Diagnostic Engine

Draw Fairness & Mechanical Bias

Chi-Square (χ²) screening across true 1st Prize results only. These cards are best used to spot follow-up candidates across time, series, and digit positions, not as standalone proof of mechanical bias.

Method note: this screen enforces a strict Minimum N policy (N ≥ 100) and applies Benjamini-Hochberg FDR correction.

A. Time Segmentation

Does the distribution heavily skew during a specific era?

2023-Present

N=1,273χ²(9)=33.97p_adj=2.71e-4V=0.054 (negligible)
Significant deviation from uniform model (FDR 5%)

2017-2019

N=969χ²(9)=20.59p_adj=2.19e-2V=0.049 (negligible)
Significant deviation from uniform model (FDR 5%)

2020-2022

N=623χ²(9)=11.72p_adj=2.30e-1V=0.046 (negligible)
Not significant after correction

B. Per-Series Isolation

Is a specific physical lottery machine generating skewed results?

NIRMAL WEEKLY LOTTERY

N=358χ²(9)=27.31p_adj=9.95e-3V=0.092 (negligible)
Significant deviation from uniform model (FDR 5%)

STHREE SAKTHI

N=413χ²(9)=17.39p_adj=1.72e-1V=0.068 (negligible)
Not significant after correction

KARUNYA

N=402χ²(9)=14.07p_adj=3.01e-1V=0.062 (negligible)
Not significant after correction

KARUNYA PLUS

N=413χ²(9)=13.27p_adj=3.01e-1V=0.060 (negligible)
Not significant after correction

POURNAMI

N=149χ²(9)=11.87p_adj=3.21e-1V=0.094 (negligible)
Not significant after correction

FIFTY-FIFTY

N=137χ²(9)=11.54p_adj=3.21e-1V=0.097 (negligible)
Not significant after correction

AKSHAYA

N=352χ²(9)=10.73p_adj=3.37e-1V=0.058 (negligible)
Not significant after correction

WIN-WIN

N=351χ²(9)=1.68p_adj=9.96e-1V=0.023 (negligible)
Not significant after correction

BHAGYATHARA

N < 100
Insufficient data for reliable statistical testing (N = 62).

DHANALEKSHMI

N < 100
Insufficient data for reliable statistical testing (N = 61).

SAMRUDHI

N < 100
Insufficient data for reliable statistical testing (N = 64).

SUVARNA KERALAM

N < 100
Insufficient data for reliable statistical testing (N = 60).

C. Position Variance

Are all drum rotors behaving the same way, or is one specific rotor sticky?

Hundreds Digit

N=2,865χ²(9)=77.21p_adj=1.73e-12V=0.055 (negligible)
Significant deviation from uniform model (FDR 5%)

Last Digit (On Ones)

N=2,865χ²(9)=47.79p_adj=4.19e-7V=0.043 (negligible)
Significant deviation from uniform model (FDR 5%)

Tens Digit

N=2,865χ²(9)=18.14p_adj=3.35e-2V=0.027 (negligible)
Significant deviation from uniform model (FDR 5%)

A Layman's Guide to Statistical Randomness in the Kerala Lottery

Understanding what the numbers on this page actually mean — and what they don't

How the Draw Actually Works

Unlike lotteries that use spinning drums full of numbered balls, the Kerala State Lottery draw is conducted differently. Each draw is supervised by a panel of judges (a minimum of three, including a Chairman), and follows a set procedure:

  • The primary method uses a draw machine with digits and letters mounted on a wheel fixed to a shaft. Before each draw begins, the panel of judges and the audience are shown exactly how the machine works, so everyone can see there's no hidden manipulation. The machine is switched on, runs for a few seconds, and stops on a number — the process is filmed and photographed as it happens.
  • If the machine has a technical fault, the department falls back on a manual method: seven rotating drums (one for each digit or series position), each filled with tokens. An authorized official spins a drum, the Chairman opens it, and draws one token by hand in front of everyone present.

Every draw is documented on a public display board and in an official register, signed by all the judges present, before it's considered final.

There are also built-in checks for edge cases. If a machine or drum happens to select a number that belongs to an unsold ticket, that particular prize draw is repeated until a sold ticket comes up. And if the same number is drawn twice within one lottery, it's cancelled outright and redrawn. Both rules exist to keep the process fair and auditable even when chance produces an awkward result — rather than leaving room for judgment calls in the moment.

This isn't a black box — it's a manually verifiable, publicly witnessed process, which is part of why it's held up as one of India's more transparent state lottery systems. It also means the statistics on this page are testing something concrete: whether a mechanical process, run thousands of times, is behaving the way pure chance predicts.

What "The Null Hypothesis" Means Here

Every statistical test on this page starts from the same basic assumption, called the null hypothesis: that each digit, 0 through 9, has an equal 10% chance of appearing in any given draw position. If the lottery mechanism is working correctly, this should hold true over a large enough number of draws — not perfectly in any single draw, but on average, over time. The tests we run simply check: do the actual results look like what we'd expect from a fair, random process?

Understanding P-Values (Without the Jargon)

A p-value is a number that answers one narrow question: 'If the process really is fair, how surprising is this result?'

A few things worth knowing before looking at any p-value on this page:

  • Small sample sizes produce noisy results. Flip a fair coin 10 times and getting 7 heads isn't unusual — it doesn't mean the coin is biased. The same logic applies to lottery digits: short-term streaks and imbalances are completely normal in a random system, not evidence of anything.
  • A 'significant' result is a flag to look closer, not a verdict. If a test does turn up a statistically unusual pattern, the honest interpretation is that it's worth checking the machine's calibration — not that the draw was rigged. Mechanical equipment used thousands of times over years can drift out of perfect balance in small ways; that's a maintenance question, not a fraud accusation.
  • Testing many things at once increases false alarms. When you run statistical tests across many digits, positions, and lotteries simultaneously, some will look 'significant' by chance alone, purely from running so many tests — this is a well-known statistical effect, not a red flag on its own.

The Honest Takeaway

These tools exist to make the lottery's fairness checkable by anyone, not to promise perfect uniformity in every single draw. Genuine randomness looks messy in the short run and only becomes predictable in aggregate, over a large number of draws. That's exactly what good statistical testing is designed to distinguish — and exactly what this page is here to help you understand, not exploit.

This page is provided for informational and educational purposes. It is not intended, and should not be used, to predict future lottery outcomes.