BuzzerBeater Forums

A glimpse into that huge brain of Princeton-trained, human BBs.

So... the human BB's are Princeton trained? I should have guessed...

Motion [...] if you want to be strong in (offensive) Rebound or Inside Scoring this is definitely the wrong tactics choice.

The fact that offensive tactics have an influence on the offensive rebounding is quite surprising to me. A long time ago we were told by BB-Charles than the defensive choice had an indirect impact on the offensive rebounding.
I don't know how you calculate the effect, but are you sure about your claim?


From the rules:

During the course of the game, your coach will make adjustments as a result of how the game is going. If you are making a lot of inside shots, he will tell you to take more inside shots. If the opponent is making a lot of inside shots, he will tell your defenders to double-team in the post more often. However, any time that the coach makes a change away from the tactics you started out with, it will be a less effective than if you had simply chosen to double-team the post all along (i.e., you will find that it is easier for your coach to tell the defenders to focus even harder than they already are on inside defense if you are already playing a 2-3 zone). The closer your submitted tactics are to the optimal choice, the better, especially because the coach is only able to make limited adjustments once the game has started.

The question is, are these adjustments visible in the team ratings?

Rebounding and Inside Shot matter less than in Base Offense, and if you want to be strong in (offensive) Rebound or Inside Scoring this is definitely the wrong tactics choice.

Does your data have strong correlation between Motion against a 3-2 and/or 1-3-1?

Would you set free the data used to train the models?

Yeah, I know that BB said that team ratings are independent of the opponent. But then, I was just wandering how could you make observations between an offensive tactic like motion and the offensive rebounds you get without looking at opponents reboundings capabilities. This answer might trigger the question to whether is interesting or not to look at defensive tactics of the teams playing against the motion. I think it might be.

---

About the data you collected. There is no need to disclosure the private information of the people who collaborated with you in order to provide usefull data.

For example, I'm not interest at all into knowing the skills of people players or where are those players playing, etc. I would have to have skills set of players with the time they played, enthusiasm, etc. You do not have to give any information like players ID, Team ID, Game ID, it is not hard to make that public, you can simply alter that info(with a simple algorithm you only you know or even a more easier way...) and mantain the relevant info.

Because, If player X have Y set skills with Z info of minutes played in some games, etc. I'm not interested into knowing how player X is in BB, not even wich is the game that you recorded. It will be really nice to know what the player ID "abfds67Wuj" did in the game ID "fjhs678hrsG" with no way to know that the player in question is (5928033) in this game (15696345). Are you getting where am I going? There is no need to disclosure private info in order to provide the data info you have collected in order to held more experiments.

Last edited by Zero, the Magi. at 11/4/2009 11:17:53 AM

I get your point about anonymization, although with the possibility of a bulk search using the BBAPI interface this wouldn't really survive a serious attempt to locate the specific matches... not that I seriously believe anyone would bother.

yeah... I don't know the BBAPI interface but it seems a really huge effort to find something that you actually do not a have a clue of what it would be(could be virtually any team!) and sometimes leading to ambiguous results.