Hey everyone! I decided to make a quick run out from my hiatus to talk about tanking. Everyone is doing it, so I thought it was worth taking a quick look at the issue. It’s been all over the TrueHoop blog recently, it was kind of discussed at Wages of Wins, and Patrick at NBA Geek just covered it today. The WoW and Geek posts are interesting in that they imply that tanking is a bad idea. This is a bit in contrast to a post I had a while back which suggests that tanking is a good idea if your goal is to make the playoffs. So let’s take another look.
Let’s start with what the discussion has been about. I think the driving force behind the TrueHoop perspective that tanking is bad is that a) it’s kind of dishonest (teams are supposed to try to win) and b) it means that fans have to watch crappy basketball. It’s hard to argue with either of those. Tanking is probably not good for fans, and may rub people the wrong way (although Henry says that as a Blazers fan, he’s happy they’re tanking since they won’t win this year and it should make them better next year).
Over at WoW, the question was how to become a contending team, meaning a team that wins 55 or more games. The answer appears to be… be a good team already! Most high-quality teams were high-quality in the preceding years as well, with 70% or so winning at least 40 games in each of the prior 4 seasons. The implication is that tanking isn’t going to make you a contender, because contenders have very rarely been bad in their recent past. However, this doesn’t quite look at tanking directly; it looks at really good teams.
Patrick’s article tackles a few issues. One is that winning brings in money, and so tanking is a bad thing to do. Hard to argue with that, but tanking isn’t really about now but about the future. If you don’t think your team is going to make the playoffs, or maybe will but then lose in the first round, the question is if it’s worth giving up some money this year to tank and then have a better team next year and maybe make that money back. So I’m not sure how that one plays out. The second issue is that tanking doesn’t guarantee that you’ll become a contender, and Patrick has a table that seems damning – only 14% of teams that win less than 20 games end up winning 50 or more within the next decade, and the number is only 22% for teams that win 20 to 24.
Between the two, these articles suggest that if you want to be great, you should start by being good. But how to you get good? My article, which looked at what kinds of teams make the playoffs, suggests that you should either already be good… or you should tank! Really crappy teams tend to make it back to the playoffs more quickly than teams that are better but not playoff quality. So there’s a bit of a disconnect.
I thought the best way to look at tanking would be to look at teams that have actually tanked. To do this, I went to bball-reference and found every team since the start of the draft lottery that had a win percentage under 31%, which basically means 25 wins or less. Of course, not every team that won less than 25 games was tanking. But, they ended up with the potential benefit of tanking: a shot at a good draft pick. Then I looked at how many wins these teams got the next year, the year after that, and so on out to 4 years later. I adjusted any numbers that happened to fall during the lockout to an 82 game season, and I dumped the teams from last year that qualified since they have no future seasons yet to contribute. That left about 100 teams with at least one future season. As a note of potential interest, the only franchises to never qualify were the Lakers, Suns, Jazz, and Rockets. The teams that showed up the most are the Clippers (9), Wolves (8), and Washington (7).
Alright, so what happens after you suck? I ran a regression of year+1 wins predicted from crappy year wins, and the result was an equation of 14.6+.67*wins, with both the slope and intercept significant. The range in wins is 11 to 25, so this predicts that after being sufficiently bad a team can expect to win between 22 and 31 games the next season. That isn’t too bad. The range of actual wins in year+1 is 11 to 66, and the average number of wins increased from 20 to 28, so teams definitely did improve.
I ran the same regression for two years later and the result was 22.6+.49*wins, although wins wasn’t significant. So the equation predicts anywhere from 28 to 35 wins, which is more than year+1, but the non-significant slope makes things a little hazy. But the range of wins in year+2 is 13 to 62 and the mean number is 32, again compared to the crappy year range of 11 to 25 and mean of 20. So teams have improved compared to their bad season, and probably improved a little compared to year+1.
For year+3 the equation is 28.4+.4*wins with the slope non-significant, making a prediction of 33 to 38 wins. The actual range is 13 to 58 wins and the mean is 36. Again the predictions and the actual mean have increased. For year+4 the equation is 24.7+.7*wins, with the slope significant, making a prediction of 32 to 42 wins. The actual range is 13 to 60 wins and the mean number of wins is 39. So again the predictions go up, as do the actual average number of wins.
The number of wins in the crappy season was only a significant predictor of future wins in two cases, but it was always positive. This suggests that if you tank, you don’t want to go all the way; a 20 win team will end up better than a 15 win team. To check on that I split the teams into two nearly-even groups by wins, so one group was the worst 51 teams and the other was the ‘better’ 52 teams. In their crappy season, the first group averaged 17 wins while the second group averaged 23 wins. The second group does do better on average in each of the next 4 years, but their 6 win advantage shrinks to 2.5 by year 4 (37.5 versus 40 wins, not a significant difference). So super-bad teams might end up a little worse than only-bad teams, but they catch up relative to how bad they started.
Let’s sum up: teams who are pretty bad, winning less than 25 games in a season, get better the next year. On average they move up from 20 wins to 28. The year after that they get to 32 wins, then 36, then 39. That obviously isn’t contender level, or even quite playoff-worthy on average (maybe in the East). But it gets you on the treadmill of mediocrity that WoW advises, or at least a move away. The worst teams in this group tend to continue being a little worse than the better teams of the group, but the difference shrinks to the point of non-statistical-significance within two years. So being ridiculously bad isn’t really any worse than being bad. This is probably a reflection of the lottery system (the worst team has typically not ended up with the best pick) and the inaccuracy of draft projections (even when they end up in the right place, teams may end up with a non-superstar player).
So let’s say you’re a borderline playoff team or on the outside looking in and you’re deciding what to do. You could ship off your good, presumably expensive players and start tanking. You’ll probably bring in less money this year while you lose, but you’ll also be paying less if you didn’t do anything stupid with your contracts. You’ll win more games next year, and your costs will also probably still be low because you’ll have a rookie getting paid rookie scale instead of the guys you got rid of. At worst you should have more money to toss at another free agent if someone becomes available. And that will continue being true the next three years; your team will tend to get better and better until you end up about back where you were before or maybe even better (44% of these teams win 41 or more games in year 4). And 10% of the time you’ll be a top contender (50 or more wins) by that third year. Alternatively, you could try to swing a trade or pick up a big free agent, but that’s unlikely to work unless it happens to be LeBron’s Decision. And given how most NBA teams spend, you’re probably just going to fill your cap for the foreseeable future for a soon-to-be-washed-up veteran. Tanking doesn’t sound so bad, does it?