The Video Assistant · Lesson 1 — The Odds ← Course

The odds — a probability distribution.

A decision isn't one fixed answer — it's whichever answer the VAR judges most likely.

Think of a weather forecast. It doesn't flatly promise rain or no rain — it says something like a 70% chance of rain. That figure is a probability: a way of putting a number on how likely something is, running from 0 (it will never happen) up to 1, or 100% (certain). A 50% chance means it could honestly go either way.

The VAR decides the same way. It doesn't lock onto a single verdict straight away. For each possible call — GOAL or NO GOAL — it weighs the evidence and settles on a probability: maybe 96% GOAL against 4% NO GOAL, or a far closer 54% against 46%. The verdict you finally see — the call — is usually whichever option came out highest.

That is also where the confidence number from the last page comes from: it is just that highest probability. When one option towers over the other — 96% versus 4% — the VAR is sure, and confidence is high. When the two sit nearly level — 54% versus 46% — neither answer is clearly more likely, so the call still gets made, just barely, and confidence is low.

There's one more twist. The call you see is usually the top bar — the most likely option. But the VAR doesn't have to take the top bar every time: a model can also draw from the odds, picking each option in proportion to its height. That's why the very same check, asked twice, can come back differently. How adventurously it draws is a dial you'll meet later.

This is the heart of how an LLM reaches every decision. It never simply "knows" the answer — it rates how likely each possible answer is, then commits to one: usually the most likely, sometimes drawn from the odds.

Drag the evidence from no goal to goal and watch the two probabilities shift. Find the point where they sit so close the call is barely more than a coin toss.

Drag the evidence, or pick a check. Watch the odds — and the call.
no goal goal
GOAL — but only just

Ask for the odds, not just the answer.

The model hands back a probability for each option. The reply is the top one; the confidence is its share. Read the spread and you can see how sure the VAR really is. 

result = var.review(evidence, return_probs=True)

result.probs        # {'goal': 0.58, 'no goal': 0.42}
result.call         # 'goal'   (the top of result.probs)
result.confidence   # 0.58
result.sample()     # 'no goal'   (drawn from the odds — can differ each time)

Underneath, every call is a set of odds.

When one outcome towers over the rest, the call is easy and the confidence is high. When the odds sit close together, the VAR still commits — just barely, with low confidence. The single number you met in The Check was always the height of the tallest bar.

Every LLM answer works this way: a set of odds across the options, with the answer usually the top bar — though it can be drawn from the odds instead. Hold onto that picture — it's where the next ideas begin (drawing more or less adventurously, and how a tall bar can still be wrong).

Next: frame by frame →
▸ Go deeper — where the odds come from optional

From scores to probabilities

For each step, the model emits a raw score — a logit — for every option in its vocabulary. A function called softmax turns those raw scores into clean probabilities that add up to 1 — the full set of odds across the options (its formal name is a probability distribution). The reply is usually the option at the top of those odds, and what we've been calling "confidence" is simply that top probability. Put plainly: confidence here means the model's predicted probability for an answer — not a statistical confidence interval, which is a different idea.

Confidence is not the same as correctness

A high probability means the patterns the model learned point strongly to that answer — not that the answer is true. Models can be confidently wrong or hesitant when right; how well a model's stated confidence matches its real accuracy is called calibration. This gap is the seed of hallucination, a few lessons on.

The odds can be tuned

Those odds aren't fixed before the model picks from them — how sharply they're peaked can itself be turned up or down, which is what makes answers steadier or more varied. That's a whole lesson later; for now, just know the odds can be tuned.

One set of odds per token

A real answer is many tokens long, and the model produces a fresh set of odds for each one, commits to a token, then does it again. The confident-looking sentence you read is a chain of these small bets — each word the most likely pick from its own little set of odds.