Tanking means deliberately losing games to improve a team’s chance of selecting near the top of the next NBA draft. Because low-win teams receive the highest selections, losing can become a rebuilding strategy rather than merely a competitive failure. section 3.0 section 3.1
The strategy targets potential superstar talent: analysis of first-round selections from 2000–2009 finds substantially stronger career indicators in the 1–5 and 6–10 pick segments than later segments. section 3.5.2 section 3.5.4
The Philadelphia 76ers rebuild
From 2013 through 2016, Philadelphia won 47 regular-season games while moving toward the top of the draft; afterward, excluding the shortened 2019–20 season, the team averaged 51 wins per season after selecting Joel Embiid and others. section 3.1
sumWS48 <- sum(draft2$WS48)
draft2 %>%
group_by(Pk2) %>%
summarize(mean = mean(WS48),
median = median(WS48),
pct = sum(WS48)/sumWS48) -> tibble4
ws3 <- ggplot(tibble4, aes(x = Pk2, y = mean)) +
geom_bar(stat = "identity", width = .8,
fill = "gold", color = "gold4") +
labs(title = "Average Win Shares per 48 Minutes",
subtitle = "First-Round Selections between
2000 and 2009 NBA Drafts",
x = "Segment",
y = "Average Win Shares per 48 Minutes",
caption = "regular season games only") +
scale_x_discrete(limits = c("1-5", "6-10", "11-15",
"16-20", "21-25", "26-30"),
labels = c("1-5", "6-10", "11-15",
"16-20", "21-25", "26-30")) +
geom_text(aes(label = round(mean, 2), vjust = -0.3)) +
geom_label(aes(label = trunc(pct*100), vjust = 1.2)) +
ylim(0, 0.13) +
theme(plot.title = element_text(face = "bold"))
ws4 <- ggplot(tibble4, aes(x = Pk2, y = median)) +
geom_bar(stat = "identity", width = .8,
fill = "gold3", color = "gold4") +
labs(title = "Median Win Shares per 48 Minutes",
subtitle = "First-Round Selections between
2000 and 2009 NBA Drafts",
x = "Segment", y = "Median Win Shares per 48 Minutes",
caption = "regular season games only") +
scale_x_discrete(limits = c("1-5", "6-10", "11-15",
"16-20", "21-25", "26-30"),
labels = c("1-5", "6-10", "11-15",
"16-20", "21-25", "26-30")) +
geom_text(aes(label = round(median, 2), vjust = -0.3)) +
geom_label(aes(label = trunc(pct*100), vjust = 1.2)) +
ylim(0, 0.13) +
theme(plot.title = element_text(face = "bold"))
ws3 + ws4 + plot_layout(ncol = 1)
Aggregate win shares per 48 minutes by draft segment, then compare means, medians, and segment contributions with a ggplot bar chart. section 3.5.4
The case is conditional, not universal: tanking appears rational within the 2000–2009 draft and career data, while lottery randomness reduces guaranteed outcomes and top-five picks do not ensure championships. section 3.1 section 3.6
