Luck vs Skill: Which Sports Are the Most Unpredictable?

Every punter has had the weekend. Your banker, a side that has dominated the opening weeks of the season, hits the woodwork twice, concedes from a deflected shot and loses 1-0 to a team in the bottom three. You look at the stats and they say your lot were better. The scoreboard says otherwise.

That isn't bad luck in the throwaway sense. It's a built-in feature of football, and it can be measured. With the 2026/27 season a couple of months old and the first "shock" results already piling up, it's a good moment to ask a simple question about luck vs skill in sport: how much of what happens on the pitch is skill, and how much is the bounce of the ball?

You'll get the best hard evidence comparing upsets across football, baseball, basketball and American football, a clear explanation of why low-scoring sports are so much more random, a fresh look at Leicester's 5000/1 title, and practical angles for accumulators, draws and underdogs.

The neatest way to think about this comes from Michael Mauboussin's book The Success Equation: Untangling Skill and Luck in Business, Sports, and Investing (Harvard Business Review Press, 2012). Picture a line. At one end sits pure luck: roulette, the lottery. At the other sits pure skill: a running race, a game of chess. Almost every sport lives somewhere in between.

There are two quick tests for where an activity belongs. First, can you lose on purpose? If you can, there's skill involved, because your choices affect the outcome. You can't deliberately lose at roulette. Second, could you lose to someone worse than you? If so, luck is in the mix. A grandmaster almost never loses to a club player. Manchester City can lose to anyone on a wet Tuesday.

The paradox of skill

Mauboussin's most striking idea is that as the overall level of skill rises, luck matters more to results, not less. When everyone at the top is brilliant, the gaps between them shrink, and the random stuff decides more contests.

It sounds backwards, but think of an elite sprint final where everyone is fast. A slight stumble out of the blocks decides the medals in a way it never would in a school race. That fits neatly with the evidence below, which shows baseball and American football becoming more competitive over time.

Opinions about which sport is "most random" are ten a penny. Proper measurement is rarer. The best comes from a paper by Ben-Naim, Vazquez and Redner titled What is the most competitive sport?, which crunched more than 300,000 regular-season games across over a century of data from five competitions: the English top flight, Major League Baseball, the NHL, the NBA and the NFL.

Their method was elegant. For every game, they rebuilt the league table as it stood that day, then asked: how often did the team with the worse record win? They called that the upset frequency. Draws counted as half a win for each side, and games between teams with no record or identical records were ignored.

Sport	Upset frequency	Better-record side's share
English football	0.452	about 55%
Baseball (MLB)	0.441	about 56%
Basketball (NBA)	0.365	about 63.5%
American football (NFL)	0.364	about 63.5%

Draws count as half a win, so football's figure includes plenty of shared points rather than outright defeats.

The headline is stark. English football was the most upset-prone sport in the study. The side higher up the table effectively takes only around 55% of the spoils. In the NBA and NFL, the team with the better record wins about 63.5% of the time.

Football's draw rate is a big part of that. Draws made up 24.6% of the football games in the data, against 14.4% in ice hockey and just 1.6% in American football. Every draw drags the favourite's share back towards a coin flip.

There's one more twist worth knowing. Over time, the paper found baseball and American football getting more competitive, while English football went the other way, with upsets becoming less frequent. The top of the English game has grown more predictable, not less.

Here's where plenty of people trip up. Look at the final NFL standings and you'll see huge gaps between the best and worst teams. Look at baseball and the records bunch up, with win fractions typically between .400 and .600. You might conclude that the NFL is a league of giants and minnows, and baseball a lottery.

The paper's authors flag exactly this trap. Part of the NFL's spread is simply the short season. If every game were a pure 50/50 coin toss, a team's win fraction would still wobble, and the wobble shrinks as the number of games grows:

Spread from pure luck = 0.5 / √(games played)

NFL             17 games   →  0.121  (12.1 points of win %)
Premier League  38 games   →  0.081
NBA             82 games   →  0.055
MLB            162 games   →  0.039

Call this the luck floor. You only see genuine skill when the real spread of results is clearly wider than it. Baseball's actual spread in the study was 0.084, roughly double its luck floor, so even in the most even-looking sport the long season sorts the good from the bad.

A 38-game Premier League season sits in an awkward middle. The luck floor of 0.081 works out at around three wins either way across a campaign. That's the difference between Europe and mid-table, and it can come down to deflections, penalty calls and the woodwork.

The same logic applies to your own betting. A few weeks of results, good or bad, are mostly noise. If you want to separate the two properly, our guide to form, variance and luck in sports betting goes deeper.

So why is football so random? The single biggest reason is that goals are scarce. When a game has only two or three scoring events, one lucky bounce can decide it. When it has dozens, the better side's quality gets room to show.

You can see this with a simple model. Treat goals as random events (the standard Poisson approach), and keep one team exactly twice as potent as the other. Then change only how many goals the game produces:

Expected goals in the game	Stronger side wins	Draw	Weaker side wins
2.7 (Premier League 2015/16 average)	58.6%	22.9%	18.5%
5.4	71.1%	13.3%	15.5%
27	95.1%	1.7%	3.2%

Same 2:1 superiority in every row. The 27-goal row simply illustrates a high-scoring game.

Same gap in quality, completely different outcomes. At a football-like scoring rate the favourite wins under 60% of the time. Pump up the scoring and the same favourite becomes almost certain. That's one big reason the NBA, with its endless possessions, sits at the predictable end of the table.

A side expected to score three goals against a team managing 0.6 still fails to win roughly one game in seven.

Even mismatches leak

Run individual fixtures through the same model and the numbers are humbling for anyone who loves a short-priced banker:

Expected goals     Win    Draw   Lose   Fair price to win
1.35 v 1.35       37.1%  25.8%  37.1%   2.70
1.8  v 1.0        56.1%  23.1%  20.8%   1.78
2.5  v 0.8        74.4%  15.7%   9.9%   1.34
3.0  v 0.6        85.0%  10.4%   4.7%   1.18

That 3.0 v 0.6 mismatch is about as lopsided as football gets, and the favourite still drops points 15% of the time. In tight games the draw sits between 21% and 26%, right in line with the 24.6% the upset study found in real football.

No discussion of football's randomness gets far without Leicester City. At the start of 2015/16 they were reported at 5000/1 to win the Premier League. They had spent much of the previous season at the bottom before rallying to finish 14th.

What followed is folklore. Claudio Ranieri's side opened with a 4-2 win over Sunderland on 8 August 2015, Jamie Vardy scored in a record 11 consecutive Premier League games on his way to 24 league goals, and Riyad Mahrez became the first African winner of the PFA Players' Player of the Year. The title was confirmed on 2 May 2016 when Chelsea and Tottenham drew 2-2 at Stamford Bridge, with Leicester still having two games to play. It was the first top-flight title in the club's history; their previous best was runners-up in 1929. It also sits among the greatest betting coups in history, at least for the few who had a fiver on at the start.

Read it the right way

We think Leicester proves something subtler than "football is random". Their players were genuinely excellent that season. What the low-scoring nature of the sport allowed was for a very good season from an unfancied side to collide with a poor one from the giants. Defending champions Chelsea sacked José Mourinho in December with the club 16th and finished 10th, the lowest finish by a Premier League title holder at that point. Eden Hazard didn't score a league goal until late April.

In basketball, where the better-record team wins about 63.5% of the time, that sort of collision is much harder to engineer. In football it can happen, once in a generation, and the magic of the cup runs in our greatest FA Cup giant killings comes from the same source: one match, few goals, plenty of room for the unexpected.

Just don't mistake 5000/1 for a fair probability. Those prices are outright quotes on rank outsiders, and long shots are where favourite-longshot bias bites hardest.

Putting the evidence together, here's how the major sports stack up for a punter. Football, baseball, basketball and the NFL lean on the upset study; ice hockey and tennis are read from the structure of each sport.

Sport	Predictability	Why	What it means for bettors
Football	Most upset-prone	Few goals, draws in about a quarter of games	Favourites leak; the draw is a real outcome
Baseball	Nearly as upset-prone	Even single games, long season sorts it out	Single games are fickle, season-long markets less so
Ice hockey	Draw-heavy	Ties in 14.4% of games in the study	Never assume the favourite simply wins in 60 minutes
American football	Favourites reliable per game	Better-record side wins about 63.5%, only 17 games	Standings exaggerate gaps; single-game favourites hold up
Basketball	Most predictable	Many scoring events, no draws	Favourites dependable; the debate moves to the handicap
Tennis	Leans towards skill	One-on-one, lots of points, no draws	Strong players get many chances to assert their class

The pattern is clear. Scoring frequency and draws are the two big luck dials. Turn either up and the better side's grip loosens.

None of this means one sport is easier to beat than another. Bookmakers price randomness in. A football favourite at 1.60 is priced as a football favourite, not a basketball one. What it changes is how you should build bets and how much variance you should expect.

Accumulators

An acca only lands if every leg wins, so the probabilities multiply:

Five legs at 80% each:  0.8 ^ 5 = 32.8%
Five legs at 60% each:  0.6 ^ 5 =  7.8%
Five legs at 50% each:  0.5 ^ 5 =  3.1%

Football favourites carry an extra failure point that basketball favourites don't: the draw. That's why five "certainties" from the Saturday 3pm kick-offs come in far less often than they feel like they should. Our breakdown of why most parlays lose covers how the bookmaker's margin compounds with every leg you add, and our football accumulator strategy guide shows how to build a weekend acca that respects the draw.

Underdogs

Low scoring makes underdogs genuinely live in football. A side given 18.5% in the model above wins nearly one game in five. That makes football underdog prices worth comparing seriously, but it's a reason to look harder at the price, not an invitation to chase big odds for the thrill of it.

Draws

A quarter of football matches end level. That isn't a freak result, it's a structural part of the sport, and it's most likely when two evenly matched sides meet. If you've never built the draw into your thinking, our draw betting strategy guide is the place to start.

Football sits at the luck-heavy end of the major sports, and we think that's a reason to bet smarter rather than to give up. These are our angles:

• Respect the draw. Around a quarter of football games finish level, so any 1X2 bet on a tight fixture should account for it, and draw-no-bet or double chance deserve a look when you rate a side but fear a stalemate.
• Keep football accas short. Every extra leg multiplies in another chance of a draw or a smash-and-grab. Two or three well-priced legs beat a seven-fold of "bankers".
• Take football underdogs seriously. The weaker side wins often enough that their prices deserve proper comparison, particularly when the favourite is short.
• Back favourites where the scoring is high. Basketball and, game by game, the NFL reward the stronger side far more reliably. If you like favourites, that's where they behave.
• Judge on xG and chances, not results. In a sport where the better team takes only about 55% of the spoils, a run of results tells you less than the underlying performances, and the best bets often come from regression to the mean when a lucky or unlucky run is priced as if it will last.
• Don't read a few weeks as proof. The luck floor shows how wide random swings can be over a 38-game season, never mind a fortnight. Give any team, system or tipster a proper sample, because sample size and yield are what separate a good tipster from a lucky one.
• Treat big outsiders as entertainment. Leicester happened, but 5000/1 shots tend to carry the fattest margins in the book.

Whatever you back, stake to a fixed, sensible unit so the luck in the short run never decides your bankroll.

• Regression to the mean: the most profitable idea in football betting
• How to tell if a tipster is actually good: sample size, yield and luck
• Favourite-longshot bias: why bettors love underdogs and lose
• Draw betting strategy: how to profit from football's most overlooked market
• Why most parlays lose: the maths behind accumulator strategy

18+. Please gamble responsibly. For free, confidential support visit BeGambleAware.org.