Do Pattern Systems Beat Guessing? 100,000 Rounds
No. Four popular rules, each scored on 100,000 fair random rounds, are right about half the time, the same as guessing. The table shows how close, and how far luck alone can push a score.
The four rules, scored
A system is a rule that picks a colour using only the results before it. The tool scores four of them on one fixed run of 100,000 fair rounds. Nothing here predicts anything: the tool only counts how often each rule would have been right. A guess is right half the time, so 50% is the line to beat.
| Rule | What you would do | Rounds scored | Right |
|---|---|---|---|
| Repeat the last result | Pick the colour that just came up | 99,995 | 49.95% |
| Pick the opposite of the last result | Switch colour every round | 99,995 | 50.05% |
| Follow the majority of the last five | Pick the colour that came up three times or more in the last five | 99,995 | 50.26% |
| Fade the majority of the last five | Pick the colour that came up less often in those five | 99,995 | 49.74% |
The first five rounds only supply history, so each rule is scored on 99,995. The run is fixed, so you will see these same numbers every time you open this page. The widest gap from 50% is 0.26 of a percentage point, called a point from here on.
Every rule has a mirror image
Read the scores in pairs. In each round exactly one rule of a pair is right and its twin is wrong, so 49.95% and 50.05% add up to 100%, and so do 50.26% and 49.74%. Whichever rule looks clever in your notes, its twin looks foolish by the same amount. Which of the two comes out ahead is down to luck.
How far can luck move a hit rate?
A hit rate is the share of guesses that are right. A rule with no power at all will still not score exactly 50%. It wobbles, and the usual size of the wobble is called a spread. The spread shrinks as the rounds pile up.
| Rounds | One spread, in rounds | One spread, in points of hit rate | Two spreads, in points |
|---|---|---|---|
| 20 | ± 2.2 | ± 11 | ± 22 |
| 100 | ± 5.0 | ± 5 | ± 10 |
| 1,000 | ± 15.8 | ± 1.6 | ± 3.2 |
| 10,000 | ± 50.0 | ± 0.5 | ± 1.0 |
| 99,995 | ± 158.1 | ± 0.16 | ± 0.32 |
Inputs: pure guessing on fair 50-50 rounds. Points are the spread divided by the rounds, times 100. Luck alone usually keeps a score within two spreads of 50%. At 99,995 rounds that is roughly ± 0.3 of a point, and all four rules sit inside it.
Two lessons come straight from the table.
- A short test shows only a huge edge. At 20 rounds, two spreads is ± 22 points. A rule that gets 14 of 20 right looks like 70%, yet in 20 rounds blind guessing reaches 14 or more 5.77% of the time, about one try in 17. Let enough people try enough rules and one of them will seem to work.
- A small edge hides for a long time. Even at 10,000 rounds, luck can move a score by a full point either way. A real edge of one point would still be hard to tell from luck.
Score a rule of your own
- Write the rule in one sentence before you look at any results, and do not change it halfway.
- Write each guess down before the result appears, so the rule never sees the answer. The colour tricks page gives the full recording method.
- Find your number of rounds in the table. A hit rate inside two spreads of 50% is what luck gives.
- Score the mirror image too. If the rule “works”, its twin “fails” by the same amount, which shows only that the rounds happened to fall one way.
- A score outside the band needs a second, fresh set of rounds before you believe it.
This test uses computer-made rounds. It says nothing about whether any platform’s rounds are fair or random, and a short record copied from an app cannot settle that either. To check a predictor’s calls rather than a rule, use the 100-round log.
The betting side
A rule only picks a colour. How much you bet on it is a separate question, and the usual answer is to bet more after a loss. Double after a loss shows what that does to a bankroll. For the streaks that tempt people into these rules, see streaks. Back to the prediction simulator.
Common questions
Which pattern system is the most accurate?
None. The four rules scored between 49.74% and 50.26% over 99,995 fair rounds, which is what guessing gives. The small gaps between them are luck, and each rule’s mirror image loses exactly what it gains.
Is 50.26% a sign that following the last five works?
No. Over this many rounds, luck alone moves a score by about 0.16 of a point either way (one spread), so 0.26 sits inside the usual band of about 0.3. Its mirror image scored 49.74%.
Why do the scores come in mirror pairs?
Each pair holds a rule and its opposite. On every round exactly one of the two is right, so what one rule gains the other loses. Their scores always total 100 per cent, whatever the rounds were.
Can I test my own rule in the simulator?
Not in the tool, which runs the four rules in the table. Test your own on paper: fix the rule first, write each guess before the result, and compare your score with the luck band for your number of rounds.
Does the 50% line apply to every colour game?
Only to a two-way bet whose sides are equally likely. If your app’s sides are not equally likely, the line moves, and its rules screen should show the chances. We cannot see inside any platform.