next_gen.Rmd

---
title: "NFL Analytics: Next Gen Stats"
output: html_notebook
---

```{r setup, include=TRUE, warning=FALSE, message=FALSE}
if (!require("pacman")) install.packages("pacman")
pacman::p_load(
    tidyverse,
    nflverse,
    gt,
    gtExtras,
    gtUtils,
    ggfx,
    rvest,
    httr,
    jsonlite,
    patchwork,
    ggtext,
    ggrepel,
    ggimage,
    dlookr,
    showtext,
    sysfonts,
    janitor,
    grid,
    ggiraph,
    glue,
    here
)    

sysfonts::font_add_google("Lexend")
showtext::showtext_auto()
```

```{r nfl_analytics_theme, include=TRUE}
nfl_analytics_theme <- function(..., base_size = 12) {
  theme(
    text = element_text(family = "Lexend", size = base_size, color = "black"),
    axis.ticks = element_blank(),
    axis.title = element_text(face = "bold"),
    axis.text = element_text(face = "bold"),
    plot.title.position = "plot",
    plot.title = element_markdown(size = 16, vjust = .02, hjust = 0.5),
    plot.subtitle = element_markdown(hjust = 0.5),
    plot.caption = element_markdown(size = 8),
    panel.grid.minor = element_blank(),
    panel.grid.major =  element_line(color = "#d0d0d0"),
    panel.background = element_rect(fill = "#f7f7f7"),
    plot.background = element_rect(fill = "#f7f7f7"),
    panel.border = element_blank(),
    legend.background = element_rect(color = "#F7F7F7"),
    legend.key = element_rect(color = "#F7F7F7"),
    legend.title = element_text(face = "bold"),
    legend.title.align = 0.5,
    strip.text = element_text(face = "bold"))
}
```

```{r teams, include=TRUE}
teams <- nflreadr::load_teams()
```

# Passing 

```{r ngs_data_passing, include=TRUE}
ngs_data_passing <- nflreadr::load_nextgen_stats(seasons = 2024,
                                                 stat_type = "passing") %>%
  # week 0 aggregates stats for you
  # it looks like that aggregation requires a minimum number of attempts
  # could always do that aggregation myself instead of waiting for NGS to do it
  filter(week == 0)

names(ngs_data_passing)
```

```{r join_teams, include=TRUE}
ngs_data_passing <- ngs_data_passing %>%
  left_join(teams, by = c("team_abbr" = "team_abbr"))
```

```{r warning=FALSE, message=FALSE}
ngs_data_passing %>% 
  ggplot(aes(x=avg_time_to_throw,
             y=avg_intended_air_yards)) +
  geom_hline(yintercept = mean(ngs_data_passing$avg_intended_air_yards),
             color = "black", size = 0.8, linetype = "dashed") +
  geom_vline(xintercept = mean(ngs_data_passing$avg_time_to_throw),
             color = "black", size = 0.8, linetype = "dashed") +
  geom_point(size = 3.5, color = ngs_data_passing$team_color) +
  scale_x_continuous(breaks = scales::pretty_breaks(),
                     labels = scales::comma_format()) +
  scale_y_continuous(breaks = scales::pretty_breaks(),
                     labels = scales::comma_format()) +
  geom_text_repel(aes(label = player_display_name),
                  family = "Lexend", fontface = "bold", size = 3.5, box.padding = 0.45,) +
  # theme_minimal() + 
  nfl_analytics_theme() +
  xlab("Average Time to Throw") +
  ylab("Average Intended Air Yards") +
  labs(title = "**Avgerage Time to Throw vs. Average Intended Air Yards**")
```

## Pro Football Reference 

```{r pfr_stats_pass, include=TRUE}
pfr_stats_pass <- nflreadr::load_pfr_advstats(seasons = 2024,
                                              stat_type = "pass",
                                              summary_level = "season")

pfr_stats_pass$team <- clean_team_abbrs(pfr_stats_pass$team)

pfr_stats_pass <- pfr_stats_pass %>%
  left_join(teams, by = c("team" = "team_abbr"))

names(pfr_stats_pass)
```

```{r}
pfr_stats_pass_filtered <- pfr_stats_pass %>% 
  filter(pass_attempts >= 25)
```


```{r warning=FALSE, message=FALSE}
pfr_stats_pass_filtered %>% 
  ggplot(aes(x = bad_throw_pct, 
             y = drop_pct,
             alpha = if_else(team == 'PIT', 'PIT', 'Other'))) +
  geom_point(color = pfr_stats_pass_filtered$team_color, size = 3.5) + 
  geom_hline(yintercept = mean(pfr_stats_pass_filtered$drop_pct),
             color = "black", linetype = "dashed", size = 0.8) +
  geom_vline(xintercept = mean(pfr_stats_pass_filtered$bad_throw_pct),
             color = "black", linetype = "dashed", size = 0.8) +
  scale_x_continuous(breaks = scales::pretty_breaks()) +
  scale_y_continuous(breaks = scales::pretty_breaks()) +
  scale_alpha_manual(values = c('PIT' = 1,
                                'Other' = .2)) + 
  nfl_analytics_theme() +
  geom_text_repel(aes(label = player),
                  family = "Lexend", fontface = "bold", size = 3.5) +
  xlab("Bad Throw %") +
  ylab("Drop %") +
  labs(title = "**QB Bad Throws % vs. WR Drops %**",
       subtitle = "2024 Regular Season<br>Justin Fields highlighted",
       caption = "*Source: Pro Football Reference*") + 
  theme(legend.position = 'none')
```

```{r weekly_qbr, include=TRUE}
weekly_qbr <- nflreadr::load_espn_qbr(seasons = most_recent_season(),
                        summary_type = 'week')
```




# Rushing 

```{r ngs_data_rushing, include=TRUE}
ngs_data_rushing <- nflreadr::load_nextgen_stats(seasons = 2024,
                                                 stat_type = "rushing") %>%
  filter(week == 0) %>% 
  mutate(actual_rush_yards = expected_rush_yards + rush_yards_over_expected)

names(ngs_data_rushing)
```

```{r join_teams2, include=TRUE}
ngs_data_rushing <- ngs_data_rushing %>%
  left_join(teams, by = c("team_abbr" = "team_abbr"))
```

```{r}
ngs_data_rushing %>% 
  arrange(-percent_attempts_gte_eight_defenders)
```


```{r warning=FALSE, message=FALSE}
ngs_data_rushing %>% 
  ggplot(aes(x = rush_yards_over_expected_per_att,
             y = actual_rush_yards,
             alpha = if_else(team_abbr == 'PIT', 'PIT', 'Other'))) +
  geom_smooth(method = lm, se = FALSE,
              color = "black",
              size = 0.8,
              linetype = "dashed") +
  geom_point(size = 3.5, color = ngs_data_rushing$team_color) +
  scale_x_continuous(breaks = scales::pretty_breaks(),
                     labels = scales::comma_format()) +
  scale_y_continuous(breaks = scales::pretty_breaks(),
                     labels = scales::comma_format()) +
  geom_text_repel(aes(label = player_display_name),
                  family = "Lexend",
                  fontface = "bold", size = 3.5) +
  scale_alpha_manual(values = c('PIT' = 1,
                                'Other' = .2)) +
  nfl_analytics_theme() +
  xlab("Rush Yards Over Expected per Attempt") +
  ylab("Actual Rush Yards") +
  labs(title = "Rush Yards Over Expected per Attempt vs. Actual Rush Yards",
       subtitle = "2024 Regular Season through Week 7") + 
  theme(legend.position = 'none')
```

## Pro Football Reference 


```{r pfr_stats_rush, include=TRUE}
pfr_stats_rush <- nflreadr::load_pfr_advstats(seasons = 2024,
                                              stat_type = "rush",
                                              summary_level = "season")

pfr_stats_rush$tm <- clean_team_abbrs(pfr_stats_rush$tm)

pfr_stats_rush <- pfr_stats_rush %>%
  left_join(teams, by = c("tm" = "team_abbr"))

names(pfr_stats_rush)
```

```{r pfr_stats_rush_filtered, include=TRUE}
pfr_stats_rush_filtered <- pfr_stats_rush %>% 
  filter(pos == 'RB',
         att >= 40)
```


```{r}
pfr_stats_rush_filtered %>% 
  ggplot(aes(x = ybc, y = yac,
             alpha = if_else(tm == 'PIT', 'PIT', 'Other'))) +
  geom_hline(yintercept = mean(pfr_stats_rush_filtered$yac),
             color = "black", linetype = "dashed", size = 0.8) +
  geom_vline(xintercept = mean(pfr_stats_rush_filtered$ybc),
             color = "black", linetype = "dashed", size = 0.8) +
  geom_point(color = pfr_stats_rush_filtered$team_color, size = 3.5) +
  scale_x_continuous(breaks = scales::pretty_breaks()) +
  scale_y_continuous(breaks = scales::pretty_breaks()) +
  nfl_analytics_theme() +
  geom_text_repel(aes(label = player),
                  family = "Lexend", fontface = "bold", size = 3.5, box.padding = .5) +
  scale_alpha_manual(values = c('PIT' = 1,
                                'Other' = .2)) +
  labs(x='Yards Before Contact',
       y='Yards After Contact',
       title = "**Running Backs: Yards Before and After Contact**<br>Minimum 40 carries",
       subtitle = "2024 Regular Season",
       caption = "*Source: Pro Football Reference*") + 
  theme(legend.position = 'none')
```


```{r}
pfr_stats_rush_filtered %>% 
  ggplot(aes(x = ybc, y = yds)) +
  geom_hline(yintercept = mean(pfr_stats_rush_filtered$yds),
             color = "black", linetype = "dashed", size = 0.8) +
  geom_vline(xintercept = mean(pfr_stats_rush_filtered$ybc),
             color = "black", linetype = "dashed", size = 0.8) +
  geom_point(color = pfr_stats_rush_filtered$team_color, size = 3.5) +
  scale_x_continuous(breaks = scales::pretty_breaks()) +
  scale_y_continuous(breaks = scales::pretty_breaks()) +
  nfl_analytics_theme() +
  geom_text_repel(aes(label = player),
                  family = "Lexend", fontface = "bold", size = 3.5, box.padding = .5) +
  labs(x='Yards Before Contact',
       y='Total Yards',
       title = "**Running Backs: Total Yards vs. Yards Before Contact**<br>Minimum 40 carries",
       subtitle = "2024 Regular Season",
       caption = "*Source: Pro Football Reference*")
```

# Receiving 

```{r pfr_stats_rec, include=TRUE}
pfr_stats_rec <- nflreadr::load_pfr_advstats(seasons = 2024, stat_type = "rec",
                                             summary_level = "season")

pfr_stats_rec <- pfr_stats_rec %>%
  left_join(teams, by = c("tm" = "team_abbr"))

names(pfr_stats_rec)
```

```{r pfr_stats_rec_filtered, include=TRUE}
pfr_stats_rec_filtered <- pfr_stats_rec %>% 
  filter(rec >= 20 | 
         tm == 'PIT' & rec > 5)

pfr_stats_rec_filtered_pit <- pfr_stats_rec %>% 
  filter(tm == 'PIT')
```

The below graphic compares the amount of yards at catch with the yards gained after catch. 

```{r message=FALSE, warning=FALSE, echo=FALSE}
pfr_stats_rec_filtered %>% 
  ggplot(.,
         aes(x=ybc,
             y=yac,
             alpha=if_else(tm == 'PIT', 'PIT', 'Other'))) + 
  geom_point(color = pfr_stats_rec_filtered$team_color, size=3.5) + 
  geom_text_repel(aes(label = player),
                  family='Lexend', fontface='bold', size=3.5, box.padding = .5) + 
  scale_alpha_manual(values = c('PIT' = 1,
                                'Other' = .2)) + 
  labs(x='Yards Before Catch',
       y='Yards After Catch') + 
  nfl_analytics_theme() + 
  theme(legend.position = 'none')
```

Najee Harris has been a weapon (through Week 5) as a pass catcher out of the backfield. His average depth of target has been behind the line of scrimmage, and he's led the league in YAC per reception (the only player to exceed 10 yards after the catch per reception). 

```{r warning=FALSE, message=FALSE}
pfr_stats_rec_filtered %>% 
  ggplot(.,
         aes(x=adot,
             y=yac_r,
             alpha=if_else(tm == 'PIT', 'PIT', 'Other'))) + 
  geom_point(color=pfr_stats_rec_filtered$team_color, size=3.5) + 
  geom_text_repel(aes(label = player),
                  family='Lexend', fontface='bold', size=3.5, box.padding = .5) + 
  labs(x='Average Depth of Target',
       y='Yards After Catch per Reception',
       caption='*Source: Pro Football Reference*<br>Steelers: minimum 5 receptions<br>Rest of League: minimum 20 receptions') + 
  ggtitle('Receiving: Average Depth of Target vs. YAC per Reception',
          subtitle = '2024 Season<br>Highlighting Steelers') + 
  nfl_analytics_theme() + 
  theme(legend.position = 'none')

```

When looking at all Steelers pass catchers, some elements stand out:

*  Pittsburgh seems to use personnel at different levels -- which is probably also common around the league. Backs are typically catching the ball behind the line of scrimmage. Tight ends and slot options are averaging around 5 yards depth of target. Wide outs are averaging a depth of target beyond 10 yards.
*  Connor Heyward is getting some greater depth in his targeting than other TEs (buoyed by his targets against the Cowboys). He's getting some switch and wheel opportunities that Fields has targeted.
*  George Pickens needs to be put in some situations to create after the catch. In the 2023 season, Pickens averaged over 6 yards after catch. So far in 2024, he's averaging less than 3 yards after the catch.

```{r pit_rec_2024_plot}
pit_rec_2024_plot <- pfr_stats_rec_filtered_pit %>% 
  ggplot(.,
         aes(x=adot,
             y=yac_r,
             alpha=rec))+ 
  geom_point(color=pfr_stats_rec_filtered_pit$team_color, size=3.5) + 
  geom_text_repel(aes(label = player),
                  family='Lexend', fontface='bold', size=3.5, box.padding = .5) + 
  labs(x='Average Depth of Target',
       y='Yards After Catch per Reception',
       subtitle = '**2024 Season**', 
       caption='*Source: Pro Football Reference*<br>Color intensity by total receptions') + 
  nfl_analytics_theme() + 
  theme(legend.position = 'none')
```

```{r warning=FALSE, message=FALSE}
pfr_stats_rec_2023 <- nflreadr::load_pfr_advstats(seasons = 2023, stat_type = "rec",
                                             summary_level = "season")
pit_rec_2023_plot <- pfr_stats_rec_2023 %>% 
  filter(tm == 'PIT') %>% 
  ggplot(.,
         aes(x=adot,
             y=yac_r,
             alpha=rec))+ 
  geom_point(color=pfr_stats_rec_filtered_pit$team_color, size=3.5) + 
  geom_text_repel(aes(label = player),
                  family='Lexend', fontface='bold', size=3.5, box.padding = .5) + 
  labs(x='Average Depth of Target',
       y='Yards After Catch per Reception',
       subtitle = '**2023 Season**',
       caption='*Source: Pro Football Reference*<br>Color intensity by total receptions') + 
  nfl_analytics_theme() + 
  theme(legend.position = 'none')
```



```{r warning=FALSE, message=FALSE, fig.width=10}
pit_rec_2023_plot + pit_rec_2024_plot + 
  plot_annotation(title = '**Steelers Receiving**: Average Depth of Target vs. YAC per Reception',
                  theme(plot.title = element_markdown()))
```


```{r pfr_stats_rec_filtered_pit, include=TRUE}
pfr_stats_rec_filtered_pit %>% 
  ggplot(.,
         aes(x=ybc,
             y=yac,
             alpha=rec)) + 
  geom_point(color = 'black', size=3.5) + 
  geom_text_repel(aes(label = player),
                  family='Lexend', fontface='bold', size=3.5, box.padding = .5) + 
  nfl_analytics_theme() + 
  theme(legend.position = 'none')
```



# Defensive stats 

## Pro Football Reference 

```{r pfr_stats_def, include=TRUE}
pfr_stats_def <- nflreadr::load_pfr_advstats(seasons = 2024,
                                             stat_type = "def",
                                             summary_level = "season")

pfr_stats_def <- pfr_stats_def %>%
  left_join(teams, by = c("tm" = "team_abbr"))

names(pfr_stats_def)
```

```{r pfr_stats_def_filtered_cb, include=TRUE}
pfr_stats_def_filtered_cb <- pfr_stats_def %>% 
  filter(pos == 'CB',
         tgt >= 20)
```

```{r}
pfr_stats_def_filtered_cb %>% 
  filter(tm == 'PIT')
```


```{r warning=FALSE, message=FALSE}
pfr_stats_def_filtered_cb %>% 
  ggplot(.,
         aes(x=yds_cmp,
             y=rat,
             alpha=if_else(tm %in% c('PIT', 'WAS'), 'PIT/WAS', 'Other'))) + 
  geom_point(color=pfr_stats_def_filtered_cb$team_color, size=3.5) +
  geom_text_repel(aes(label = player),
                  familiy='Lexend', fontface='bold', size=3.5, box.padding = .5) + 
  labs(y='QB rating allowed',
       x='Yards per completion') + 
  scale_alpha_manual(values = c('PIT/WAS' = 1,
                                'Other' = .2)) + 
  nfl_analytics_theme() + 
  theme(legend.position = 'none')
```



# FTN Charting 

```{r ftn_data, include=TRUE}
ftn_data <- nflreadr::load_ftn_charting(2024)

play_by_play_2024 <- load_pbp(seasons = most_recent_season())
```

```{r join_pbp_ftn, include=TRUE}
play_by_play_2024 <- play_by_play_2024 %>%
  left_join(ftn_data, by = c("game_id" = "nflverse_game_id",
                             "play_id" = "nflverse_play_id"))
```

This is copied from the book:

```{r}
airyards_accu <- play_by_play_2024 %>%
  filter(season_type == "REG") %>%
  filter(complete_pass == 1 | incomplete_pass == 1, !is.na(down)) %>%
  group_by(passer) %>%
  mutate(total_passes = n()) %>%
  filter(total_passes >= 25) %>%
  ungroup() %>%
  group_by(passer, posteam, air_yards) %>%
  summarise(
    total_att = n(),
    total_catchable = sum(is_catchable_ball == "TRUE", na.rm = TRUE),
    total_pct = total_catchable / total_att,
    .groups = 'drop') %>%
  filter(air_yards >= 0 & air_yards <= 25)
```


```{r}
airyards_accu <- airyards_accu %>%
  left_join(teams, by = c("posteam" = "team_abbr"))

airyards_accu <- airyards_accu %>%
  filter(passer %in% c("L.Jackson", "D.Watson", "J.Fields", "J.Burrow", "R.Wilson"))
```

```{r}
ggplot(data = airyards_accu, aes(x = air_yards, y = total_pct)) +
  geom_smooth(se = FALSE, aes(color = posteam)) +
  nflplotR::scale_color_nfl(type = "primary") +
  scale_x_continuous(breaks = scales::pretty_breaks()) +
  scale_y_continuous(breaks = scales::pretty_breaks(),
                     labels = scales::label_percent()) +
  facet_wrap(vars(passer), scales = "free_y") +
  xlab("Air Yards") +
  ylab("Total % of Catchable Attempts") +
  labs(title = "**Total % of Catchable Attempts Based on Air Yards**",
       subtitle = "*Minimum 100 Attempts  |  Through Week 5 of 2023 Season*",
       caption = "*Introduction to NFL Analytics With R*<br>**Bradley J. Congelio**") +
  nfl_analytics_theme()
```