2. Data Acquisition and Resilience

The foundation of any robust model is a reliable data source. Following the unexpected shutdown of advanced data feeds on FBref in January 2026, this project successfully pivoted its data infrastructure to the American Soccer Analysis (ASA) API, demonstrating data resilience and adaptability—key traits for a modern data scientist.

# --- 2.1 Data Extraction (ASA API - Final Verified Version) ---
library(itscalledsoccer)
library(tidyverse)

# 1. Initialize the ASA API client
asa_client <- AmericanSoccerAnalysis$new()

# 2. Extract 2025 MLS Match Data
mls_match_data_raw <- asa_client$get_game_xgoals(leagues = "mls")

##   INFO: New data found. Clearing session cache and refreshing `AmericanSoccerAnalysis` class.

# 3. Filter for 2025 and Clean using the correct column names
mls_match_data <- mls_match_data_raw %>%
  # Use the correct date column name
  filter(grepl("2025", as.character(date_time_utc))) %>% 
  select(
    Date = date_time_utc,
    Home_ID = home_team_id,
    Away_ID = away_team_id,
    Home_Goals = home_goals,
    Away_Goals = away_goals,
    Home_xG = home_team_xgoals,
    Away_xG = away_team_xgoals,
    Home_xPts = home_xpoints,
    Away_xPts = away_xpoints
  )

# 4. Map Team IDs to Names (Optional but recommended for clarity)
# ASA usually provides a separate function for team names if needed:
# teams <- asa_client$get_teams(leagues = "mls")
# mls_match_data <- mls_match_data %>% 
#   left_join(teams %>% select(team_id, Home_Name = team_name), by = c("Home_ID" = "team_id"))

print(paste("Successfully extracted", nrow(mls_match_data), "matches with xPts!"))

## [1] "Successfully extracted 540 matches with xPts!"

head(mls_match_data)

# --- 2.2 Mapping Team IDs to Names ---

# 1. Pull the team list from ASA
teams_raw <- asa_client$get_teams(leagues = "mls")

# 2. Create a clean mapping table
team_map <- teams_raw %>%
  select(team_id, team_name)

# 3. Join names to your match data
mls_match_data_final <- mls_match_data %>%
  left_join(team_map, by = c("Home_ID" = "team_id")) %>%
  rename(Home_Team = team_name) %>%
  left_join(team_map, by = c("Away_ID" = "team_id")) %>%
  rename(Away_Team = team_name)

# 4. Aggregate the Total 2025 xPts per Team
# This is the '60%' foundation of your 2026 prediction!
total_xpts_2025 <- mls_match_data_final %>%
  pivot_longer(cols = c(Home_Team, Away_Team), names_to = "Location", values_to = "Team") %>%
  mutate(
    xPts_Earned = case_when(
      Location == "Home_Team" ~ Home_xPts,
      Location == "Away_Team" ~ Away_xPts
    )
  ) %>%
  group_by(Team) %>%
  summarise(
    Total_xPts_2025 = sum(xPts_Earned, na.rm = TRUE),
    .groups = 'drop'
  )

print("Total xPts for 2025 calculated successfully!")

## [1] "Total xPts for 2025 calculated successfully!"

head(total_xpts_2025 %>% arrange(desc(Total_xPts_2025)))

# --- 4. Final 2026 Prediction Model ---

# 1. Create a table with your Verified 2025 Actual Points
# (I've put the top teams here as an example; add the rest of your verified list!)
actual_standings_2025 <- tibble(
  Team = c("Philadelphia Union", "FC Cincinnati", "Inter Miami CF", "San Diego FC", "Vancouver Whitecaps FC", "Los Angeles FC"),
  Actual_Points = c(66, 65, 65, 63, 63, 60),
  Actual_Rank = c(1, 2, 3, 1, 2, 3) # Note: San Diego/Vancouver/LAFC are West
)

# 2. Add your Qualitative Adjustments (Roster and Schedule)
# Scale: -5 to +5 for Roster, -2 to +2 for Schedule
qualitative_adjustments <- tibble(
  Team = c("Philadelphia Union", "FC Cincinnati", "Inter Miami CF", "San Diego FC", "Vancouver Whitecaps FC", "Los Angeles FC"),
  Roster_Adj = c(-1, 0, 4, 1, 2, 1), # e.g., Miami +4 for roster strength
  SoS_Adj = c(0, 0, 1, 0, 0, -1)     # e.g., LAFC -1 for hard schedule
)

# 3. Combine and Calculate 2026 Predicted Points
final_2026_prediction <- total_xpts_2025 %>%
  inner_join(actual_standings_2025, by = "Team") %>%
  inner_join(qualitative_adjustments, by = "Team") %>%
  mutate(
    xPts_Delta = Actual_Points - Total_xPts_2025,
    # The 60/30/10 Formula:
    # 60% weight on xPts, 30% on Actual, plus Roster (x2) and SoS (x1) points
    Predicted_Points_2026 = (Total_xPts_2025 * 0.6) + 
                            (Actual_Points * 0.3) + 
                            (Roster_Adj * 2) + 
                            (SoS_Adj * 1)
  ) %>%
  arrange(desc(Predicted_Points_2026)) %>%
  mutate(Predicted_Rank_2026 = row_number())

# 4. View your Portfolio-Ready Table
final_2026_prediction %>% 
  select(Team, Actual_Points, Total_xPts_2025, xPts_Delta, Predicted_Points_2026, Predicted_Rank_2026) %>%
  gt() %>%
  tab_header(title = "MLS 2026: Predicted Performance Model") %>%
  fmt_number(columns = c(Total_xPts_2025, xPts_Delta, Predicted_Points_2026), decimals = 2)

Team	Actual_Points	Total_xPts_2025	xPts_Delta	Predicted_Points_2026	Predicted_Rank_2026
MLS 2026: Predicted Performance Model
Inter Miami CF	65	57.60	7.40	63.06	1
Vancouver Whitecaps FC	63	61.05	1.95	59.53	2
Los Angeles FC	60	63.59	−3.59	57.15	3
Philadelphia Union	66	60.04	5.96	53.83	4
San Diego FC	63	52.67	10.33	52.50	5
FC Cincinnati	65	42.48	22.52	44.99	6

# --- FINAL 30-TEAM INTEGRATION ---

# 1. Get the full team list from ASA for all 30 teams
all_teams <- asa_client$get_teams(leagues = "mls") %>%
  select(Team = team_name)

# 2. Your Verified 2025 Actual Standings (Full 30 Teams)
# I have populated this based on the official standings you provided
full_actual_2025 <- tibble(
  Team = c("Philadelphia Union", "FC Cincinnati", "Inter Miami CF", "Charlotte FC", "New York City FC", 
           "Nashville SC", "Columbus Crew", "Chicago Fire FC", "Orlando City SC", "New York Red Bulls", 
           "New England Revolution", "Toronto FC", "CF Montréal", "Atlanta United FC", "D.C. United",
           "San Diego FC", "Vancouver Whitecaps FC", "Los Angeles FC", "Minnesota United FC", "Seattle Sounders FC", 
           "Austin FC", "FC Dallas", "Portland Timbers", "Real Salt Lake", "San Jose Earthquakes", 
           "Colorado Rapids", "Houston Dynamo FC", "St. Louis City SC", "LA Galaxy", "Sporting Kansas City"),
  Actual_Points = c(66, 65, 65, 59, 56, 54, 54, 53, 53, 43, 36, 32, 28, 28, 26, 
                    63, 63, 60, 58, 55, 47, 44, 44, 41, 41, 41, 37, 32, 30, 28),
  Conference = c(rep("East", 15), rep("West", 15))
)

# 3. Apply the 2026 Model to the Full League
# Note: I'm setting Roster/SoS to 0 for teams we haven't 'scouted' yet; 
# you can manually update these in your script!
final_30_team_prediction <- total_xpts_2025 %>%
  inner_join(full_actual_2025, by = "Team") %>%
  mutate(
    xPts_Delta = Actual_Points - Total_xPts_2025,
    # Standard prediction baseline (can be adjusted with Roster/SoS)
    Predicted_Points_2026 = (Total_xPts_2025 * 0.6) + (Actual_Points * 0.4)
  ) %>%
  group_by(Conference) %>%
  arrange(desc(Predicted_Points_2026)) %>%
  mutate(Predicted_Rank_2026 = row_number()) %>%
  ungroup()

# 4. Generate the Final Master Table
final_30_team_prediction %>%
  select(Conference, Predicted_Rank_2026, Team, Actual_Points, Total_xPts_2025, xPts_Delta, Predicted_Points_2026) %>%
  arrange(Conference, Predicted_Rank_2026) %>%
  gt() %>%
  tab_header(title = "MLS 2026: Complete League Projections",
             subtitle = "Based on 2025 xPts Performance & Historical Regression") %>%
  fmt_number(columns = c(Total_xPts_2025, xPts_Delta, Predicted_Points_2026), decimals = 2)

Conference	Predicted_Rank_2026	Team	Actual_Points	Total_xPts_2025	xPts_Delta	Predicted_Points_2026
MLS 2026: Complete League Projections
Based on 2025 xPts Performance & Historical Regression
East	1	Philadelphia Union	66	60.04	5.96	62.43
East	2	Inter Miami CF	65	57.60	7.40	60.56
East	3	Nashville SC	54	59.53	−5.53	57.32
East	4	Orlando City SC	53	57.91	−4.91	55.94
East	5	Columbus Crew	54	51.43	2.57	52.46
East	6	Chicago Fire FC	53	50.71	2.29	51.63
East	7	New York City FC	56	48.54	7.46	51.53
East	8	FC Cincinnati	65	42.48	22.52	51.49
East	9	Charlotte FC	59	43.02	15.98	49.41
East	10	New York Red Bulls	43	43.00	0.00	43.00
East	11	New England Revolution	36	37.92	−1.92	37.15
East	12	Atlanta United FC	28	43.15	−15.15	37.09
East	13	Toronto FC	32	36.91	−4.91	34.95
East	14	CF Montréal	28	39.48	−11.48	34.89
East	15	D.C. United	26	40.49	−14.49	34.69
West	1	Los Angeles FC	60	63.59	−3.59	62.15
West	2	Vancouver Whitecaps FC	63	61.05	1.95	61.83
West	3	San Diego FC	63	52.67	10.33	56.80
West	4	Seattle Sounders FC	55	54.91	0.09	54.95
West	5	Minnesota United FC	58	48.71	9.29	52.43
West	6	San Jose Earthquakes	41	53.30	−12.30	48.38
West	7	Austin FC	47	42.89	4.11	44.53
West	8	FC Dallas	44	43.34	0.66	43.60
West	9	Colorado Rapids	41	44.85	−3.85	43.31
West	10	Real Salt Lake	41	41.66	−0.66	41.40
West	11	Houston Dynamo FC	37	44.14	−7.14	41.29
West	12	St. Louis City SC	32	42.99	−10.99	38.59
West	13	LA Galaxy	30	41.23	−11.23	36.74
West	14	Sporting Kansas City	28	31.96	−3.96	30.38

# --- FINAL 30-TEAM FIX ---

# 1. Correct the name for Portland in your verified standings
full_actual_2025 <- full_actual_2025 %>%
  mutate(Team = if_else(Team == "Portland Timbers", "Portland Timbers FC", Team))

# 2. Re-run the join and calculation
final_30_team_prediction <- total_xpts_2025 %>%
  inner_join(full_actual_2025, by = "Team") %>%
  mutate(
    xPts_Delta = Actual_Points - Total_xPts_2025,
    Predicted_Points_2026 = (Total_xPts_2025 * 0.6) + (Actual_Points * 0.4)
  ) %>%
  group_by(Conference) %>%
  arrange(desc(Predicted_Points_2026)) %>%
  mutate(Predicted_Rank_2026 = row_number()) %>%
  ungroup()

# 3. Verify you have 30 teams
print(paste("Final Team Count:", nrow(final_30_team_prediction)))

## [1] "Final Team Count: 30"

# 4. View the complete, professional table
final_30_team_prediction %>%
  select(Conference, Predicted_Rank_2026, Team, Actual_Points, Total_xPts_2025, xPts_Delta, Predicted_Points_2026) %>%
  arrange(Conference, Predicted_Rank_2026) %>%
  gt() %>%
  tab_header(title = "MLS 2026: Complete League Projections",
             subtitle = "The 'inData' Model: Regression Analysis & Performance Baseline") %>%
  fmt_number(columns = c(Total_xPts_2025, xPts_Delta, Predicted_Points_2026), decimals = 2)

Conference	Predicted_Rank_2026	Team	Actual_Points	Total_xPts_2025	xPts_Delta	Predicted_Points_2026
MLS 2026: Complete League Projections
The 'inData' Model: Regression Analysis & Performance Baseline
East	1	Philadelphia Union	66	60.04	5.96	62.43
East	2	Inter Miami CF	65	57.60	7.40	60.56
East	3	Nashville SC	54	59.53	−5.53	57.32
East	4	Orlando City SC	53	57.91	−4.91	55.94
East	5	Columbus Crew	54	51.43	2.57	52.46
East	6	Chicago Fire FC	53	50.71	2.29	51.63
East	7	New York City FC	56	48.54	7.46	51.53
East	8	FC Cincinnati	65	42.48	22.52	51.49
East	9	Charlotte FC	59	43.02	15.98	49.41
East	10	New York Red Bulls	43	43.00	0.00	43.00
East	11	New England Revolution	36	37.92	−1.92	37.15
East	12	Atlanta United FC	28	43.15	−15.15	37.09
East	13	Toronto FC	32	36.91	−4.91	34.95
East	14	CF Montréal	28	39.48	−11.48	34.89
East	15	D.C. United	26	40.49	−14.49	34.69
West	1	Los Angeles FC	60	63.59	−3.59	62.15
West	2	Vancouver Whitecaps FC	63	61.05	1.95	61.83
West	3	San Diego FC	63	52.67	10.33	56.80
West	4	Seattle Sounders FC	55	54.91	0.09	54.95
West	5	Minnesota United FC	58	48.71	9.29	52.43
West	6	San Jose Earthquakes	41	53.30	−12.30	48.38
West	7	Austin FC	47	42.89	4.11	44.53
West	8	FC Dallas	44	43.34	0.66	43.60
West	9	Colorado Rapids	41	44.85	−3.85	43.31
West	10	Real Salt Lake	41	41.66	−0.66	41.40
West	11	Houston Dynamo FC	37	44.14	−7.14	41.29
West	12	Portland Timbers FC	44	38.12	5.88	40.47
West	13	St. Louis City SC	32	42.99	−10.99	38.59
West	14	LA Galaxy	30	41.23	−11.23	36.74
West	15	Sporting Kansas City	28	31.96	−3.96	30.38

MLS 2026 Regular Season Prediction

Luigi Demasi

2026-01-30

1. Introduction: The ‘inData’ Philosophy

2. Data Acquisition and Resilience