Item the first:
 

Now that we have picked a player, we must determine what he will do with the ball. There are three things he can do: turn the ball over (3), shoot (5) or be fouled (4). Every player has a real-life turnover percentage, field goal attempt percentage and times fouled percentage that is relative to his total number of actual possessions. At this point, many factors can adjust those percentages and affect whether we go to step 3, 4 or 5. Generally, we will view the chance that he shoots as the amount remaining after modifications to fouled percentage and turnover percentage. These factors include: the man defending the player, the rest of the defense, the type of offense, the type of defense, a player’s fatigue value, any team under-possession penalty and the player’s over-possession penalty.

If a player has committed a turnover, he may have committed a ball-handling mistake, poor pass or offensive foul. (ed. 2 of these things could lead to stoppage in time BTW but that's for another day)The percentage chance that an offensive foul has occurred is relative his actual personal foul rate. This is figured into the chance in the previous section. Ball-handling and passing turnovers each use a fixed percentage of the remaining chance that is based on historical averages for these turnovers. If it is one of these two types of turnovers there is a chance for a steal. This chance is based on the defenders and is relative the steal weighting used in the decision above. If a steal has occurred, there is a chance for a breakaway. Either way, the player who steals the ball is determined after the steal event is known. Each defensive player’s chance for a steal is relative to his contribution to the steal weighting in #2. If no steal has occurred, the ball simply goes to the other team.

I would challenge the attendees to discuss and unpack their relevant observations regarding the following thesis points:

are turnovers subject to too much aberrant/outlier behavior(more or less than other stat categories)?

should WiS make TOV% a searchable stat (this one's for you ash)?

consider the weighting of other potential modifying factors (opp stl%, pace et al) - what could they be? how should they be weighted in your opinion?

Posted by felonius on 2/14/2014 3:10:00 PM (view original):

I would challenge the attendees to discuss and unpack their relevant observations regarding the following thesis points:

are turnovers subject to too much aberrant/outlier behavior(more or less than other stat categories)?

should WiS make TOV% a searchable stat (this one's for you ash)?

consider the weighting of other potential modifying factors (opp stl%, pace et al) - what could they be? how should they be weighted in your opinion?

Posted by felonius on 2/14/2014 5:14:00 PM (view original):
thoughts re Stl% 

stl % is flawed as an indicator of likelihood of a steal occurring in a given play as it is actually a measure of what % of a team's steals a given player was responsible for not a measure of how many times a player created a steal per play (let alone a direct head-to-head metric of steal vs opp TOV) - the one guy who occasionally creates steals on a team otherwise not prone to doing so will have a higher stl% than a true ball hawk amongst ball hawks

in other words it's a bad metric in the first place and it's being used in a way that doesn't really apply - they'd be better off using a steals per minutes played figure in its stead

Posted by felonius on 2/14/2014 5:13:00 PM (view original):
also I think TOs have a higher magnitude of fluctuation than many other stats* - a swing of 10 turnovers from game to game is like a swing of shooting 20% to 80% from the field game to game and yet it happens all the time - I just had back-to-back games of TOs in the 30s against no press and with a team that has an aggregate TOV somewhere around 14%


*magnitude of fluctuation is a serious oversight on ******'s part as far as I'm concerned - the smaller the stat unit the great the fluctuation - consider the problem of fatigue amongst scrubs for one - a guy who plays 3mpg who averages 3.5mpg for instance

Posted by ncmusician_7 on 2/14/2014 6:07:00 PM (view original):

Posted by felonius on 2/14/2014 5:13:00 PM (view original):
also I think TOs have a higher magnitude of fluctuation than many other stats* - a swing of 10 turnovers from game to game is like a swing of shooting 20% to 80% from the field game to game and yet it happens all the time - I just had back-to-back games of TOs in the 30s against no press and with a team that has an aggregate TOV somewhere around 14%


*magnitude of fluctuation is a serious oversight on ******'s part as far as I'm concerned - the smaller the stat unit the great the fluctuation - consider the problem of fatigue amongst scrubs for one - a guy who plays 3mpg who averages 3.5mpg for instance

If both teams run uptempo, each team will have around 100 FGA (generally speaking).  Even if you consider a TO the equivalent of 2 missed FGA, a 10 TO swing would be more like a FG% swing of 40% to 60% IMO.

Posted by ncmusician_7 on 2/14/2014 6:00:00 PM (view original):

Posted by felonius on 2/14/2014 5:14:00 PM (view original):
thoughts re Stl% 

stl % is flawed as an indicator of likelihood of a steal occurring in a given play as it is actually a measure of what % of a team's steals a given player was responsible for not a measure of how many times a player created a steal per play (let alone a direct head-to-head metric of steal vs opp TOV) - the one guy who occasionally creates steals on a team otherwise not prone to doing so will have a higher stl% than a true ball hawk amongst ball hawks

in other words it's a bad metric in the first place and it's being used in a way that doesn't really apply - they'd be better off using a steals per minutes played figure in its stead

That isn't how STL% is measured.  If a player gets 3 steals per 100 possessions by the opponents, he will have a STL% of 3 (regardless of how many steals his teammates get).