I thought I would look into the woes of the bottom of the NBA standings Dallas Mavericks. Seeing how my automatic scouting algorithm could identify how to help Dallas become more competitive after the first 11 games of the season (not that they probably want to )
Yep they certainly appear to be tanking with the salaries they are paying. They would love to get off the Matthews contract but that won’t happen.
This season and next are going to be rough, but heading into 19/20 season they will have Dennis Smith, top 5-10 pick and the potential for 2 max players to go with Barnes. That has the making of something.
I was surprised that after 11 games the algorithm found some meaningful stuff to work with.
From a lineup perspective, they are getting much better production out of Powell and Harris. Barnes and Matthews as a 2 man lineup was like -13 per game. Matthews is having a terrible start to the season.
Without any analysis and purely an opinion. Ceiling for Barnes feels like the 3rd best player on a championship calibre team.
In terms of machine learning for tanking, no haven’t looked at that. I think you will find players never tank, the way to tank is basically by purposefully having a terrible roster and keeping guys from suitIng up. A coach is never going to tank either, they are going to put the best lineups they feel will win the game. Front offices do the tanking.
Take the mavs for example. No way carisle is tanking and the players aren’t either, the front office though has made a clear decision to pay the minimum cap, maintain cap flexibility in future years with the obvious outcome being losses in the short term. That’s how tanking happens.
In the short term, I will publish a few more machine learning scouts. I have the process down to just under 10 minutes to collect, format and run the algorithm which gives the high level result. It then takes me approx. 45 minutes to manually find data to backup the high level findings from the agorithm and then write up the scout with details. So all up, I can have a pretty detailed scout based on the algorithms findings in under an hour.
I wonder how that compares to the normal scout coaches/scouts put together?
After a few more team scouts there are a few different machine learning projects I’ll put together and share.
So playing Matthews - Barnes over and over, big, awful minutes is trying to win but just consistently wrong? Is Carlisle ignoring the data / believing otherwise or oblivious? Has any Dallas or national media asked him?
If minutes truly "have to earned" and team results and / or RPM matter, Barnes should be nailed to bench and Matthews-Barnes should never be seen again. They have fallen to -19 per 100 possessions. If it is not tanking, it is extremely unsuccessful coaching. Expected W-L is 4 wins, they have 2 wins. Barnes' top 10 player pairs are all negative, most as horrible as with Matthews. His best pairs are with bench players who aren't the future and he still loses.
He or others might say "no choice not to play him" but that isn't true. What could perform worse? Try anything else before resignation to this? Pushing for a trade or wishing and hoping someday it changes?
2 games ago M-B played 26 minutes, was -8 and was the Mavs 40th best pair that night. Last night was almost identical in minutes and results. Is he going to keep doing this all season?
Totally agree with the Matthews/Barnes combo based on lineup +\- data. I looked at it when diving deeper Into the scout.
They were a positive last season (although only slight if I recall), so perhaps carisle/mavs believe it has worked with some success in the past and giving it every opportunity. Agree though, it can’t last much longer.
They will never be able to trade Matthews (contract is far to unfavourable) and would have to attach a 1st rounder in a salary dump trade. The first round pick is never going to be traded. So they are stuck with him.
He’s earning 18m this year, tough pill to swallow to banish him. I get why they are giving him every opportunity right now.
)