Scoring systems in a prediction game — and what they really do to your season
Tipperoo — we build the ad-free prediction game and write down why.
In short:
- 4/3/2 (exact 4, goal difference 3, tendency 2): a narrow spread. The best predictor wins over 34 matchdays. The table is calm and fairly just.
- 6/2/1 (exact 6, goal difference 2, tendency 1): a wide spread. An exact result is worth six correct tendencies here — at 4/3/2 it’s worth two. The table stays open until May, and luck decides more of it.
- Kicktipp’s rule system: the same three levels, but points can be set per outcome (home win, draw, away win) — plus a dedicated draw level with no goal-difference tier.
- The joker (×2) doubles conviction, not risk: it doubles points you already earned. Zero times two is still zero.
The one decision underneath: should the best predictor win, or should the league stay alive for everyone until May? You cannot have both.
The three systems in one sentence
Almost every German prediction game uses the same three hit levels: exact result, goal difference, tendency. All that differs is the numbers behind them — and those decide everything.
| Level | What it means | Example, result 2:1 |
|---|---|---|
| Exact result | predicted exactly right | Prediction 2:1 |
| Goal difference | right winner, right margin | Prediction 3:2 |
| Tendency | right winner, wrong margin | Prediction 4:1 |
| No hit | wrong winner | Prediction 1:1 or 0:2 → 0 points |
On Tipperoo the two presets are called Standard (4/3/2) and Hard (6/2/1), and there is a Kicktipp template that rebuilds Kicktipp’s rule system. You can also set the three values yourself.
And that is the whole kit. Everything that follows is the question of what those numbers do to you in March.
What an exact result really rewards
An exact result is not an achievement you can plan. Tendency you can be good at — anyone who follows football calls more winners correctly over a season than someone rolling dice. An exact score you can only hit.
The reason is structural and needs no statistics: predict 2:1 in a home win and you have to pick, out of every plausible home win — 1:0, 2:0, 2:1, 3:1, 3:2, 4:1 … — the one that actually happens. Even a perfect read of the match (“narrow home win”) leaves half a dozen scorelines open. Two people who read the game exactly as well as each other land on different numbers — and one of them takes the full points.
So the size of the exact tier is really a luck dial. The more points you put on the exact result, the more of your season you hand to chance. That can be exactly right — it should just be a decision, not an accident.
Goal difference: the tier nobody thinks about
The middle tier decides whether your game has three levels or, in truth, one. You only see it when you lay the steps side by side:
| System | Tendency | Goal difference | Exact | The steps |
|---|---|---|---|---|
| Standard | 2 | 3 | 4 | +1, +1 — a staircase |
| Hard | 1 | 2 | 6 | +1, +4 — a step and a cliff |
At 4/3/2 the goal difference, at 3 points, sits right next to the exact result (4). Reading how a team wins — narrowly, comfortably, by two — pays almost as well as hitting the number. That rewards real judgement.
At 6/2/1 the goal difference (2) is a single point above tendency (1) and four points below the exact result (6). The middle tier is all but worthless. “Hard” is not a three-tier game any more; it is a one-tier game with two consolation prizes: exact, or near enough irrelevant.
That is the most important sentence in this article, and it is in nobody’s feature list.
Why 6/2/1 makes a season harder than the name suggests
“Hard” sounds like the system for the experts. It is the system that rewards the experts least.
Work it through on one Bundesliga matchday — nine matches:
- You read the whole matchday right: nine correct tendencies. Under Hard that is 9 points (9 × 1).
- Someone else gets seven matches wrong, lands one exact result, and had their joker on it: 12 points (6 × 2).
One lucky hit beats a perfectly read matchday. And that is not a bug in the system — it is the system.
The same matchday under Standard: nine correct tendencies are 18 points (9 × 2), and the jokered single hit brings 8 (4 × 2). Here the graft wins, and it wins clearly.
So the choice between the two is not about difficulty. It is about variance:
- Narrow spread (4/3/2): the table is stable. The best player leads and stays there. The price: someone 30 points down after twelve matchdays realistically can’t come back — and stops predicting.
- Wide spread (6/2/1): the table keeps tipping over. Nobody is safe, nobody is out. The price: the winner may be whoever’s 3:1 happened to land — and the best predictor in the league will notice.
For a big, mixed league with lots of casual predictors the wide spread is often the socially smarter call: it keeps people in. For a small league of experts it is the fastest way to devalue the win.
The joker: an amplifier, not a gamble
The joker doubles the points for one match of a matchday. On Tipperoo: ×2, once per matchday, on or off.
And then, by matchday two at the latest, somebody in every league says: “Joker on the crazy match.” That is precisely backwards. The joker doubles no risk — it doubles a hit. Zero times two is zero. It belongs on the match you are surest about, not the one with the wildest scoreline. It rewards conviction, not courage.
Where it gets dangerous is in combination with the spread: under Hard a jokered exact hit is 12 points — more than an entire, perfectly read matchday. Under Standard it is 8, and the matchday is still worth 18. Same joker, two completely different seasons.
Two practical rules we wrote down while building it:
- A joker freezes at kickoff. A joker you can still move after kickoff is not a joker, it is an insurance policy. Ours is nailed down from kickoff — like the prediction underneath it.
- Kicktipp’s rule system has no joker. That is not a criticism but a configuration note: in our Kicktipp template it is therefore off by default. Switching it on means deliberately playing a different game.
And what if the joker is on a cup tie that is only settled in extra time or on penalties? Which result counts for your prediction then is something we took apart in its own piece: when a prediction-game result is really final.
The draw problem, and why it needs a tier of its own
A predicted draw has one hit level fewer than a predicted win. That is not taste, it is arithmetic: predict 1:1, and if it ends 2:2 you have the goal difference right by construction — zero is zero. In a draw there is nothing to hit between “exact” and “tendency”. So a draw prediction has two routes to points, and a win prediction has three.
The consequence, if every outcome pays the same: calling a draw is structurally the worse deal. It has one fewer chance at points, at the same price. A league that doesn’t correct for that quietly trains its players to always back the favourite. And a table in which everyone backs the favourite isn’t a table — it’s a copy.
That is exactly what the dedicated draw level is for: it pays back the missing tier. Kicktipp turns this into an explicit recommendation — alongside its standard points rule (the same points for all three outcomes) its help pages carry a “modified 9er rule” in which draws and away wins score higher, recommended for leagues under roughly 50 members, expressly to make always backing the favourite less attractive (Kicktipp help 2/213, as of 13 July 2026).
Ours: the Kicktipp template pays more for a correct draw tendency than for a correct win tendency — for exactly that reason. What is not yet possible: entering your own values per outcome (home / draw / away, the “9er” shape). The engine handles it; the input screen is still being built. Today you set your three values and they apply to every outcome — the way Kicktipp’s standard points rule works. The moment the input ships, we’ll say so here.
Where the systems come from — and why they exist at all
Every one of these systems answers the same social question: how much should luck decide — and should it be free to always back the favourite?
- The three levels (exact / goal difference / tendency) are the German prediction-game tradition. Kicktipp publishes them as its standard points rule: the same points for all three outcomes, in the range of two to four points (help 2/213).
- Points per outcome (the “9er” shape) answer the favourite problem above.
- A wide spread like 6/2/1 answers a different problem: a league that falls apart in November because half of it sees no way back.
- The quota rule is the third answer, and the most elegant: the points for an outcome follow from your own league’s predictions — call the draw when nobody else does and you get more. Important, and we say it plainly: those are your league’s predictions, never a bookmaker’s odds. Ours ships in September; Kicktipp recommends it for leagues from about 15 members up.
- The joker doesn’t come from this tradition at all, but from the newer apps. It is an amplifier, not a correction.
Lay them out side by side and it becomes obvious: point values aren’t the accounting of a prediction game — they are its dramaturgy. They are the only lever a founder has over how a season plays out.
What we built — and where our check ends
We implemented all of them: Standard, Hard, Kicktipp’s rule system, your own values, the joker, plus knockout scoring after 90 minutes or after extra time and penalties. Every one of those rules is pinned by a reference test suite that checks the same cases against the scoring engine and against the app. If one side drifts, the code doesn’t ship.
And now the honest part, because the rest would be worthless without it: we check against Kicktipp’s published rules, not against Kicktipp’s servers. Kicktipp has no single points rule at all — every league there configures its own values. So there is nothing for our numbers to be measured equal to. What there is is a rule system, and a rule system can be rebuilt.
An example where exactly that caught us out. On penalty shootouts we had assumed the winner gets credited one goal: a 2:2 after extra time would have come out as 3:2 on Tipperoo. Kicktipp’s own help pages carry the worked example: Portugal–England, Euro 2004 — 2:2 after extra time, 6:5 on penalties, recorded as 8:7. Every goal counts, including the ones from the shootout. We were wrong. We changed the engine — before a single match had been scored. You can read the example at Kicktipp yourself: help 2/209.
That is the kind of care you never notice in a prediction game — until the matchday it isn’t there. What comes across when you move an existing league, and what honestly doesn’t, is on our Kicktipp alternative page.
Which system fits your league?
No recommendation without the league attached. Four cases, and in each one the reasoning matters more than the number.
| Your league | Take | Why |
|---|---|---|
| 8–15 friends, everyone follows football | Standard 4/3/2 | Narrow spread, three real tiers. The person who predicted best wins — and in a small league that is exactly what you want. |
| Office, mixed, lots of casual players | Standard 4/3/2, joker off | Keep it simple and forgiving. The joker is one more rule for someone to miss — and at work you lose people to every rule that needs explaining. |
| 20+ people, very mixed knowledge | Hard 6/2/1 | The wide spread keeps the table open. Someone 40 points down in November still has a story in January. You pay for it with more luck in the game — deliberately. |
| Big league with the favourite problem | Kicktipp’s rule system — or the quota rule from September | Both aim at the same thing: it should pay to commit against the majority. |
And the rule above all rules: lock the system before matchday 1 and don’t touch it until May. Change the points for draws on matchday 12 and you retroactively change who is leading. No league forgives that — and it is the second most common reason prediction leagues fall apart (the most common is the forgotten matchday).
If you’re planning a league right now: the six decisions ahead of you — matches, people, points, deadline, invites, missed matchdays — are here: How to set up a prediction game. You cannot create your league on Tipperoo yet; we launch with the season. Sign up and we’ll tell you in good time.
Statements about Kicktipp are based on Kicktipp’s publicly accessible help pages (as of 13 July 2026) and serve a factual, truthful comparison. Not affiliated with the DFL, the Bundesliga or the Premier League; Kicktipp is a trademark of its respective owners.