Where the data comes from

All of the data on this page comes directly from Major League Baseball's own servers — the same systems that power MLB.com and the Ballpark app. MLB makes this information freely available, and Scootercam pulls it automatically, several times a day, without any manual effort.

Once an hour, Scootercam checks for updated scores, standings, and season statistics. Twice a day — morning and evening — it specifically checks which pitchers are scheduled to start upcoming games, then runs each one through a scoring formula that weighs their ERA, walks, strikeouts, and hits allowed. Overnight, a third automated check retrieves final scores and pitching lines from completed games, building a running record that lets the model evaluate its own predictions over time. Nothing here is manually entered or editorially adjusted. The numbers come from MLB, the scores are calculated fresh each run, and the page updates on its own. We start collecting league-wide data on May 5, 2026.

Cubs & Sox combos and prediction markets

Prediction markets like Kalshi let you bet on whether the Cubs, the Sox, or both will win on any given day. The most interesting play — and the one this page is built around — is the combo: a single bet that pays out only if both teams win. The math works against you more than people expect. Given the Cubs' and Sox's current records, the odds of both winning on the same day by pure chance are shown right on the page. The combo score is the model's attempt to find days when the pitching conditions meaningfully improve those odds — when both teams are facing hittable starters and the environment favors offense. A high combo score doesn't guarantee a win; it means the conditions are better than average. A low one means the opposite, and the model says to sit it out.

How pitcher scores inform predictions

The pitcher score — 0 to 100 — is a pre-game read on how tough an opposing starter will be to score against. A score in the 70s or 80s means you're looking at an elite arm: low ERA, few baserunners allowed, high strikeout rate. That's a pitcher who makes winning harder and suppresses run totals. A score in the 30s or 40s means the starter is hittable — more walks, more contact, more opportunities. When you're evaluating a prediction market position, the pitcher score gives you a sharper lens than the team's overall record alone. A team that wins 60% of their games still wins far less often against an 85-scored starter than a 35-scored one.

The Kalshi market probability shown next to each game is an independent check — it reflects what real-money bettors currently think, factoring in things the pitcher score doesn't see: lineup changes, weather, bullpen fatigue. When the pitcher score and the Kalshi number point in the same direction, the signal is stronger. When they diverge sharply — say, a pitcher scored 80 but the market has the opposing team favored — it's worth pausing to understand why before acting.

What this section is

The Scout Validation section is a visual accuracy test for the MLB Scout model. Before each game, the model calculates a pitcher score (0–100) for every probable starter — a number that reflects how dominant or hittable that pitcher is expected to be. This section takes those pre-game predictions and stacks them up against what actually happened once the game was played.

The scatter plot — reading the axes

Every dot on the scatter plot is one pitcher's start. The two axes tell the complete story:

• Horizontal axis (X) — Pre-Game Pitcher Score (0–100). This is what the model predicted before the game. A dot on the right side of the chart (score 80–100) means the model considered that pitcher elite for that day — low ERA, few baserunners allowed, high strikeout rate. A dot on the left side (score 0–30) means the model flagged that pitcher as hittable.
• Vertical axis (Y) — Earned Runs Allowed. This is what actually happened on the mound. Low on the Y axis means a good outing — few runs allowed. High on the Y axis means the pitcher got hit around.

What a working model looks like

If the scorer is doing its job, the dots should form a pattern that slopes downward from left to right. Pitchers with low pre-game scores (left side) should cluster toward the top of the chart — they got hit hard, as predicted. Pitchers with high pre-game scores (right side) should cluster near the bottom — they shut down the opposing lineup, as predicted.

You won't see a perfect diagonal line. Baseball is unpredictable. But over enough starts, the overall tendency should lean in that direction.

The trend line

The dashed line cutting through the dots is a best-fit line — a mathematical average of the relationship between scores and outcomes across all the data. It tells you the overall story at a glance without your eyes having to find the pattern in dozens of individual dots.

• Green trend line — The slope runs downward (high score → fewer runs allowed). This is the correct direction. The steeper the green line, the stronger the model's predictive power.
• Red trend line — The slope runs upward or flat. This means high-scored pitchers aren't actually outperforming low-scored ones — a diagnostic signal that the formula needs attention.

Early in the season, the line may be red or nearly flat simply because there aren't enough data points yet. As starts accumulate through the season, the line should settle into its true direction.

The larger dots

Some dots are slightly bigger than others. These represent quality starts — defined as 6 or more innings pitched with 3 or fewer earned runs allowed. If the model is working, you'd expect larger dots to appear more frequently on the right side of the chart (high scores) than the left.

Outliers — what they mean

No model is perfect, and you'll see dots that defy the trend:

• Top-right outlier — A high-scored pitcher who got lit up. The model overrated them. One or two of these is noise; a cluster suggests the formula may be over-weighting a particular stat without accounting for the opponent's lineup.
• Bottom-left outlier — A low-scored pitcher who had a great outing. The model underrated them. These are worth noting — there may be an intangible the scorer isn't capturing.

The call accuracy table

Below the scatter plot, games are grouped into Favorable, Neutral, and Pass buckets and compared against the actual run environment — the total runs scored in each game. If the model is working, Favorable games (hittable pitching on both sides) should average more total runs than Pass games (elite pitching on both sides). The wider the gap between those two averages, the more the model is seeing something real. "High-scoring" is defined as 9 or more total runs in the game.

The pitcher leaderboard

The leaderboard can be toggled between two views:

• Most Appearances — pitchers ranked by how often they've shown up in the dataset. Useful for identifying arms you'll encounter frequently throughout the season.
• Biggest Surprises — pitchers ranked by how far their actual performance diverged from their predicted score. A large positive surprise (red) means they allowed many more runs than expected — the model was overconfident. A large negative surprise (green) means they were much better than predicted. Consistent surprises in the same direction for a specific pitcher point to a formula flaw worth investigating.

How to use this day-to-day

The scatter plot isn't a game-by-game tool — it's a calibration instrument. Check it periodically (weekly or bi-weekly) to answer one question: Is my model still predictive? If the trend line is green and the dots loosely follow the slope, trust the daily scores. If the trend line has gone red over a stretch of 30+ starts, it's time to look at the formula's inputs and weightings.

Think of it like a gun sight — the scatter plot tells you whether the barrel is still pointed in the right direction, so you can trust (or adjust) your aim on any given day's slate.

Using the leaderboard to handicap borderline calls

The pitcher score handles the easy decisions — a score in the 70s or 80s is a clear pass, a score in the 30s or 40s is a favorable environment. The gray area is the middle. When both opposing starters land in the 45–65 range, the model isn't sending a strong signal, and that's when the leaderboard earns its keep.

Before acting on a borderline day, look up each opposing starter in the leaderboard and weigh three things:

• QS% — a pitcher scoring 55 who delivers quality starts 60% of the time is a durable arm the score may be underselling. A pitcher scoring 55 with 20% QS% is volatile — capable of both dominant outings and early exits. Volatile pitchers create variance that cuts both ways; durable ones are more predictable.
• Avg IP per start — correlated with QS% but tells you something distinct: can this pitcher protect the bullpen? An avg IP below 5.0 means even decent starts often end early, handing the game to relievers — a variable the pre-game score doesn't capture.
• Surprise — a pitcher who consistently outperforms their pre-game score (negative surprise, shown in green) is being structurally underrated by the formula; treat their effective difficulty as higher than the number suggests. Persistent positive surprise (red) is the warning sign: they look better on paper than they pitch in practice.

Two starters both scoring 52 look identical on the main page. But if one has 65% QS%, 6.2 avg IP, and a −1.2 surprise while the other has 22% QS%, 4.7 avg IP, and a +0.9 surprise, those are very different environments. The first matchup is a genuine coin flip. The second is riskier than the score alone suggests — and on a borderline combo day, that difference is worth knowing before you commit.