A striker averaging 0.60 goals per 90 minutes can look attractive in an anytime goalscorer market. A winger averaging 3.5 shots per 90 may look equally appealing for Over 2.5 Shots.
But neither statistic answers the most important question:
How long is the player actually expected to be on the pitch?
A player expected to complete 85 minutes has far more time to accumulate shots and scoring opportunities than one likely to be substituted after 60 minutes. A substitute expected to enter in the 75th minute may have an excellent goals-per-90 rate, but only 15–20 minutes in which to convert that rate into an actual goal.
Player-minute analysis is therefore one of the most important adjustments when forecasting:
- anytime goalscorers;
- first goalscorers;
- total player shots;
- shots on target;
- player goal totals.
The basic principle is straightforward:
Player rates describe how often events occur while the player is on the pitch. Expected minutes determine how much opportunity the player is likely to receive in the next match.
This is why a useful player prediction should estimate minutes first and statistical output second.
Why Player Minutes Matter So Much?
Football player props are opportunity-dependent.
A player cannot register a shot after being substituted. He cannot score from the bench. If he enters with only 12 minutes remaining, his historical 90-minute numbers cannot simply be treated as though he has another full match to reproduce them.
This matters especially for goals because scoring is a relatively infrequent event. A player’s underlying scoring ability may be strong, but cutting his playing time from approximately 90 minutes to 45 minutes removes roughly half of his available exposure before tactical context is considered.
Shots behave similarly, although usually at higher frequencies. A forward who attempts four shots per 90 minutes may be well suited to an Over 2.5 Shots line when expected to play almost the full match, but the same line becomes harder to reach if the manager routinely substitutes him around the 60th minute.
Research on football substitutions also shows why playing time should not be treated as completely linear. Substitutes can produce different per-minute physical and technical output from starters because they enter with different tactical instructions, energy levels and match states. Studies have found substitutes capable of producing elevated relative physical output and, depending on position and context, changes in technical actions related to attacking performance.
So minutes are essential, but minutes × per-90 rate is a starting estimate rather than a complete prediction model.
The Three Numbers You Need Before Predicting a Player Market
For most goalscorer and shots predictions, start with three separate pieces of information.
1. Probability the Player Starts
Do not assume a player starts simply because he is normally first choice.
Consider:
- recent starting line-ups;
- injuries or fitness concerns;
- suspension returns;
- competition priority;
- fixture congestion;
- rotation patterns;
- expected formation.
A striker with an 85% starting probability should be analysed differently from one virtually certain to start.
2. Expected Minutes If He Starts
Some starters regularly complete 85–90 minutes. Others are routinely removed between 55 and 70 minutes.
This can depend on:
- position;
- player fitness;
- tactical role;
- age;
- competition schedule;
- bench quality;
- whether the team is leading or losing.
3. Expected Minutes If He Is a Substitute
If the player does not start, when is he normally introduced?
There is a meaningful difference between a player commonly appearing around:
- 46–55 minutes;
- 60–70 minutes;
- 75–85 minutes.
These scenarios create very different scoring and shooting exposure.
Calculate Expected Minutes Before Expected Shots
A practical model can combine starting and substitute scenarios.
Suppose a forward has:
- 80% probability of starting;
- 72 expected minutes when starting;
- 20% probability of beginning on the bench;
- 24 expected minutes when used as a substitute.
His approximate expected playing time is:
(0.80 × 72) + (0.20 × 24)
= 57.6 + 4.8
= 62.4 expected minutes
This is much more useful than simply assuming 90 minutes because his name appears in the predicted XI.
The same framework can be extended when there is also a probability that he does not play at all.
For example:
| Scenario | Probability | Minutes |
| Starts | 75% | 75 |
| Substitute appearance | 20% | 22 |
| Does not play | 5% | 0 |
Expected minutes:
(0.75 × 75) + (0.20 × 22) + (0.05 × 0)
= 60.65 minutes
This becomes the exposure estimate used for the next stage of the player prediction.
How Minutes Affect Shots Predictions?
Suppose a winger averages:
3.6 shots per 90 minutes
and is expected to play:
65 minutes
A simple minutes-adjusted estimate is:
3.6 × (65 ÷ 90)
= 2.6 expected shots
That does not mean the player will record exactly 2.6 shots. It means that, under a simple constant-rate assumption, his average opportunity would be around that level.
The distinction matters when the sportsbook line is:
Over 2.5 Shots
A raw 3.6 shots-per-90 statistic initially looks comfortably above the line.
Once expected minutes are incorporated, the expected value becomes approximately 2.6, almost directly on the threshold.
A player who looks like a strong Over candidate from his per-90 number may therefore become marginal after substitution risk is incorporated.
This is one reason NaijaScore9’s player shots betting analysis should focus on expected role and minutes rather than simply ranking players by shots per 90.
A Better Example: Same Player, Different Minutes
Assume the same player averages 4.0 shots per 90.
| Expected minutes | Minutes-adjusted shot expectation |
| 90 | 4.00 |
| 80 | 3.56 |
| 70 | 3.11 |
| 60 | 2.67 |
| 50 | 2.22 |
| 30 | 1.33 |
The player’s underlying rate has not changed.
Only his expected playing time has changed.
Yet a betting line of Over 2.5 Shots moves from looking comfortably achievable at 80–90 expected minutes to considerably less attractive around 50 minutes.
That demonstrates why predicted line-ups alone are insufficient. Knowing that a player starts is only half the question; knowing how long he is likely to remain on the pitch is the other half.
Shots on Target Are Even More Sensitive to Minutes
Shots on target occur less frequently than total shots for most players, so losing 20 or 30 expected minutes can have a significant effect on threshold probabilities.
Suppose a striker averages:
- 3.2 total shots per 90;
- 1.4 shots on target per 90.
At 90 expected minutes, the raw shot-on-target expectation is 1.4.
At 60 expected minutes:
1.4 × 60 ÷ 90 = 0.93
At 45 expected minutes:
1.4 × 45 ÷ 90 = 0.70
A market such as 1+ shot on target may still be realistic, but a 2+ shots on target prediction becomes increasingly demanding as the expected playing time falls.
The official definition of a shot on target can also matter for settlement. Bet365 and Betfair, for example, specify what qualifies as a shot on target and use official data-provider definitions for these player-statistic markets.
For that reason, a shot’s prediction should consider both the player’s opportunity and the operator’s settlement definition.
How Minutes Affect Anytime Goalscorer Predictions?
Goalscorer analysis requires a slightly different approach because a goal is a binary event for an anytime scorer market: the player either scores at least once or does not.
Suppose a striker has an underlying expected-goals rate of:
0.60 xG per 90
and is expected to play 60 minutes.
A simple proportional expected-goals estimate becomes:
0.60 × 60 ÷ 90 = 0.40 xG
If we use a basic Poisson approximation, the probability of scoring at least once from an expected-goal value of 0.40 is:
1 − e⁻⁰·⁴⁰ ≈ 33%
If the same player were expected to complete 90 minutes:
1 − e⁻⁰·⁶⁰ ≈ 45%
So reducing expected playing time from 90 to 60 minutes materially changes the estimated anytime-scoring probability even though the player’s per-90 attacking quality remains unchanged.
This is an illustrative model, not a guarantee or a complete goalscorer formula. Penalties, position, opposition, game state and shot quality still need to be incorporated.
For match-specific selections, NaijaScore9’s anytime goalscorer tips should therefore be evaluated using realistic expected minutes rather than full-match statistics by default.
Why Goals per 90 Can Mislead?
Suppose Player A has:
0.75 goals per 90
and Player B has:
0.50 goals per 90
At first glance, Player A appears clearly stronger.
But now consider expected playing time:
- Player A: 35 expected minutes;
- Player B: 82 expected minutes.
A simple minutes-adjusted scoring-rate estimate gives:
Player A
0.75 × 35 ÷ 90 = 0.292
Player B
0.50 × 82 ÷ 90 = 0.456
Player A has the better rate.
Player B has the greater expected scoring exposure.
This is why player leaderboards based only on goals per 90 should not be copied directly into goalscorer predictions.
A high per-90 rate can come from a relatively small number of minutes, substitute appearances or a short period of unusual finishing. The rate tells you something about performance while playing; it does not tell you how much the player is expected to play next.
Starting Status Matters Differently Across Bookmakers
This is one of the most practical reasons to check player minutes and starting status before placing player-market selections: sportsbook participation rules are not identical.
For example, Bet365’s current rules state that its pre-match Player to Score at Any Time selections are void when the chosen player does not start, outside specified rolling-substitution competitions. Its First/Last Player to Score rules also require the selected player to start.
Betfair currently uses a different structure for several markets. Its Anytime Scorer rules generally regard a player as a participant if he enters at any stage of normal time plus stoppage time, while its player shots and shots-on-target rules similarly treat an appearance from the bench as participation.
That means a predicted substitute can create different practical exposure depending on where the selection is offered.
For Nigerian users comparing player markets, do not assume:
“If he doesn’t start, every bookmaker refunds the bet.”
Check the exact current settlement rules first.
This is especially important before using first goalscorer predictions, because first-scorer participation rules can be stricter than anytime-scorer rules.
First Goalscorer Is More Sensitive to Starting Minutes
Anytime goalscorer and first goalscorer are not identical minute problems.
A substitute introduced after 65 minutes may still have time to score and potentially win an anytime market under an operator whose rules keep substitute selections active.
But if the match already contains a goal before he enters, he obviously cannot become the actual first goalscorer.
Some bookmaker rules explicitly account for this situation. Betfair, for instance, voids first/last goalscorer bets on players who have not entered before the first goal in circumstances specified by its current rules.
When estimating first-goalscorer probability, consider:
Probability player is on pitch before first goal × probability his team scores first × player’s share of team scoring opportunities
That makes starting status particularly important.
A high-quality substitute may be a perfectly reasonable anytime-scoring candidate while being a poor first-goalscorer candidate.
Expected Minutes Are Not Always the Same as Recent Minutes
A common mistake is taking a player’s last five appearances and calculating an average.
Suppose the player logged:
90, 88, 85, 27, 19 minutes
The simple average is:
61.8 minutes
But what actually happened?
Perhaps:
- the first three matches were starts;
- the final two were substitute appearances after returning from injury.
Or perhaps the opposite occurred: he began the period as a substitute and has now regained a starting position.
The historical average can therefore describe a role that no longer exists.
Expected minutes should be forward-looking.
Ask:
- Is he expected to start this match?
- Is he now fully fit?
- Has his position changed?
- Is another starter returning?
- Is this a stronger or weaker competition?
- Is rotation expected?
- Does the manager normally remove him when protecting a lead?
Recent minutes are evidence. They are not automatically the forecast.
Substitution Patterns Matter More Than the Average
Two players can both average 70 minutes per start and have different substitution risk.
Player A
Usually completes 90 minutes but was substituted twice early because of tactical situations.
Player B
It is almost automatically removed between 65 and 72 minutes in every match.
Their average can look similar, but Player B has a much more predictable ceiling.
This distinction becomes useful for shots overs.
If a player needs three shots to win a market and is almost always removed by minute 68, the analyst knows that he has a fairly defined opportunity window.
If another player has a realistic chance of playing the full 90, his upper-tail opportunity is larger even if the recent averages are similar.
Game State Changes Expected Minutes
Expected minutes should not always be one fixed number because substitution timing often depends on the score.
Imagine a star striker.
If his team is:
Winning 3–0 after 60 minutes: he may be removed early.
If the match is:
1–1 after 75 minutes: he may stay on.
If his team is:
Losing 1–0: the manager may leave him on and add another attacker.
A better model can therefore use scenarios.
| Match scenario | Estimated minutes |
| Team comfortably ahead | 65 |
| Close game | 82 |
| Team chasing result | 90 |
If the opponent is weak and the player’s team is heavily favoured, there is a subtle trade-off.
The favourable matchup may increase his expected scoring and shooting opportunities per minute, while the greater probability of an early comfortable lead may reduce his expected total minutes.
Good player modelling considers both effects.
Fixture Congestion Can Reduce Expected Minutes Without Removing the Player From the XI
Rotation analysis should not stop at whether a player starts.
During a congested schedule, a manager can select a first-choice striker but plan to replace him after approximately 60 minutes because another match follows three days later.
That creates a player who is:
- officially starting;
- still likely to be popular in goalscorer markets;
- nevertheless carrying reduced minute expectations.
Research on football workload and substitution behaviour supports treating playing time as a meaningful component of physical exposure rather than assuming every starter receives a full match.
NaijaScore9’s guide on how fixture congestion affects football match probabilities explains the broader schedule effect. For player props, the key additional question is:
Will congestion change this specific player’s expected minutes?
Penalty Takers Need Separate Treatment
Minutes are especially valuable for penalty takers because penalty opportunities exist only while the player is on the field.
Suppose a striker takes all penalties when playing.
If he is substituted in the 65th minute, any penalty awarded afterwards cannot help his anytime-goalscorer selection.
A replacement penalty taker may instead receive the opportunity.
This means expected minutes affect not only open-play shot volume but also a player’s share of set-piece scoring opportunities.
When modelling a goalscorer, check:
- first-choice penalty taker;
- second-choice penalty taker;
- whether both are likely to start;
- substitution order;
- whether penalty responsibility has recently changed.
A player’s scoring probability can decline meaningfully when he is expected to leave before a large part of the match has been played.
Position Changes Can Make Historical Per-90 Numbers Misleading
Minutes alone cannot correct a role change.
Suppose a winger averaged 3.2 shots per 90 while regularly playing as an inside forward.
He is now expected to play:
- wider;
- deeper;
- as a wing-back.
Even if he still receives 80 minutes, the old shot rate may no longer be appropriate.
The same issue appears when a central midfielder moves into an attacking midfield role or a striker is shifted to the wing.
The correct process is therefore:
Role-adjusted rate × expected minutes
rather than:
Historical overall rate × expected minutes
Football technical performance can vary substantially with match context and tactical environment, and research has cautioned against interpreting shots and other technical actions without that context.
Why Substitutes Cannot Simply Be Given Their Full-Match Per-90 Rate?
A substitute entering for the final 20 minutes is playing in a different football environment from a starter playing the opening 20.
Late in a match:
- one team may be chasing;
- the opponent may be defending deeply;
- players may be fatigued;
- transitions can increase;
- the score may already be comfortable;
- the substitute may have explicit instructions to attack.
Research has found that substitutes can produce greater relative high-speed output per minute than players who have been on the pitch for longer periods.
Therefore, simply taking a substitute’s season-long shots per 90 and multiplying it by 20/90 can miss the effect of his usual game state.
A better approach is to separate:
- performances as a starter;
- performances as a substitute;
- minutes entered;
- score at entry;
- tactical role after entering.
The sample will be smaller, so it should not be overinterpreted, but it provides useful context.
Which Statistics Should Be Adjusted for Minutes?
For goalscorer predictions, useful rate statistics include:
- goals per 90;
- non-penalty goals per 90;
- expected goals per 90;
- shots inside the box per 90;
- big-chance involvement;
- penalty share.
For shots predictions, focus more on:
- shots per 90;
- shots on target per 90;
- shots inside the box;
- average shot locations;
- team shot share;
- role and formation.
But every rate should eventually be combined with expected playing time.
A player averaging 4.5 shots per 90 is not automatically better for a shots market than someone averaging 3.4 if the first player is expected to receive 50 minutes and the second 88.
Do Not Confuse Expected Shots With Probability of Clearing a Line?
Suppose your model estimates:
2.7 shots
for a player whose market line is:
Over 2.5 Shots
It is tempting to interpret this as a strong Over.
But an average expectation does not tell you the complete distribution.
The player might produce:
- 0 shots in some games;
- 1–2 frequently;
- 4–5 in others.
To estimate the probability of recording at least three shots, you need a distribution around the expected value, not just the mean itself.
The same applies to shots on target.
This is why an advanced player model should eventually estimate:
P(shots ≥ required line)
rather than only:
expected shots = 2.7
The expected-minute calculation remains the foundation because it determines how much opportunity feeds into that distribution.
A Practical Player-Minutes Prediction Example
Consider a fictional forward, Player A.
His recent underlying numbers are:
- 3.8 shots per 90;
- 1.5 shots on target per 90;
- 0.55 xG per 90;
- first-choice penalty taker.
The available player markets are:
- Over 2.5 Shots;
- 1+ Shot on Target;
- Anytime Goalscorer.
Scenario 1: Expected to Play 88 Minutes
Simple minutes adjustment:
Shots
3.8 × 88/90 = 3.72
Shots on target
1.5 × 88/90 = 1.47
xG
0.55 × 88/90 = 0.54
The player’s exposure remains close to his full-match rates.
Scenario 2: Expected to Play 62 Minutes
Shots
3.8 × 62/90 = 2.62
Shots on target
1.5 × 62/90 = 1.03
xG
0.55 × 62/90 = 0.38
The player has exactly the same per-90 profile but a substantially weaker case for the Over 2.5 Shots line.
Scenario 3: Expected to Start on the Bench
Suppose he has only a 35% probability of starting and otherwise enters around minute 70.
His overall expected minutes could fall below 40.
At that point, using the full-season per-90 numbers without a large minutes adjustment would materially overstate his opportunities.
This example captures the central lesson:
The player’s quality did not change. His opportunity changed.
A Practical Checklist Before Making Goalscorer or Shots Predictions
Before publishing or using a player-market prediction, verify:
| Question | Why it matters |
| Is the player expected to start? | Determines initial exposure |
| What is his start probability? | Predicted line-ups are uncertain |
| How many minutes does he usually play when starting? | Sets the likely opportunity window |
| When is he normally substituted? | Identifies a practical minute ceiling |
| What happens when his team leads? | Can cause earlier substitution |
| What happens when his team trails? | May extend minutes or increase attacking role |
| Is fixture congestion involved? | Can reduce minutes |
| Is he fully fit? | Fitness can limit playing time |
| Is his tactical position unchanged? | Historical rates may otherwise mislead |
| Does he take penalties/set pieces? | Increases scoring opportunity while on field |
| What are his shots/xG rates per 90? | Provides the underlying event rate |
| Has the rate been adjusted for expected minutes? | Converts performance into next-match exposure |
| What are the bookmaker participation rules? | Determines whether a substitute is active or void |
| What probability does the price imply? | Allows value comparison |
After estimating the final player probability, the NaijaScore9 implied probability calculator can be used to compare your estimate with the break-even percentage represented by the available odds.
Common Player-Minutes Mistakes
Assuming Every Starter Plays 90 Minutes
Many attacking players have predictable substitution patterns. Starting confirms the opening role, not the full-match duration.
Using Per-90 Statistics Without Adjusting Them
Per-90 numbers standardise historical performance. They do not guarantee 90 future minutes.
Ignoring the Chance the Player Is Benched
A player with an uncertain starting position should have his substitute scenario incorporated before the market is evaluated.
Ignoring Game State
Heavy favourites can offer excellent attacking opportunities while also creating early-substitution risk if the match becomes comfortable.
Treating Substitute Rates as Identical to Starting Rates
Substitutes often enter different tactical and physical match conditions.
Ignoring Penalty-Taker Substitution
A player cannot benefit from a late penalty if he has already left the pitch.
Looking Only at Goals
Goals can fluctuate substantially over short periods. Shot volume, expected goals, role and playing time help explain whether recent scoring is supported by repeatable opportunity.
Ignoring Bookmaker Rules
Goalscorer and shots participation rules can differ by operator and market. Bet365 and Betfair currently illustrate that players who do not start can be treated differently depending on the product.
Conclusion
Player-minute analysis solves one of the biggest weaknesses in simplistic goalscorer and shots predictions: per-90 statistics assume a standardised amount of playing time, while actual football players rarely receive exactly 90 minutes every match.
A forward averaging 0.60 expected goals per 90 is not carrying the same scoring probability when projected for 45 minutes as when projected for 88. Likewise, a winger averaging four shots per 90 can move from a strong Over 2.5 Shots candidate to a marginal one when his expected playing time falls to around an hour.
The correct order of analysis is therefore:
starting probability → expected minutes → tactical role → per-minute attacking rate → opponent adjustment → market probability → available price.
Expected minutes should also be scenario-based where necessary. A player may stay on longer when his team needs a goal, leave early when the game is comfortable or have reduced minutes because another important fixture follows.
Most importantly, separate player quality from player opportunity. A high-quality scorer with limited expected minutes can have a lower match-specific scoring probability than a slightly weaker forward expected to play almost the entire match.
For NaijaScore9’s broader forecasting framework, use the football prediction methodology alongside match-specific line-ups, player roles and current team information.