Articles
Original analysis, explainers, and arguments about the numbers behind the games. New pieces are added regularly — recent essays at the top.
-
NBA
Analytics
Blocks measure risk, not rim protection
The block looks like the cleanest measure of rim protection, but it rewards the gamble and ignores the shots a defender deletes by just standing there. Why the swat leaders aren’t the best protectors.
-
NFL
Analytics
Third-down conversion rate is a distance stat
Third-down percentage gets read as clutch execution, but it mostly records how far a team had to go — and that was decided two downs earlier. Read it by distance, not as a verdict.
-
MLB
Analytics
The RBI measures lineup spot, not hitting
A hundred RBI reads like a great season, but the total mostly records how many runners were on base when a hitter came up — a function of lineup spot and teammates, not hitting.
-
NHL
Analytics
The game-winning goal is hockey’s emptiest stat
The game-winning goal sounds clutch, but it’s assigned by counting up from the loser’s total — often a meaningless early goal. Strip the label and it’s just goals with a better name.
-
EPL
Analytics
Key passes inherit the finisher’s luck
Chances created feels like a clean creativity stat, but the count depends on whoever receives the ball choosing to shoot, and treats a tap-in setup like a hopeful half-chance. Expected assists fixes it.
-
EPL
Analytics
Pass completion rewards the safest pass
Completion percentage pays players for declining difficult passes. Why centre-backs top the leaderboard, creators sit at the bottom, and volume stats reveal what the percentage hides.
-
EPL
Analytics
Goal contributions add two different skills
G+A welds a skill a player controls to one he only half controls, ignores minutes, and prices a penalty like a wonder goal. Split the sum back apart and read the profile instead.
-
EPL
Analytics
The appearance is football’s emptiest counting stat
A full ninety and a stoppage-time cameo both count as one appearance. In the five-sub era the column no longer measures playing time — minutes do, and the division rewrites most durability narratives.
-
EPL
Analytics
Yellow cards measure reputation as much as fouls
A booking is a referee's judgment shaped by position, tactical assignment, reputation, and suspension math. The card column is an eyewitness account, not a character record.
-
EPL
Analytics
Five-game form is noise wearing a trend line
The form strip shows five results and gets read as momentum. Five matches in a low-scoring sport mostly display variance and fixture luck — season-long numbers beat the badges every week.
-
NBA
Analytics
The assist is a scorekeeper's opinion
Every other box-score stat has a hard trigger. The assist is awarded by a human scorer making a judgment call in real time — which is why arena bias and the hockey-assist blind spot are baked in. How to read playmaking without leaning on it.
-
NFL
Analytics
Completion percentage is a depth-of-target stat
71% completions sounds accurate. Mostly it means he threw short. Why a checkdown artist and a vertical gunslinger can't be compared on the same number — and what completion percentage over expected actually measures.
-
MLB
Analytics
OPS adds two numbers that don't belong together
Baseball's favorite advanced stat sums two fractions with different denominators and weights them as equals when on-base is worth nearly twice as much. What OPS gets right, what the addition distorts, and why wOBA fixes it.
-
NHL
Analytics
PDO is regression wearing a jersey
Shooting percentage plus save percentage gets cited as a pure luck dial. It mostly regresses to 100 — but not always. How to tell real finishing and goaltending talent from variance.
-
EPL
Analytics
The big chance is a subjective stat
"Big chances missed" looks like hard data sitting next to xG. It's a human analyst's judgment call that double-counts the model and discards the gradation xG was built to keep. How to read finishing fairly.
-
NBA
Analytics
Usage rate is a workload stat, not a skill stat
A 32% usage rate sounds like a star. It's often a description of the supporting cast's poverty as much as the lead guard's gravity. Here's why usage measures role and roster context, not player skill — and what to pair it with when reading a player's offensive impact.
-
NFL
Explainer
Passer rating was built for 1973 football
The NFL passer rating ignores sacks, ignores rushing, caps its inputs, and weights yards the same regardless of down and distance. The formula was the right answer in 1973. Here's why the broadcast number no longer matches the game being played underneath it.
-
MLB
Analytics
ERA is half a pitching stat
The pitcher's name sits on top of earned run average, but 40-50% of the variance in any season's ERA comes from defense, park, and sequencing. Here's why FIP and SIERA predict next-year ERA better than current-year ERA does, and what that says about who's really on the leaderboard.
-
NHL
Defense
Hits don't measure defense in hockey
You can only hit a player who has the puck. Teams with the highest hit totals are usually the teams that don't have it. Here's why the broadcast graphic praising physicality is mostly logging a possession deficit, and what actually predicts run prevention on a hockey rink.
-
EPL
Tactics
Tackles are a positioning failure stat
Van Dijk and Rodri win individual honors with tackle counts that look pedestrian. The league's high-tackle defenders are mostly the ones whose teams' structure repeatedly forces duels. Here's why a low tackle count is often the signature of an elite defender, not a disengaged one.
-
NBA
Methodology
The 3-and-D archetype is the most overvalued role in basketball
Every front office wants three or four 3-and-D wings. Most of the players who get the label don't actually deliver both halves of the package at the level their contract assumes. Here's why the market keeps overpaying for an archetype built more on body type and one hot shooting season than on sustained two-way performance.
-
NFL
Methodology
The 40-yard dash has stopped predicting NFL careers
For fifty years the 40 was the headline number of the combine. The correlation between 40 time and NFL career success has been declining for two decades, and across most positions is now statistically zero. Smart front offices have moved to the 10-yard split and tracking data. The broadcast hasn't caught up.
-
MLB
Methodology
Defensive WAR is the noisiest part of WAR
WAR is the closest baseball has to a single-number player evaluation. Its defensive component has 2-3x the standard error of its batting component. Most MVP-by-WAR arguments rest on gaps smaller than that error. Here's why the WAR totals you see quoted deserve much wider uncertainty bars than they get.
-
NHL
Methodology
Time on ice is a coaching alibi, not a stat
Top defensemen are credited with workload based on TOI totals their coaches engineered. The same player under a balanced-rotation coach would post 3-4 fewer minutes a night and look like a lesser horse. Here's why TOI is closer to a coaching deployment log than a player skill measurement.
-
EPL
Methodology
The 'big six' label is statistically incoherent
The gap inside the big six is now six times the gap between its bottom edge and seventh place. The label survives as a brand convention, not a statistical category, and it's been miscalibrating fixture-difficulty analyses and transfer-market pricing for years. Here's what replaces it.
-
Cross-sport
Big picture
The dynasty is dead, and the math killed it
No MLB team has repeated as champion since 2000. No NBA team since 2018. The Chiefs back-to-backed and lost their three-peat bid. The Lightning back-to-backed and stopped. The drought across every major sport isn't a coincidence — it's the structural output of salary caps, free agency, the analytics floor, and the variance of multi-round playoffs.
-
NBA
Defense
Defensive rating doesn't measure defense
The cleanest-looking advanced defensive stat in basketball is mostly a team rating with a single player's name attached. A defender who gets traded loses 8 points off his rating without changing his defensive play. Here's why the column on the back of the card is a team stat in disguise.
-
NFL
Methodology
Time of possession is the most misleading stat in football
TOP correlates with winning — but as a downstream effect, not a cause. Teams that lead run more clock; teams that trail throw incompletions that stop it. Once you control for who led at any point, the TOP-to-winning correlation collapses to zero. The graphic is recording the consequence of having already played well.
-
MLB
Methodology
The save is a structurally broken stat
A sportswriter invented the save in 1969 to give relievers some recognition. The rule, as written, rewards the wrong inning and the wrong leverage, and it has reshaped bullpen deployment in ways that demonstrably cost teams wins. Here's why the structure itself is the problem.
-
EPL
Methodology
Pressing intensity stats are coaching marketing
PPDA, high turnovers, and counter-press recoveries are the analytics-flavored framing for a generation of fashionable managers. Most of these stats count attempts, not successes, and several of the most-cited "pressing sides" finish in the bottom half on xG against. Here's what the headline stats are hiding.
-
NBA
Methodology
Clutch stats are the smallest sample size in sports
The "clutch" graphic that appears on every late-game broadcast is computed off roughly 60 field-goal attempts — about two full games of basketball spread over six months. The math says it should swing 13 percentage points season to season for the same player. The discourse has been ignoring that for forty years.
-
NFL
Methodology
Yards per carry is a trap for evaluating running backs
A 5.0 yards-per-carry average reads as obviously good. A 3.4 reads as obviously bad. Both numbers tell you almost nothing about the back. Here's how the stat is confounded by long-run variance, blocking, and game script — and which numbers actually predict.
-
MLB
Defense
Fielding percentage hides what defenders actually do
The oldest defensive stat in baseball still appears on broadcast graphics, still decides Gold Gloves, and still rewards passivity. A defender who never attempts a tough chance posts a better fielding percentage than one who attempts everything — and is worse at defense. Here's what replaced it for serious evaluation.
-
NHL
Methodology
Faceoff percentage is the most overrated number in hockey
Coaches obsess over it. Broadcasters open every period with it. Contracts get partially justified by it. A faceoff win is actually worth about 0.015 expected goals — real value, but a tiny fraction of what the airtime implies. Here's what the stat is and isn't doing for your team.
-
EPL
Defense
The clean sheet is mostly a defender alibi
A clean sheet against an opponent that took three open-net chances and skied all three looks identical, on the box score, to one earned against a defense that allowed zero. Both produce the same defender reputation. Here's why the headline stat hides the most important defensive information in the match.
-
NBA
Defense
The closeout is modern basketball's defense
The three-point revolution didn't just change offense. It turned the entire NBA defense into one repeated action — the closeout — and made the players who do it well the most undervalued defenders in the league.
-
NFL
Roster building
The draft trade value chart needs a rewrite
The Jimmy Johnson chart has governed NFL draft trades for thirty years. The economics literature has known it's wrong for almost as long. Here's how it's wrong, why it survives, and what the teams that actually understand the math have started doing instead.
-
MLB
Analytics
Why early-season ERA tells you almost nothing
Every May, baseball broadcasts overflow with Cy Young narratives built on six or seven starts of data. Almost none of those ERAs will survive August. Here's why early-season ERA is mostly noise — and the small set of pitching numbers that actually mean something this early.
-
NHL
Methodology
Stanley Cup overtime is the most random format in sports
Sudden-death, full-strength, no-shootout playoff hockey is the most dramatic format in North American sports. It's also the highest-variance championship format we have. Here's the math on how much of an OT result is luck — and how to read playoff overtime without being fooled by the narratives.
-
EPL
Analytics
The Premier League run-in is mostly a story we tell ourselves
The "run-in" is one of the most pundit-favored framings in soccer broadcasting and one of the least informative. Most of the late-season form people credit to mental strength is fixture difficulty plus small-sample variance. Here's what actually decides the table in May.
-
Cross-sport
Methodology
Every analytics revolution is a per-possession revolution
Every major sport went through the same analytical shift: from counting stats per game to rate stats per opportunity. Per-100 in basketball, per-drive in football, per-60 in hockey, per-plate-appearance in baseball, per-90 in soccer. The same idea, the same resistance, the same outcome.
-
MLB
Analytics
Pitcher wins are a useless stat (and what to use instead)
The pitcher win is the oldest individual stat in baseball and one of the worst. Here's why it survived a century of obvious flaws, what it actually measures, and what to use instead when evaluating a starting pitcher.
-
NBA
Game evolution
Why every NBA team plays faster now
NBA pace bottomed out in the late 1990s and has been climbing ever since. Today every team plays faster than the fastest team of the early 2000s. Here's what drove the revolution, what it changed, and what it cost.
-
NHL
Analytics
Plus-minus is the worst widely used stat in hockey
Plus-minus is the most-cited individual hockey stat that survives in the modern game and the noisiest one we have. Here's why it's almost useless on its own, what it's actually measuring, and what to look at instead.
-
EPL
Tactics
Why possession stats lie
Possession percentage is the most-cited team stat in soccer broadcasts and the most misleading. Here's why the team with more of the ball is often losing on purpose, what possession actually correlates with, and what to look at instead.
-
NBA
Explainer
True shooting percentage and the death of FG%
Field goal percentage was the gold standard for scoring efficiency for fifty years. Then the three-pointer broke it. Here's how true shooting took over, what it actually measures, and where it still falls short.
-
NBA
Analytics
The mid-range jumper isn't dead. It just got expensive.
The story that analytics killed the mid-range jumper is half right. It collapsed for most players, but the league's best scorers still hunt it — because they understand a piece of the math the league-average version misses.
-
NBA
Awards
How load management broke the All-NBA vote
The 65-game threshold was supposed to fix load management. It mostly just changed who got snubbed. Here's what the data shows about rest, awards, and what the rule actually optimized for.
-
NFL
Analytics
The QB stat that actually predicts winning
Passer rating is everywhere. Wins-as-a-starter is everywhere. Both tell you very little. Here's the quarterback metric that actually correlates with team success — and why the broadcasts still won't lead with it.
-
NFL
Tactics
Why every NFL team scripts the first 15 plays
Bill Walsh started scripting the first 15 plays of his offensive game plans in the 1970s. By the 2020s, every NFL team did it. Here's why the script works — and why it isn't really about the plays it calls.
-
NFL
Defense
The best defensive linemen never get sacks
The interior defensive line is the most undervalued position group in the NFL because the box score doesn't have a column for what it does. Here's why pressure rate and run-stop win rate matter more than sack totals.
-
MLB
Analytics
Why batting average refuses to die
Sabermetrics has spent thirty years explaining why batting average is the wrong way to evaluate a hitter. The stat keeps showing up on every broadcast anyway. Here's why it survives — and what it actually does and doesn't tell you.
-
MLB
Rule changes
What one full year of the shift ban actually proved
The shift ban was supposed to revive batting averages and bring back base hits. What actually happened is more subtle and a useful case study in how rule changes interact with optimization.
-
NHL
Analytics
Corsi, Fenwick, and the shot quality revolution
Corsi was the stat that taught a generation of hockey fans to think about possession instead of scoring. Fenwick was the refinement. Expected goals replaced both. Here's how the shot-quality revolution worked, and what's coming next.
-
NHL
Methodology
Why goalies are the hardest position in sports to evaluate
Goaltenders are the most leveraged position in hockey and the hardest one to evaluate. Their performance is a confounding mix of skill, defense quality, sample size, and equipment trends. Here's why every public goalie stat is somewhat broken.
-
EPL
Tactics
What xG misses about set pieces
xG revolutionized soccer analytics but has always struggled with set pieces. Here's why corners, free kicks, and penalties confound the model, and how the smartest teams have built whole tactical philosophies around exploiting the gap.
-
EPL
Analytics
The relegation six-pointer is a myth
The "six-pointer" between two relegation-threatened sides is a cliche so embedded in broadcast vocabulary that nobody questions whether it makes mathematical sense. It mostly doesn't. Here's what the math actually says about late-season survival.
-
NFL
Analytics
Going for it on fourth down: how the math finally won
For most of NFL history, fourth down was a cultural decision, not a mathematical one. The math had been screaming the opposite answer since the 1970s. Here's why coaches resisted, what changed, and what the analytics still don't capture.
-
NBA
Explainer
How to actually read an NBA box score
Every column on the box score is supposed to tell you something. Most of them are also lying to you, or at least withholding important context. Here's what each number is really saying — and what to look at first when you only have ten seconds.
-
NHL
Analytics
Save percentage is the most-quoted, least-useful goalie stat
Every NHL broadcast leans on save percentage like it's a verdict. It isn't. It's a ratio that treats every shot as the same, which is the one thing every shot definitely is not. Here's the case against it, and what to use instead.
-
EPL
Analytics
Why expected goals outgrew the sport that invented it
xG started as a niche stat football traditionalists rolled their eyes at. A decade later it's on every broadcast graphic and every recruiting board. What it does well, what it still misses, and why it spread to every sport that lets you take a shot.