams to run Skills). And AP is better than the other tie breakers. It is also more accurate with more matches.

    Would anyone want Qual order + Skills Ranking order with lowest number to determine finals ranking. Except that every team does not run Skills, it could be used as the next tie breaker before SP.

  16. @Rick TYler This would be the end of defensive strategies. No use winning 10-5 if it kills your tie-breaker score.

    That's a good point, but usually defensive bots tank in quals anyhow. 62A was 2-3 at states and 4-6 at worlds. They were also division champions :P

    @rpm Can't really disagree. Though there is no good replacement for WP. At least an EP can make it more accurate with more matches. (Though I'm guilty of not doing so to make it easier for teams to run Skills). And AP is better than the other tie breakers. It is also more accurate with more matches.

    Would anyone want Qual order + Skills Ranking order with lowest number to determine finals ranking. Except that every team does not run Skills, it could be used as the next tie breaker before SP.

    Chess has flexible qualification matches where players face other players with the same record. This helps keep teams from being carried to good records and reduces the number of matches qualifications take to create a reasonable normal curve that represents the ability of teams. (And maybe even frees up some time for bo3 : )

    And yeah, APs are a *way* better tie breaker than SPs.

  17. Kashan hussain

    Jan 8 Washington 8931C

    If I understood this right, there is one problem that I see which is also something that deterred any other SP strategy and that is that SP would have to be calculated after every match has been played. This could cause problems as the placement that everyone has throughout the tournament would be inaccurate and it wouldn't be in place at all at the start of the tournament.
    I do agree with the issue of how it doesnt show the stronger team in tiebreakers.
    I honestly dont know. SP has its issues, but just about every other system would also have some also. Its sortof like Video Review in sports. There could be a possible error, but unless it is clear and obvious on video replay, only then will the call on the field be reversed, and if there isnt, even if there is some evidence on replay that the call could be wrong, then the original is favored. This analogy works with SP being the existing call on the field, while other systems are the replays for video review. There really isnt a 100% working method that would be worth a whole system change.

  18. 536Mentor

    Jan 8 Event Partner, V5 Beta Tester Appleton, WI 536

    @Anomaly Chess has flexible qualification matches where players face other players with the same record. This helps keep teams from being carried to good records and reduces the number of matches qualifications take to create a reasonable normal curve that represents the ability of teams. (And maybe even frees up some time for bo3 : )

    And yeah, APs are a *way* better tie breaker than SPs.

    As an EP, the difficulties with any form of flexible schedule that attempts to match up teams of equal record are several. One, how many matches would be needed to establish "equal records"? This adds to the number of qualifying matches, not reduces the number. Also, if the chess is is being played in a sanctioned tournament, the level of the players are based on the player's US Chess Federation rankings or FIDE rankings. I would have an issue with ranking robots on their past performances as a team with a less than stellar robot may arrive at the next competition with a very much improved or very different robot. In chess your ranking is a pretty good indicator of your ability and it is unlikely that someone ranked as an average player will become a Master player in a week or two. Two, from a spectator standpoint, many spectators have access to the match schedule and plan their viewing based on that. It might very difficult and confusing for a family member to know just when to watch, if the match schedule is fluid. Three, from a team standpoint, having a set match schedule lets a team plan their day. If they have a big gap in their schedule, for example, they know that they have time to do major repairs or programming or eat during that time. With a fluid match schedule, that becomes difficult.
    I don't disagree that SPs are not a great system. Teams scoring for their opponents is also confusing for spectators. I have yet to see a reasonably implementable replacement for SPs so far.

  19. last week

    Got a Screw Loose

    Jan 8 Ohio 44833B

    What about just removing SPs entirely? Based on the game, sometimes this could be more useful and accurate than a point-swinging game.

  20. Robo_Eng_13

    Jan 8 Indiana 1483

    VEX/RECF already made a big step in the right direction by maing AP the second tie breaker and moving SP down to third. Axing SP and jumping then to highest match score, then to electronic coin flip would be generally reasonable.

    However, it should be possible to get a vague strength of schedule metric without incentivizing teams to hold back or score for the other team (to me, the other team scoring points for you is way more demoralizing than getting beaten in a landslide).

    The total number of WP for all opponents so far? Maybe even multiple levels deep. This would give a good indication of a team's true overall strength of schedule while also being more difficult to deliberately manipulate.

    The best (to me) would be something like the Google Page Rank algorithm, unfortunately it would not be valid until part of the way through the matches, and it would put another constraint on the match schedule generation to ensure no independent groups of competitors who never interacted even indirectly with another group.

    I have run the Page Rank algorithm over completed data and it had pretty good predictive power (using worlds qualifier data to determine rank, then worlds elimination data to test accuracy, generally about 5% more accurate than just betting on the higher seed to win)

    It would leave some ambiguity in the rankings during the qualification matches, but could produce a solid and reasonable third tie breaker before alliance selection. Besides, rankings during qualifications are very nebulous and poorly interpreted anyway, with the first team to perform their 7th match when the others have only done 6 seeing themselves in 1st place and celebrating.

  21. Newer ›
 

or Sign Up to reply!