How the Iceberg Simulation Predicts a Game
How the Iceberg Simulation Predicts a Game
A headline pick gives you one number: this team wins 58% of the time. Useful — but it's the tip of the iceberg. Underneath is the full range of ways the game could actually unfold, and that's what our Iceberg Simulation surfaces.
What "Monte Carlo" means
A Monte Carlo simulation answers questions about an uncertain event by playing it out at random, thousands of times, and tallying the results. Instead of trying to calculate one perfect answer, you let chance do the work: simulate the game 10,000 times, see how often each thing happens, and those frequencies become your probabilities. The name comes from the casino — it's literally rolling dice, just very fast.
How we run it
For each game, the Iceberg engine starts from the consensus and the market lines, derives each team's expected scoring, then plays the game out 10,000 times using a shape that fits the sport:
- Baseball — runs follow a "negative binomial" distribution (lots of low scores, occasional crooked numbers).
- Hockey and soccer — goals follow a Poisson distribution (rare, clustered events).
- Basketball and football — scoring margins follow a normal bell curve.
Each simulated game produces a final score. After 10,000 of them, we count.
What the distribution tells you
A single win probability can't show you these, but the simulation can:
- Most likely final score — the exact result that came up most often.
- Over/Under probability — in what share of the 10,000 games did the total clear the posted line.
- Cover probability — how often each side beat the spread.
- The shape of the total — the bar chart on each game page is the actual distribution of simulated totals. A tall, narrow peak means a predictable game; a wide, flat spread means anything can happen.
Why a range beats a point
Two games can both have a 9.5-run total projection, but one might be a tight pitchers' duel (most outcomes near 8–10) and the other a volatile slugfest (4 to 14). A point estimate hides that; the distribution shows it. For totals and props especially, knowing the shape of likely outcomes is often more valuable than the single most likely number.
The Iceberg is our own model — so it doesn't get a vote in the independent consensus (that would double-count) — but it's free on every game page, and like every model here, it's graded against real results so you can watch how it does.
FreezyPicks aggregates independent models, sharp-money data, and our own Iceberg simulation into free, graded picks — for entertainment, not betting advice. See today's picks or the full disclaimer. 21+ and where legal.