Methodology

Sports projection methodology
for NBA and MLB research.

Courtside projections are designed for decision support: clear assumptions, transparent ranges, and enough context to challenge the model instead of blindly trusting it.

What the numbers suggest
ShaiGilgeous-Alexander
Shai Gilgeous-Alexander · Points·-110·projects 33.4 pts
vs Denver Nuggets·Wed, May 13, 10:00 PM EDT
VerdictLean Over+7.5% · strong
Implied
50.0%
de-vigged
Model
57.5%
med conf
EV / $100
+$9.77
expected return
Kelly suggests
$20 / $1k
per $1k bankroll
−5%0%5%10%
+7.5%solid
Not financial advice · probability estimates, not predictions · bet responsibly
● Live componentProjection feeding a live verdict
NBA
Short-term + long-term player projections
MLB
Recent-form projection + Statcast context
DFS
Fantasy point scoring + ceiling ranges
Betting
Projection inputs feed verdict cards
Primary keyword
sports projection methodology
Search intent
Understand how Courtside creates player projections, ranges, and model context for sports research.
Updated
May 30, 2026
How it works4 steps

From question to inspectable answer.

  1. I
    Step 1
    Baseline

    Start with what a player normally does.

    A long-run baseline uses career performance, age, and league context. That keeps one hot or cold stretch from overpowering the projection.

    Example

    “What is this player's long-term baseline projection for points and assists?”

  2. II
    Step 2
    Update with recency

    Recent form changes the median.

    Newer games get more weight while the baseline still matters. Role changes, injuries, and rotation shifts can move the projection.

    Example

    “How much does the last 10 games move this prop projection vs season average?”

  3. III
    Step 3
    Add uncertainty

    Floor and ceiling, never one number.

    Courtside publishes a projection range, not just one number. The range comes from past player volatility and game-specific factors.

    Example

    “Show floor, median, and ceiling for this NBA points projection.”

  4. IV
    Step 4
    Anchor and audit

    Market anchoring + cited drivers.

    Where market lines exist, a blend reduces drift. Every projection can be inspected for the inputs that moved it most.

    Example

    “Why is this DFS projection higher than the player's season average?”

What it doesCapabilities

Three layers, one conversation.

01 / Baseline

A projection starts with what a player usually is.

Courtside uses historical performance and role context as a baseline before recent form gets a vote. This reduces the risk of overreacting to one noisy game.

  • I

    Longer samples help stabilize rate stats and efficiency assumptions.

  • II

    Age, league context, and regression are considered where applicable.

  • III

    Role and minutes or playing time remain central to the final range.

02 / Recent form

Short-term signal matters, but it has to be weighted.

Recent games can reveal usage changes, injuries, matchup adjustments, or lineup shifts. Courtside treats that signal as useful but uncertain.

  • I

    Recent games get more influence without discarding the longer-term baseline.

  • II

    Market context can anchor DFS and prop expectations when available.

  • III

    Floor and ceiling ranges communicate uncertainty better than a single number.

03 / Explainability

The model is most useful when you can argue with it.

Courtside exposes the projection, the range, and the surrounding evidence so an analyst can ask why the number moved or what would change it.

  • I

    Ask for drivers behind a projection increase or decrease.

  • II

    Compare recent-form projections against season-long baselines.

  • III

    Use citations and rich components to inspect the supporting data.

GlossaryTerms used throughout this page

A short dictionary for sports projection methodology.

01
Recent-form average
A recent-form average where newer games count more and older games fade gradually instead of disappearing all at once.
02
Baseline
A longer-term estimate of a player's usual level. It helps protect the projection from overreacting to a small hot or cold stretch.
03
Market anchor
Blending a small share of the betting market's implied expectation into a projection. It can steady the estimate without overriding the model.
04
Floor / ceiling
A practical low-end and high-end outcome range. It helps show risk and upside instead of pretending one number is exact.
Common questionsAnswered

Frequently asked, cleanly answered.

Q1.

Does Courtside provide projection ranges?

Yes. Courtside can show projected outcomes with floor and ceiling context, not only a median estimate.

Q2.

Are projections the same for DFS and betting?

They share underlying player expectations, but DFS converts stats into scoring rules while betting verdicts compare probability against market price.

Q3.

Can I inspect why a projection changed?

Yes. You can ask Courtside to explain the drivers behind a projection, including role, matchup, recent form, and market context.

Q4.

How does the model handle small samples?

Short samples are pulled toward the baseline through regression-to-the-mean. As a player accumulates plate appearances or minutes, the recent-form signal carries more weight.

ContinueRelated research

Adjacent reading for the next step.

Read next

DFS lineup methodology

How projections become constrained rosters.

Open
Read next

Bet verdict methodology

How projections feed edge, expected value, and recommendation language.

Open
Read next

NBA player prop research

See the projection feeding a live prop research workflow.

Open

Stop dashboard-hopping.
Ask a question.

Start askingJoin Discord

Join the Discord for pick notes, slate questions, product drops, and smarter NBA and MLB conversation.