Table of Contents
Start With a Question, Not a Stats Page
Most people open a box score, scan the columns, and form an opinion somewhere around the fourth number. That order is backwards, and it’s why arguments about players go in circles. The useful sequence is: question first, stat second, context third.
Say a hitter is batting .340 in mid-May and you want to know whether to believe it. “Is he good?” is too vague to answer with a number. “Is his power real, and is he controlling the strike zone?” is answerable. That question points you at isolated power, barrel rate, and strikeout-to-walk ratio instead of batting average, which is the noisiest number on the page.
Here’s a rough translation table I use constantly:
- Is this hitter’s power real? Look at ISO, barrel rate, and average exit velocity, not home runs.
- Is this goalie good or just busy? Compare save percentage against post-shot expected goals allowed.
- Is this guard’s scoring helping the team? True shooting percentage plus usage rate, not points per game.
- Is this defender as good as the highlight reel? Tackles and interceptions per 90, plus progressive passes received.
- Is this receiver’s breakout sustainable? Targets per route run and average depth of target.
If you’re new to the idea that the same stat can mean two different things depending on how it’s built, this breakdown of what player stats tell you and what they hide is worth ten minutes before you go further.
Step 1: Decide How Much Data You Need Before You Believe Anything
This is the step almost everyone skips, and it’s the one that saves you from looking foolish in group chats.
The 60-plate-appearance problem
In MLB, strikeout rate becomes reasonably stable at roughly 60 plate appearances. Walk rate needs about twice that. Batting average needs somewhere near 500 plate appearances before it says much about the player rather than the luck. A hitter who is 18-for-52 in April has shown you almost nothing.
Exit velocity is the opposite: about 40 batted balls and you have a real signal, because it’s measuring the swing itself rather than the outcome.
The same logic in other sports
In the NBA, three-point percentage needs several hundred attempts before it stops bouncing around. A 12-game stretch at 44% from deep is a heater, not a skill change. In soccer, shooting conversion and save percentage both take the better part of a season to settle, which is why xG-based metrics were invented in the first place.
Write your thresholds down. Mine are: strikeout rate 60 PA, walk rate 120 PA, ISO 200 PA, batting average 500 PA. When a number hasn’t crossed its threshold, I treat it as a rumor.
Step 2: Put Every Number on the Same Unit
Raw totals punish players who play less, and bench players always look worse than they are. Convert before you compare.
Player A scores 18 points in 30 minutes. Player B scores 18 in 22 minutes. Same box score line, very different players. Per-36 or per-75-possession tells you B is the more productive scorer, though it also raises the question of whether he can hold up with more minutes and tougher matchups.
Basketball uses per-100-possession or per-36. Soccer uses per 90. Hockey uses per 60. Baseball is the outlier because the plate appearance is already a fair unit, which is why rate stats like OBP and slugging work so cleanly.
One trap: don’t mix counting and rate stats in the same argument. “He has more rebounds” and “he rebounds at a higher rate” are two different claims, and only one of them is context-free.
Step 3: Add the Context Layer Before You Judge
A stat with no context is a number with a costume on. Three adjustments cover most of what you need.
Park and venue
Coors Field inflates offense by roughly 15–20% compared with a neutral park. A .280/.350/.480 line in Denver is meaningfully less impressive than the same line in San Diego. Baseball Reference publishes park factors; use them before you crown anybody.
Opponent quality
A quarterback who carved up a bottom-five pass defense in Weeks 3 and 4 hasn’t proven much. Note the opponent next to the number in your notes, or you’ll forget by November.
Role and usage
Usage rate in basketball, target share in football, and touches in the final third in soccer all shape what a player is even allowed to produce. A midfielder whose tackles dropped by half may not have declined at all. He may have been pushed 15 yards higher up the pitch.
Step 4: Compare Against a Baseline You Actually Trust
Numbers mean nothing in isolation. Pick one of three baselines and stick with it for the whole comparison.
- League average tells you whether the player is above or below the middle of the pack.
- Positional average is usually better, since a shortstop and a first baseman are judged on different scales.
- The player’s own history is best for detecting change. Two hundred games of his own past beats any league-wide benchmark.
This is the core of judging athletes fairly across different sports, and it’s why comparing a defensive midfielder to a striker on goals is a waste of everyone’s afternoon.
Step 5: Build a Ten-Minute Weekly Log
Tracking player stats only pays off if you do it on a schedule. Ad-hoc checking produces recency bias; you remember the last game and forget the previous nine.
My Tuesday routine, which borrows heavily from a coach-style Tuesday standings review, takes about ten minutes:
- Pick five players you’re actually following. No more.
- Log one rate stat, one volume stat, and one context note per player.
- Write a single sentence explaining any change from last week.
- Flag anything that looks like a role change, not a skill change.
- Ignore everything that hasn’t crossed its sample-size threshold.
A real entry from last season: “Winger, 14 matches, 0.31 xG per 90, down from 0.48. Touches in the box halved. New manager playing him as a wing-back.” That one line saved me from a bad take.
Step 6: Read the Opponent’s Numbers Too
Half of what looks like a player’s improvement is a schedule quirk. A pitcher with a 1.90 ERA over six starts against three of the worst offenses in the league has a story his ERA isn’t telling.
Cross-check against what the other side did. If a defense allowed 30 points in three straight games and your running back went off against them, that’s a matchup note, not a breakout. Learning to read what final scores hide sharpens this fast, because it forces you to look at how the numbers were produced rather than just what they totalled.
The scout habit is simple: pull the last five opponents, note their defensive rank, and write it next to the stat line. If a player’s hot streak came entirely against bottom-ten units, discount it. If it came against top-ten units, take it seriously.
When the Numbers and Your Eyes Disagree
Sooner or later you’ll watch a player who passes the eye test and see mediocre stats, or the reverse. The instinct is to pick a side. The better move is to assume one of you is measuring the wrong thing.
Three explanations cover most disagreements. First, the stat is measuring outcomes rather than process: a defender who consistently closes down space may record few tackles because opponents stop passing into his zone. Second, the role has shifted since the number was set: a player moved off his best position will look worse in almost every metric. Third, the sample is too small, and you’re arguing about a 40-plate-appearance heater.
The fix is to go find the stat that matches what your eyes are seeing. If he looks slower, check sprint speed and burst metrics. If he looks like he’s creating more but the assists are flat, check shot-creating actions or expected assists, which credit the pass before the pass. Once you locate the missing measurement, the disagreement usually disappears, and you’ve added a tool that works for the next twenty players too.
Keep the log going, keep the sample sizes honest, and let the numbers argue with your eyes instead of overruling them.
