Sr. Manager, Football Ops Data Scientist at @NFL. Alum of @Columbia (Data Science) and @UWMadison (Physics/Astronomy). Born / raised in Oakland, CA 🇬🇾🇺🇸.
New York, NY
- Tweets 1,626
- Following 642
- Followers 7,271
- Likes 6,963
Using linear optimization, here is how the 65 Power 5 teams from last season could be put into 5 conferences (each with 13 teams):
139
193
2,003
884,985
Using linear optimization, here is how NFL divisions could be recreated if they were built to minimize distance between teams:
(due to large computational times required for confirming divisional grouping with lowest distances, this map may not represent global minimum)
97
99
1,227
536,939
Using our framework, 2023 KC had the most win prob added via luck of any playoff team since 2018.
Most significant plays for KC:
- BUF missed FG with 1:47 left (+18.7%)
- BUF dropped 54 yd pass with 8:23 left (+18.3%)
- Recovery of SF muffed punt with 2:42 left in Q3 (+15.6%)
91
164
1,222
808,708
For each NFL team, here are the top 5 colleges in terms of % of snaps played during the 2021 season. Which college does your team have the most snaps from?
(Credit to @JEvanReid for idea)
47
153
765
For each NFL team, here is the % of snaps by the player's team that drafted them or signed them as undrafted FA:
30
107
709
Using our analysis where we estimated value of travel, rest and playing at home while accounting for team strength, here are wins added via matchups and schedule advantages for each team in 2022:
16
54
331
Here is how NFL defensive personnel groupings have changed across seasons.
While the 4-3-4 (4 DL, 3 LBs, 4 DBs) was the most popular grouping in 2007, the 4-2-5 (4 DL, 2 LBs, 5 DBs) was the most popular grouping in 2020. As teams pass more, we see more DBs on the field.
12
71
327
More NFL kickoffs than ever have resulted in a touchback in 2023, giving offenses more of an opportunity to pick where the want to start their possession. Here's what they've selected:
7
22
301
240,438
How does your team's head coach use their timeouts? Which head coach uses their timeouts most efficiently? We explore timeout usage at the #ExtraPoint.
operations.nfl.com/gameday/a…
35
44
247
Many things can affect the outcome of an NFL game - some of which are out of a team's control or based on "luck".
Here we analyze how luck can affect NFL games by looking at events that are purely determined by opponent performance or lucky bounces.
operations.nfl.com/gameday/a…
25
44
250
Using @cfbfastr, @CFB_Data for win probability and Wikipedia for rivalries, here are the excitement scores (calculated by sum(win probability added) over game) for every rivalry 2014-2022.
The common wealth cup has highest median excitement score over this time.
Someone should make boxplots of each NCAA football rivalry's excitement score (sum of win probability changes) since 2000 to identify what is the actual best rivalry.
17
34
247
155,900
The 2022 NFL Schedule has been released. Here are the rest differences experienced by each team ordered by most advantageous to least advantageous. #NFLScheduleRelease
18
54
222
Some personal news, today I was promoted to the manager level at the @NFL league office.
Thanks to @StatsbyLopez and others at the @NFL for helping me get this far! I am excited to continue to do exciting stuff here at the league!
31
9
214
2023 Net WPA by Luck thru week 11:
- GB benefited from five 10% or more WPA dropped passes by LAC
- KC dropped TD with 1:50 left vs PHI highest WPA play of wk 11 (+42.4% WPA for PHI)
- DEN's fumble going out of bounds with 1:30 left vs MIN worth +21.8% for DEN and -21.8% for MIN
25
39
190
345,637
Luck dashboard updated thru week 6:
- LAC leads in WPA and benefitted from highest WPA play of '23 (MIN dropped game winning TD with 7 secs left which was intercepted)
- CLE benefitted from 2nd highest WPA play (missed game winning FG by SF)
18
34
189
306,766
Also fun facts:
- KC is in the AFC West, but east of DAL which is in the NFC East
- IND is in the AFC South, but North of CIN which is in the AFC North
4
8
159
18,877
We are hiring someone to work in Football/Media Data & Analytics at the NFL - please reach out if you are interested
5
15
167
46,709
Net Win Prob Added for Week 13.
Top plays of week 13:
- DAL dropped pass on 4th and 2 with 1:08 left in Q3 (+20.5% for SEA)
- CIN dropped interception which was deflected and caught by JAX for TD with 5:33 left in Q3 (+17.3% for JAX)
23
20
152
556,441
Net Win Probability Added table through week 10.
CIN WR Tyler Boyd dropped TD pass at with 1:42 left in 4th quarter was luckiest play of week 10 (+33.5% WP for HOU)
6
21
134
214,705
Here are unique starting OL combinations for each team during the 2020 season sorted from least combinations (LAR) to most combinations (PHI).
Notice that the entire bottom row (the 8 teams with the most OL combinations) missed the playoffs.
8
25
126
Using our team strengths model and @nflseedR, here are the ten most likely sets of seeds for each conference
8
13
117
62,059
Using @nflseedR , here are the most likely matchups (with no regards to seeds) in this season's wild card round and divisional round:
8
21
119
90,231
Earlier in 2022, we saw a 33-point comeback (largest in NFL history). When is a lead "safe"?
Here is a table with observed win %s of teams with given leads throughout periods of a game as well as a plot that shows the largest points comeback in a given game minute since 2011:
3
20
121
30,271
Dots for Aidan Hutchinson strip sack/safety/celebratory punt.
Launch speed of Hutchinson punt was ~25 MPH.
@StatsbyLopez @DataWithBliss can either of you give me the dots and stats on the Aiden Hutchinson punt?
3
8
118
25,155
With the help of @StatsbyLopez , I am happy to release weather data for every NFL Regular / Post Season game 2000 - 2019.
The raw data can be found here:
github.com/ThompsonJamesBlis…
Documentation on how to acquire the data can be found here:
datawithbliss.com/weather-da…
7
19
114
Using @nflseedR , here is the probability that each game will have a post season impact for each team ordered by most impact to least impact.
Denver has the most on the line while Detroit has the least.
6
21
113
With the release of the 2020 NFL schedule, here are the short rest advantages (1-4 days of rest difference and teams are not both coming off byes) and bye rest advantages (6+ days of rest difference) by team.
3
29
112
With the college football returning, here is how Power 4 conferences would look if they were created purely to minimize the average distance between teams (17 teams per conf).
Avg distance between teams in same conference:
- Current: 787 miles
- Optimal: 479 miles
5
17
115
24,377
I hope everyone is excited as I am for the start of the 2021 #BigDataBowl 😃😃😃!!!!! For those of you who are figuring out how to start, I put together a tutorial on Kaggle and made it public! Enjoy! kaggle.com/tombliss/tutorial
3
10
98
With the growing popularity of "NFL crossover" games, here are the number of players who have played on each pair of teams since 2002 (when HOU became 32nd franchise).
Since 2002:
- 58 players played for both NE and MIA (most)
- 8 players played for both PIT and MIN (least)
6
13
98
33,857
Come work with me!
The NFL Football Data & Analytics team is hiring a summer intern based in NYC. Looking for someone who knows R and/or Python.
(select "Football Data and Analytics (NY)" as your choice of department on the 2nd page of the application)
hdmm.fa.us6.oraclecloud.com/…
3
45
101
Final iteration of Luck table using data thru week 18.
Biggest luck play in week 18 was SEA Missed Field Goal at end of regulation (+28.9% for LAR).
9
20
93
182,655
Using results through week 15, here is the probability that each team will be playing for seeding or playoffs (not locked into seed or mathematically eliminated) entering week 18.
BUF@MIA has an 87% chance to have both teams still playing for something entering the final week.
5
19
87
70,502
WPA added via luck rank of each Super Bowl champion
2018: NE (79)
2019: KC (24)
2020: TB (6)
2021: LAR (68)
2022: KC (7)
2023: KC (1)
4
9
90
27,391
Hello NYC area sports analytics friends,
@StatsbyLopez @Ally_Blake3 and I have organized a sports analytics meet up this Monday Dec 19th at @BierhausNYC at 6pm.
Please come through if you are in the area and want to join us for some good conversation and delicious drinks/food!
9
15
93
A common 4th down strategy is to have offense attempt to draw the defense offside.
Here is the breakdown of 4th & ≤5 plays with:
- An offensive timeout and then punt/FG
- A delay of game/false start and then punt/FG
- A defensive offside
Since '16, 7% have led to first downs.
9
14
90
81,641
Using data thru last night's SNF game, KC's stadium at night remains the #NFL location/time with the highest drop rate of catchable passes.
Chiefs’ Arrowhead Stadium has a weird night-game quirk. Will a recent move change that?
kansascity.com/sports/nfl/ka…
3
8
87
34,602
While everyone is excited about a new college football video game, I am excited about a new college football statistical model!
Using our Bayesian state space model framework on CFB & sports betting data, I estimate magnitudes of various advantages commonly experienced in games.
1
11
91
The 2022 #BigDataBowl is live and the NFL football operations group is excited to provide some demo code analyzing field goal kickers to get you started!
This year we have demo code in both R and Python 🔥🔥🔥
R: kaggle.com/tombliss/analyzin…
🐍: kaggle.com/dhritiyandapally/…
1
17
89
With the MLB All-Star game tonight, here is how MLB divisions would look if they were created purely to minimize the average distance between teams.
Avg distance between teams in same division:
- Current: 606 miles
- Optimal: 421 miles
3
4
89
25,394
This NFL season is the first where teams will play 17 games.
In terms of making the playoffs, what are a teams chances if they start 3-0? What are a team's chances if they start 0-3?
We explore in this new article of the #ExtraPoint.
operations.nfl.com/gameday/a…
3
20
82
KC Receivers (2016 thru 2023 week 12):
Home:
- Early Afternoon: 6.7%
- Late Afternoon: 6.6%
- Night: 9.9%
Away:
- Early Afternoon: 6.8%
- Late Afternoon: 6.1%
- Night: 6.4%
5
11
79
91,340
How did betting markets perceive home-field advantage in 2020? Noticeably lower than any of the past 15 seasons. Check out our newest post in the #ExtraPoint: operations.nfl.com/gameday/a…
2
15
77
Whenever a team fumbles the ball, the recovery rate is ~50% for each team.
By calculating the win prob observed after fumble and subtracting the win prob if the other team recovers (assume 0 yd return) and summing over the season, we get wins added due to fumble recovery luck:
7
4
75
Week 1 and week 2 of 2023 each set records for lowest kickoff return rate since 1991 (and likely NFL history)
7
20
75
39,150
Highest WPA via luck plays from 2018-23 postseason:
- PHI dropped pass which was intercepted with 2 mins left in 2018 (+32.7% WP for NO)
- NE dropped pass which was intercepted with 8 mins left in 2018 (+28.7% for KC)
- CIN made game winning 52-yd FG in 2021 (-27.8% for TEN)
4
3
77
22,009
Using NGS tracking data, here is where each team's QB spent their time with the ball (during dropbacks or designed rushes) during the '23 season.
CHI QB Fields spent the least amount of time in pocket (80.5%), while LV QB O'Connell spent the most amount of time in pocket (98.5%)
4
16
71
30,335
I enjoy the discussion of whether a team that literally won x games is truly an x-win team.
Using the output from our state space model, here is the distribution of strengths for each win count - also highlighting the "worst team", the median value and the "best team".
3
10
69
With #Madden23 ratings being released, I decided to plot Madden 23 player Speed ratings vs observed player max sustained speeds (from NGS).
4
7
69
Using our team strengths model and simulation, here is the playoff leverage of each game remaining for every team that is still "in the hunt".
No team has a game more important to its playoff hopes than Baltimore's contest at Cleveland this weekend.
6
12
69
Using @nflseedR with info through week 17, here are the most likely matchups (with no regards to seeds/home or away) in this season's Wild Card Round, Divisional Round, Conference Championship Round and Super Bowl.
4
8
68
22,907
So far the 2020 NFL season has had a lot of close games and a few notable comebacks. Using our Win Probability model, I take the median WP of every game (ignoring OT) and sum it to calculate "estimated wins" and "estimated losses" for each team. Here are the results:
10
16
68
With Madden 22 ratings being released, I wanted to compare observed speeds / accelerations in game to Madden speed / acceleration ratings for the 2020 season.
For each player, I compare the avg max speed/acc reached during the 1st 2 secs of go routes to their Madden 21 speed/acc
7
12
70
Updated by plot that shows the top 5 colleges in terms of % of snaps played for each NFL team for 2022.
Which college does your team have the most snaps from?
For each NFL team, here are the top 5 colleges in terms of % of snaps played during the 2021 season. Which college does your team have the most snaps from?
(Credit to @JEvanReid for idea)
2
7
69
29,164
Final Dashboard of '23:
Top plays wk 18:
- ARI missed FG w/ 0 secs left (+52.3% for SEA)
- IND dropped pass on 4th down w/ 1:06 left (+47.3% for HOU)
Top plays 2023:
- CHI dropped TD pass w/ 0 secs left (+92.8% for CLE)
- MIN dropped TD pass w/ 12 secs left (+90.5% for LAC)
5
12
67
43,458
T-1 Day until NFL Football!!!
While applying our new 7-Team playoff format, here are observational probabilities from 2002 - 2019 of making the playoffs for every possible record a team can have throughout a 16 game season:
2
14
66
Since 2001, Aaron Rodgers has the most Hail Mary TDs, Hail Mary attempts and highest Hail Mary TD percent of all QBs with 5+ attempts.
7
15
68
26,174
Opposing Teams' Receivers playing at Arrowhead:
- Early Afternoon: 6.3%
- Late Afternoon: 6.9%
- Night: 10.8%
2
10
68
15,412
Inspired by work from @erinpsajdl during her summer internship at the NFL, here is a plot showing QB hits per 100 dropbacks split by Quarterback ordered by their average across seasons.
7
12
59
Inspired by the recent discourse on NFL players playing in the NBA/NBA players playing in the NFL, we decided to compare the two groups.
On average, NFL players weigh 29 lbs more than NBA players, but are 4" shorter.
More details in the #ExtraPointBlog:
operations.nfl.com/gameday/a…
6
11
64
12,260
We are about a month into the 2022 #BigDataBowl and we are happy to provide you more demo code to help you with your submissions. This time, we analyze player speed on kickoffs!
Again, we have this in both R and Python.
R: kaggle.com/tombliss/speed-on…
🐍: kaggle.com/dhritiyandapally/…
2
6
57
Update on coach timeout usage ahead of championship weekend.
Andy Reid #ChiefsKingdom and Nick Sirianni #FlyEaglesFly are amongst the lowest in unnecessary timeout usage while Shanahan #FTTB and Taylor #RuleTheJungle are in the upper 50%.
5
19
60
87,855
I was curious about that too... it was this play:
piped.video/mge5U1H5kY8?t=292
2
56
21,352
NFL win prob added via Luck dashboard updated through week 7:
- Dotson dropped pass on 4th and 5 at end of WAS-NYG is highest WPA via luck play of week 7
- Despite slightly net negative week for LAC and slightly net positive week for MIN, both remain first and last respectively.
1
17
52
19,575
Using Next Gen Stats tracking data for all QB designed runs/dropbacks in 2021, here is the % of time spent with ball for each team's most played QB:
5
8
52
Since 2011, NFL QBs who received snaps have become
- Faster by -0.10 secs (40-yd dash)
- Shorter by 1.0 inches
- Lighter by 8 lbs
4
6
55
11,627
Revisiting my analysis on laterals per opportunity in the NFL:
Laterals per pass reception has increased slightly in 2024 to a rate of nearly 2 per 1000 catches.
Unsurprisingly, in the #NFL, the lateral rate is much higher by defensive players as well as kickoff/punt returners than an offensive player during an ordinary scrimmage play.
(On rush plays, only downfield laterals are counted. Last play of regulation filtered out.)
2
7
52
22,848
We are hiring a graduate student intern for Summer 2024 who will work on projects that relate to NFL competitiveness, pace of play, officiating & health/safety.
To apply, select "Football Data & Analytics" as top choice of department in the application:
hdmm.fa.us6.oraclecloud.com/…
4
18
49
38,501
Happy Super Bowl Sunday!
Tampa Bay's Raymond James Stadium is among several NFL stadiums with a noticeable difference in Field Goal Make Rate by North/South direction.
It is easier to kick North (towards the pirate ship) than South.
3
11
47
For those submitting to the #BigDataBowl, here is a list of potential topics. This list is not to be thought of as complete and there are many great topics not listed that we would encourage participants to do:
kaggle.com/competitions/nfl-…
1
4
49
When a team is kicking a late field goal, they will often run a play to center the attempt for their kicker.
Using NGS tracking data with a binomial GAM on FG attempts, there does not seem to be a strong benefit to do so - if anything centering the kick lowers the make rate.
11
10
46
Using data from @MaxPreps , I analyze home advantage (home score - visit score) in high school varsity football.
AK (large, spread out state) has highest H.A. while NH (small state) has lowest H.A.
This makes sense considering the trend between travel and home point diff.
2
7
47
The 2023 #BigDataBowl has started on Kaggle (sign up: kaggle.com/competitions/nfl-…).
To help participants get started, we put together a demo notebook analyzing edge rushers which will show how to explore the data, animate a play and create a few basic plots:
kaggle.com/code/tombliss/pas…
1
10
44
Unsurprisingly, in the #NFL, the lateral rate is much higher by defensive players as well as kickoff/punt returners than an offensive player during an ordinary scrimmage play.
(On rush plays, only downfield laterals are counted. Last play of regulation filtered out.)
WHAT JUST HAPPENED?
AN ABSOLUTLEY WILD @Chiefs TOUCHDOWN IN FRANKFURT.
📺: #MIAvsKC on NFL Network
📱: Stream on #NFLPlus bit.ly/46VrzAY
3
5
42
60,436
Apply to the 2025 Graduate Student Internship with my team (select "Football Data & Analytics" as department preference).
We are looking for someone who knows & loves the sport of football and is skilled in R and/or Python.
job-boards.greenhouse.io/nfl…
1
16
45
21,456
College latitude/longitude data is from @masseyratings.
Same caveat as yesterday: due to large computational times required for confirming conference grouping with lowest distances, this map may not represent global minimum.
2
42
22,625
Dynamic kickoffs in the '24 preseason featured a 70.5% return rate (+15.7% from last yr) and an avg drive start from the 28.8 (+4.4 from last yr).
Kickoff teams had the most success when the ball landed from the 1-5 yard line.
More in the Extra Point:
operations.nfl.com/gameday/a…
2
11
44
26,599
Another day, Another #BigDataBowl dataset! Thanks to @TelemetrySports, we give you coverage schemes for each Week 1 game and play in the data!
Example Notebook 😀🔥😄:
kaggle.com/tombliss/addition…
Data 🤓🤓🤓:
kaggle.com/tombliss/nfl-big-…
6
42
Luck dashboard update for weeks 1-15.
NYG@MIN this Saturday is a matchup between the two luckiest teams.
9
8
40
60,624
Hahaha - yeah to be honest when I saw my solution I thought I could’ve figured out the ACC and PAC-12 pretty easily and maybe just done the other 3 from there…
Doodling is sometimes easier/just as good as difficult mathematical processes
33
11,555
It was awesome to be part of the process of building this model. Credit to @benbbaldwin for creating an incredible baseline framework and to @MBandNFL for working with AWS to get it into production.
For fans, the best football defense is one you can read. 🏈☁️💡
Powered by #AWS, Coverage Classification is a first-of-its-kind stat that uses over 60,000 @NFL plays to identify the defense seconds after a play. Lean into the huddle. @NextGenStats
🖇 go.aws/3yDKVeO
2
4
40
Using our play by play data (which goes back to 1999), I found one other game where this happened.
It was a Raiders @ Chiefs game in the 1999 season:
pro-football-reference.com/b…
Chiefs scored 17 points before 2nd offensive possession with punt return TD, pick six, and field goal.
There can't be very many games in NFL history where a team has scored 16 points before their 2nd offensive possession
4
2
39
26,716
Here are histograms showing the locations of field goal and extra point make/miss locations in 2022 with kickers sorted by avg offset from center.
Badgely and Slye (competing this preseason to be Washington's starting kicker) have average offsets of 1.3 and 1.7 respectively.
2
3
39
12,451
All starting QBs having unique first names is uncommon - 36/367 weeks since '03.
Last time was wk 13:
Bailey, Baker, Brock, Bryce, CJ, Dak, Derek, Desmond, Gardner, Geno, Jake, Jalen, Jared, Joe, Jordan, Justin, Kenny, Kyler, Matt, Patrick, Russell, Sam, Tim, Trevor, Tua, Will
2
1
40
17,346
Luck Dashboard thru week 12.
Most impactful plays:
- WAS missed game-typing XP with 0:21 left in Q4 (+37.6% for DAL)
- HOU missed game-tying FG with 1:56 left in Q4 (+27.1% for TEN)
- SF dropped pass intercepted by GB with 6:00 left in Q3 (+17.9% for GB)
3
5
38
74,930
20th out of the 32 teams:
Final Dashboard of '23:
Top plays wk 18:
- ARI missed FG w/ 0 secs left (+52.3% for SEA)
- IND dropped pass on 4th down w/ 1:06 left (+47.3% for HOU)
Top plays 2023:
- CHI dropped TD pass w/ 0 secs left (+92.8% for CLE)
- MIN dropped TD pass w/ 12 secs left (+90.5% for LAC)
4
35
15,490