BuzzerBeater Forums

I am only saying that the data should be collected under normal circumstances to get rid of unwanted effects.
For the sake of science :)

hrudev cites one reason why the data could be damaged.

In a Patient, shot distribution can dramatically change.
Princeton is a slow-paced attack, even slower than Patient and Low Post going by the manuel.
Pace and its consequences are elements difficult to understand for many players.
Your data is quite interesting but we should remain cautious on an interpretation.

Seeing as you have this data I would love to hear the truth about just how slow the pace of it is. There is a lot of debate about what is the slowest tactic with some saying the order is motion, low post, Princeton and patient being the slowest but others disagree.

As you said I don't think your data on shot distribution is affected by the quality of teams running Princeton as much as the success percentge would be and perhaps it is presumptious to assume that all good teams run look inside and only crappy teams run Princeton.

Just saying if you have data on 30,000 shot atempts taken using base and 5-6,000 using Princeton I and many others would love to see a list of how many shot attempts each offence generated under your study.

PS: I bet if you have 30,000 base you have 130,000 look inside. It would also be interesting but of little use to see how popular each offence is.

If I had to guess I would speculate something like:
LI 40%
Push 15%
Patient 15%
Motion 10%
Run and Gun 5%
Low Post 5%
Base 5%
Princeton 2%
Inside Isolation 2%
Outside Isolation 1%

I will reword it.
The game engine is not built on attacks having aims for shot distribution but primarily on individual skills in relation to the team, the opponent and the matchup.
The choice in attack will alter the basic probable shot distribution for a particular game.

This data on shot distribution is not relevant if the teams are not well set up.
It's evident in the case of Cs playing PG and PGs playing C, the shot distribution will be different than in a "normal" game.
We don't know for sure which skills at which position are expected to make a Princeton work (this alone should have the Princeton reworked).

In short, if you want your PF to take 3s, tactics or skills ?

Seeing as you have this data I would love to hear the truth about just how slow the pace of it is. There is a lot of debate about what is the slowest tactic with some saying the order is motion, low post, Princeton and patient being the slowest but others disagree.

We're moving kind of beyond the scope of this thread, but briefly... you don't actually need shot distribution data for this, just boxscores, which are available through the bbapi. Which means you can base it on far more data than what I referred to above. When I ran those numbers a while ago, it looked like:

An average possession in Base lasts 15.1 seconds.

Patient/LP/Princeton increased this by ~9%. (longer possessions = fewer possessions)
LI/Motion increased it by ~4%.
RnG shortened it by ~3%.
Push was basically identical.

That LI was slower than base, and essentially equivalent to motion, always sort of surprised me. but this data is very old (definitely pre-season 15, maybe much earlier)... if some enterprising bbapi user wants to run more recent numbers to decide whether *that* should also have a bug thread, feel free.

Thanks that is already incredibly interesting and goes against 100% of "conventional wisdom" which says LI is a fast tactic. I know this is off topic again but do you have the breakdown of the frequency of use of each offence?

I know they are "easily available through bbapi" but having no idea what that is or how to use it is a bit of an issue for a noob like me.