Does normalisation increase or decrease my marks?

Either, depending on how your shift performed relative to the others. If your shift was harder than average, scores in it are generally adjusted upward. If it was easier, they are generally adjusted downward. The purpose is to make scores from different papers comparable, not to reward or punish anyone.

Can I choose an easier shift?

No, and it would not help. Shift and date are allotted by the board, and normalisation exists precisely to remove the advantage an easier shift would otherwise confer. Trying to game this is wasted energy.

Why does the cut-off change every year?

Because it is an outcome, not a target. It is whatever the normalised score of the last selected candidate happens to be, which depends on vacancies, the number of candidates, and the difficulty of that year's papers. Nobody sets it in advance, which is why no honest site can predict it.

RRB normalisation explained: why your raw score is not your final score

If you have taken any railway exam you will have seen the phrase “normalised marks” on your scorecard, usually attached to a number that differs from the one you calculated yourself from the answer key. Understanding why removes a great deal of unnecessary anxiety, and it changes one real decision about how you attempt the paper.

The problem normalisation solves

RRB NTPC is taken by millions of candidates. There is no venue capacity anywhere that allows them all to sit at once, so the exam runs across many days and multiple shifts per day, and every shift gets a different question paper.

Different papers are never exactly equally difficult. However carefully a paper is set, one shift’s mathematics section will turn out slightly harder than another’s. If raw marks were compared directly, a candidate who happened to be allotted a hard shift would be at a disadvantage created entirely by the allotment, not by their ability.

Normalisation is the statistical adjustment that removes that unfairness. Its goal is that a given level of ability produces the same normalised score regardless of which shift you sat in.

What it does, in plain terms

The board does not publish the exact formula it applies, and any site claiming to reproduce it precisely is guessing. What is publicly stated is the principle, and the principle is enough to reason with.

The method compares the performance distribution of each shift against the overall distribution. Roughly:

For each shift, statistics such as the mean and spread of raw scores are computed, along with the performance of the top candidates in that shift.
A shift whose distribution sits lower than the overall pattern is treated as having had a harder paper, and raw scores in it are scaled up.
A shift whose distribution sits higher is treated as having had an easier paper, and raw scores in it are scaled down.

The important implication is that your normalised score depends on other people. Not on their individual answers, but on the aggregate performance of your shift relative to every other shift. This is why you cannot compute your normalised score yourself, no matter how carefully you check the answer key.

What it does not do

It does not change your rank within your own shift. Everyone in a shift is scaled by the same adjustment, so if you outscored someone who sat beside you, you still do.

It does not create marks out of nothing. A candidate who attempted forty questions correctly is not adjusted up to the level of one who attempted seventy correctly. The adjustments are modest, typically a few marks, not transformative.

It does not apply to a qualifying test. The typing skill test is pass or fail against a stated speed. The computer based aptitude test for Station Master and Traffic Assistant requires a minimum T-score in every battery, and that requirement is absolute rather than normalised across shifts.

The one thing it should change about your strategy

Because you cannot know your shift’s difficulty relative to others, and because you cannot know the cut-off in advance, there is no coherent way to aim for a target score. Candidates who spend the exam mentally tracking “am I at 65 yet” are allocating attention to a calculation they cannot complete.

The strategy that follows from normalisation is simply: maximise correct answers and minimise wrong ones, and stop modelling the cut-off. If the paper was hard, it was hard for your shift and the adjustment accounts for it.

Negative marking arithmetic, which is the real decision

The penalty in both computer based tests is one third of a mark for a wrong answer. This is where a small amount of expected-value reasoning genuinely pays.

Consider a four-option question:

Pure guess, no elimination. One chance in four of gaining one mark, three chances in four of losing a third. Expected value: a quarter minus a quarter, which is zero. So a blind guess is break-even, not harmful, on average.
One option eliminated. One chance in three of gaining a mark, two in three of losing a third. Expected value: a third minus two-ninths, which is positive.
Two options eliminated. One chance in two of gaining a mark, one in two of losing a third. Expected value: a half minus a sixth, clearly positive.

So in railway exams, if you can eliminate even one option, attempting is the correct decision on average. This is genuinely different from banking exams, where the quarter mark penalty makes a blind guess a losing proposition and you need two options eliminated before attempting becomes clearly worthwhile.

That difference is worth internalising before you sit down, because it is the kind of decision you will make thirty times in ninety minutes, and thirty small correct decisions is worth several marks.

What to actually do with this

Three practical conclusions:

Do not compute a predicted score from the answer key and then panic. It is not your final score and cannot be.
Do not chase a cut-off number you read on a coaching site. Nobody sets the cut-off; it emerges from the results.
In a railway paper, attempt anything where you can rule out an option. In a banking paper, be stricter. The penalties are different and the correct behaviour is different.