Question 1
27 marks
Boosted actions in a game: geometric and rising-probability models
- Statistics and probability
(a)(i)
2 marks
The first two actions are not boosted and the third is.
(a)(ii)
3 marks
Use the complement: no boosts in six actions has probability .
(b)(i)
1 mark
The first boost is on action exactly when the first actions are not boosted, each with probability , and then action is boosted, with probability . The actions are independent, so the probability is .
(b)(ii)
1 mark
The expected value is the sum of each value times its probability.
(c)(i)
4 marks
- Differentiate the left side of term by term, with the constant term vanishing.
- Differentiate the right side, .
(c)(ii)
2 marks
In put , and .
(d)
2 marks
With .
(e)
2 marks
The chance of a boost is 0.2 on the first action, 0.4 on the second and 0.6 on the third. The first boost on the third action needs two actions not boosted, then a boost.
(f)
1 mark
After four actions that were not boosted, the chance of a boost on the fifth action is , so the fifth action is certain to be boosted. The first boost cannot come later than the fifth action.
(g)(i)
2 marks
- For : not boosted on the first, then boosted with chance 0.4.
- For : not boosted three times, then boosted with chance 0.8.
(g)(ii)
2 marks
Sum .
(g)(iii)
2 marks
Use .
(h)(i)
1 mark
Set .
(h)(ii)
1 mark
Use the variance formula.
(h)(iii)
1 mark
The two models have the same mean, but . The second model has the smaller spread, so it gives a more consistent experience.

