What are the effects of tanking?

[kucoach7]

Nathan Sliver here again with sevenfortyseven.com. After a grueling and riveting league-wide conversation on tanking that ended as all league debates end, a lot of hurt feelings and funny troll posts by NOLA and Loco, I got to thinking, 'What does the data say about tanking?' Well, who better to answer that than me.

My first question was simple. How quickly can a team recover from tanking? I took two approaches to the issue. In my first graph I plot the average winning percentage of teams that finished with the worst 10 records in the league for 6 years after their bottom 10 finish. This does not take into account draft position. A tanking team has much less control over this, they only lower their winning percentage to give themselves the best chance at a good draft pick.

Tank All Top 10.GIF (22.92 KiB) Viewed 918 times

I notice a few things with this graph.
1: The teams with the top two picks tend to do significantly better than the rest of the bunch, although the worst record teams tend to tale off after peaking in the fourth year. This suggests that GMs that have had the worst record may only be able to maintain a winning team for a short while.
2: Picks 3-7 all hang together very tightly. They don't improve very much because they aren't bad enough to get the help they need from the draft but they still rely on the draft, thinking that they will get the help they need.
3: Picks 8 and 10 actually get worse. These teams really don't get much help in the draft but they sure seem to be counting on it so they actually have to get worse and get better draft picks before they can get better.

The second graph only looks at teams that go from not being a bottom 10 pick to being a bottom ten pick, meaning that repeat "offenders" don't get recounted. This is valuable because is truly measures what happens to teams that sell considerable assets and how long it takes to recover, regardless of how many times they have to go through the lottery.

Tank 1st Timers.GIF (23.12 KiB) Viewed 918 times

As you can see, the lines here are much flatter. The first and second picks still gain a significant advantage but for everyone else growth will be slow. What does this mean? Tanking is high risk high reward. If you can manage to score one of the top two picks, you'll probably be good soon. If you don't manage one of the top two picks, there will probably be ping pong balls with your name on them the next year too.

Now let's move on the the league-wide effects.

[kucoach7]

The debate on tanking has centered on the fun of the league, which (based on the idea that tanking limits fun) can be measured based on the competitiveness of the league. I look at three different measures of competitiveness to examine the relationship between how bad the worst teams are and how much fun there is to be had.

The most likely culprits behind the complaining about tanking are probably those lower seed playoff teams that have worked hard to build a solid squad and feel like their efforts are being squandered by teams with absurdly good rosters built by tank trades. I have plotted the average winning percentage of the bottom 5 teams against the difference between the average winning percentage of the top 8 seeds vs that of the bottom 8 seeds.

Tank Playoff Seeds.GIF (11.87 KiB) Viewed 912 times

As you can see there is essentially no evidence that in years where tanking is the worst, that the playoffs become overly top heavy. But seeds don't really matter, do they. It is all about who wins it all. What if the likelihood of a top seed winning it all increases as tanking gets worse? Well lets look at this plot.

Tank champ seed.GIF (12.06 KiB) Viewed 912 times

As you can see, 1 seeds win a lot, regardless of how bad the worst teams were. But what if not all one seeds are created equal. Well, let's look at the average winning percentage of the top 5 teams in the league vs. that of the bottom 5.

Tank Top and Bottom.GIF (12.26 KiB) Viewed 912 times

While this one provides the greatest evidence that tanking hurts competitiveness in the league it still isn't much.

So, there you have some pretty pictures and numbers. What do they mean? That is open for interpretation but I think that tanking probably doesn't hurt the competitiveness of the league and that tanking is super risky.

[42PhD]

Good work. The data is very fuzzy, so anything is subject to "But but but" from any armchair statistician, but I can tell you that I appreciate this.

Here's what I see that is more relevant that you have not addressed in your text from the first post:

-- The improvement is improvement to like 50 wins, maybe a few more. The teams do not become dominant.
-- The improvement, if it appears, is temporary, most of those teams are losers again after the rookie contracts end.
-- Many of the tankers just never end up getting out of it.

Punchline: Many more people tank to make it worthwhile for all of them, but they don't understand that. (maybe that helps).

Tip: Specifics matter. Improvement from loser to loser only matters if it is part of a systematic improvement from loser to . . . . to winner. Improvement only matters if people who when to expect it. Season 1, suck . . . . Season 5 (4th after draft), finally good . . . Season Six, suck. Are people this patient? The data says, "no," and what people CHOOSE is more important here than anything.



Here's what I see that is more relevant that you have not addressed in your text from the second post:

- The bottom graph is 100% driven by the extreme data point, nothing to see here. The extreme data point in the upper right of the "blob" is a huge counter example to the inferred trend, as well.

Tips: You may want to do a simple binning presentation. Show how many teams in teach conference fall within some "win bands" I think what you will find that is with a dominant team, you'll see the overall winning percentage drop in that conference the next season.. It may not sustain, but I would be that is what you'll see. What that would likely say is that GM's just give up.




I think trying to lose is an artifact of trying to find a way to win in the presence of dominance and not knowing how to build a team. It's tough (at least for me), and it's not clear to me that I'm the only one who feels the pain . . . just maybe the only one dumb enough to say it.

[kucoach7]

Good work. The data is very fuzzy, so anything is subject to "But but but" from any armchair statistician, but I can tell you that I appreciate this.

Here's what I see that is more relevant that you have not addressed in your text from the first post:

-- The improvement is improvement to like 50 wins, maybe a few more. The teams do not become dominant.
-- The improvement, if it appears, is temporary, most of those teams are losers again after the rookie contracts end.
-- Many of the tankers just never end up getting out of it.

Punchline: Many more people tank to make it worthwhile for all of them, but they don't understand that. (maybe that helps).

Tip: Specifics matter. Improvement from loser to loser only matters if it is part of a systematic improvement from loser to . . . . to winner. Improvement only matters if people who when to expect it. Season 1, suck . . . . Season 5 (4th after draft), finally good . . . Season Six, suck. Are people this patient? The data says, "no," and what people CHOOSE is more important here than anything.



Here's what I see that is more relevant that you have not addressed in your text from the second post:

- The bottom graph is 100% driven by the extreme data point, nothing to see here. The extreme data point in the upper right of the "blob" is a huge counter example to the inferred trend, as well.

Tips: You may want to do a simple binning presentation. Show how many teams in teach conference fall within some "win bands" I think what you will find that is with a dominant team, you'll see the overall winning percentage drop in that conference the next season.. It may not sustain, but I would be that is what you'll see. What that would likely say is that GM's just give up.




I think trying to lose is an artifact of trying to find a way to win in the presence of dominance and not knowing how to build a team. It's tough (at least for me), and it's not clear to me that I'm the only one who feels the pain . . . just maybe the only one dumb enough to say it.

Thanks for the input. You bring up some great points and I totally agree. My second post really isn't great because I have 12 data points so I appreciate the suggestion.

As for struggling to build a good team, I'm right there with you. Things are good in Portland right now but I had to lose four years in a row and acquire a very good pick for the fifth year of that sequence to get here and how long can I sustain it?

[42PhD]

Good work. The data is very fuzzy, so anything is subject to "But but but" from any armchair statistician, but I can tell you that I appreciate this.

Here's what I see that is more relevant that you have not addressed in your text from the first post:

-- The improvement is improvement to like 50 wins, maybe a few more. The teams do not become dominant.
-- The improvement, if it appears, is temporary, most of those teams are losers again after the rookie contracts end.
-- Many of the tankers just never end up getting out of it.

Punchline: Many more people tank to make it worthwhile for all of them, but they don't understand that. (maybe that helps).

Tip: Specifics matter. Improvement from loser to loser only matters if it is part of a systematic improvement from loser to . . . . to winner. Improvement only matters if people who when to expect it. Season 1, suck . . . . Season 5 (4th after draft), finally good . . . Season Six, suck. Are people this patient? The data says, "no," and what people CHOOSE is more important here than anything.



Here's what I see that is more relevant that you have not addressed in your text from the second post:

- The bottom graph is 100% driven by the extreme data point, nothing to see here. The extreme data point in the upper right of the "blob" is a huge counter example to the inferred trend, as well.

Tips: You may want to do a simple binning presentation. Show how many teams in teach conference fall within some "win bands" I think what you will find that is with a dominant team, you'll see the overall winning percentage drop in that conference the next season.. It may not sustain, but I would be that is what you'll see. What that would likely say is that GM's just give up.




I think trying to lose is an artifact of trying to find a way to win in the presence of dominance and not knowing how to build a team. It's tough (at least for me), and it's not clear to me that I'm the only one who feels the pain . . . just maybe the only one dumb enough to say it.

Thanks for the input. You bring up some great points and I totally agree. My second post really isn't great because I have 12 data points so I appreciate the suggestion.

As for struggling to build a good team, I'm right there with you. Things are good in Portland right now but I had to lose four years in a row and acquire a very good pick for the fifth year of that sequence to get here and how long can I sustain it?

Yeah, and not trying to throw stones and your or anyone else's team building. Everyone knows I am a novice coach, maybe slightly better GM, but lack of ability to perform does not mean I can't see things as they are. Lots of bad musicians can tell the difference between good and bad music, they just can't fix theirs. The numbers are there, and it's silly to ignore the message of them. I might be wrong (which means you are), but I'd like to see an alternative explanation that is some kind of observable evidence so bump against the ones listed.

[IamQuailman]

Nerds. lol JK Some interesting information here man!

AWARDED 5 POINTS! THANKS FOR YOUR CONTRIBUTION!