Glossary

1

Home win

X

Draw

2

Away win

The same three codes on every screen: a match card, the board, the radar and a decision receipt all write a pick this way.

WinPIQ probability

WinPIQ's estimate of the chance this outcome happens, based on the information available right now.

It's the starting point for every decision on the page — every other number is compared against it.

A 70% call still loses about 3 times in 10 — it is not a promise.

When it is missing: This figure hasn't been published.

Decision confidence

How sure WinPIQ is about this call, combining model agreement, data and lineup certainty into one label.

It tells you how much weight to put on the call itself, separate from the price.

High confidence with a short price is a different situation than high confidence with a long one.

When it is missing: The confidence read will appear once the analysis is complete.

Analysis strength

How complete the picture behind this call is, expressed in words rather than a raw number.

A strong call and a well-supported one are not always the same thing — this says how well-supported it is.

The lowest reading is 'assessment still forming.'

When it is missing: Not shown until the analysis is complete.

Decision quality

How well a decision was made, graded after the result but not by it: the state you acted on, the price you took, how fresh the information was, and whether you acknowledged the uncertainty.

A good decision can lose and a bad one can win. This is the only number on the page that a single result cannot flatter or punish.

Taking a fair price on an actionable call scores well even when the match is lost.

When it is missing: No graded decision yet.

Decision stability

How likely this recommendation is to stay the same as new information arrives.

A stable call is safer to act on early; a sensitive one is worth rechecking closer to kickoff.

'Very stable', 'sensitive to changes' and 'not stable' are the words used here — never a raw score.

When it is missing: Not shown until the decision has run at least once.

Conviction

How strongly the independent systems — quant, scout, market, models — agree.

It's a second, independent read from the probability itself — the two can and do disagree.

'High' conviction means the systems reached the same call from different angles.

When it is missing: Conviction hasn't been calculated for this selection yet.

Uncertainty

How wide or unstable the reading is: lineup, volatility, draw or upset risk, and how settled the estimate behind it is.

High uncertainty means the same call could look different by kickoff, even if nothing 'goes wrong.'

'Low' uncertainty means the estimate is settled; 'high' means it could still move.

When it is missing: Uncertainty hasn't been calculated for this selection yet.

Current price at the market

The best odds a bookmaker is offering right now, with its source and freshness.

It's the price you'd actually get if you acted right now.

2.00 means a 10-unit stake returns 20 if it wins.

When it is missing: No live price right now.

Fair odds by WinPIQ

The odds WinPIQ's own probability implies, before the operator's margin is added.

It's the benchmark for whether a live price is worth taking.

50% implies fair odds of 2.00.

When it is missing: The fair price hasn't been calculated for this selection.

The lowest price at which this selection still has value

Below this price, the edge no longer covers the risk.

It marks where the price stops being worth the risk.

If the minimum price is 1.85 and the market offers 1.80, the edge is gone.

When it is missing: The minimum price hasn't been set for this selection.

Probability implied by the market's odds

The probability the market's current odds imply, once the bookmaker's own margin is stripped out.

It's the number WinPIQ's own estimate is compared against — the gap between them is where a decision starts.

56% at the market versus WinPIQ's 63% is a meaningful gap.

When it is missing: No market price to compare against right now.

Probability gap

WinPIQ's probability minus the market's probability, in percentage points — only where a market price exists.

It's the plain-language size of the disagreement between WinPIQ and the market.

56% versus 49% is a gap of 7 percentage points.

When it is missing: No market price to compare against.

Model-implied expected return

The mathematical return a bet would carry if WinPIQ's probability is right — a probabilistic calculation, not a promised profit.

It converts the probability gap into the number that actually matters for a stake.

6.4% expected return.

When it is missing: No market price to calculate this against.

Price movement

How the live price has changed since the last check, without guessing why.

A moving price tells you the market is reacting to something, even before the reason is confirmed.

The price fell from 2.10 to 1.92 — the market now rates this result more likely.

When it is missing: No earlier price to compare against yet.

How long the edge might last

An estimate of how long the current edge is likely to remain before the price or the news catches up — never a guarantee.

It turns an opportunity into a timing decision, not just a yes/no one.

About 2 hours, based on how fast this kind of edge has moved before.

When it is missing: Not shown when there's no active edge right now.

Squad readiness

How much of the team's usual starting lineup is expected to be available, in words rather than a raw number.

A missing regular starter can matter far more than a headline number would suggest.

'Minor absences' versus 'significant absences' — named, not just counted.

When it is missing: Team news hasn't been reviewed yet for this match.

Confidence in the expected lineup

How official the expected lineup is: confirmed team news, or WinPIQ's own projection when the team hasn't confirmed yet.

A projected lineup can still change before kickoff — this says how likely it is to hold.

8 of the 11 projected positions are considered likely.

When it is missing: No projected lineup yet for this match.

Expert consensus

How many independent experts picked this result, always shown with the count they were drawn from.

A share without its count can be misleading — 3 of 4 and 30 of 40 are not the same signal.

12 of 20 experts picked a home win.

When it is missing: No expert picks recorded yet for this match.

Analyst rank

How this analyst's picks have scored over time, weighted by sample size and consistency — never by popularity.

A track record separates an analyst worth reading from one who is simply well known.

68% accuracy across 34 of the last 50 predictions.

When it is missing: This analyst's rank will appear once they've made enough picks to score.

Model consensus

How strongly WinPIQ's independent models agree on the same result.

Agreement across independent models is a different kind of signal than any one model being confident.

8 out of 10 models favour the home team.

When it is missing: This reading appears once all models have voted.

Calibration accuracy — lower is better (Brier score)

How closely stated probabilities match what actually happened, across many predictions.

A model can pick winners and still be poorly calibrated — this measures the second thing.

A lower score means the stated probabilities were closer to reality on average.

When it is missing: This score appears once enough predictions have settled.

Log loss — lower is better (Log loss)

A second calibration measure that penalises confident, wrong predictions more heavily than calibration accuracy does.

It catches overconfidence that a simpler calibration score can miss.

A confident, wrong call costs far more here than a cautious, wrong one.

When it is missing: This score appears once enough predictions have settled.

Closing line value (CLV)

Whether the price taken was better than the market's final price before kickoff — a measure of timing, not of the pick itself.

Beating the closing price consistently is one of the more reliable signs of a well-timed process.

Taking 2.05 when the market closed at 1.95 is positive closing-line value.

When it is missing: not published — we do not record closing snapshots

Sample size

The number of predictions, matches or opinions a figure above is based on.

The same percentage means different things over 10 cases and over 500.

34 of the last 50 predictions.

When it is missing: No count available for this figure yet.

Opportunity rank (WinPIQ Score)

A single composite ranking of how strong this opportunity looks relative to others, for comparing across a list.

It's useful for sorting a list; the reasons behind any one row still matter more than the number.

Sorting a watchlist by this number surfaces the strongest opportunities first.

When it is missing: This ranking will appear once enough opportunities have been scored.