BEWARE Increased premium after NO FAULT accident

Spiderham

mikey72 wrote: »

or statistically am I somewhere on the bottom of the curve?

Correlatively speaking this probably. Bear in mind a significant number of claims do come from completely "clean" customers so there is always an implied latent risk.

Insurers don't know what causes claims (ie causal links), but they can get quite good at spotting statistical trends and by linking similar drivers together get an implied cost for various factors. Put these together, plus a bit for solvency, wages and (haha, we should be so lucky) profit margin and voila, you have a premium.

That is more or less how it works in layman's terms.

starrystarry

Spiderham wrote: »

But is it? I'm less than convinced.

Ok. If you were to roll a dice, the odds of you rolling a 3 are 1 in 6. Agreed? Let's say you roll a 2, what are the odds of you rolling a 3 next time? Are they less than, the same as, or greater than 1 in 6?

starrystarry

mikey72 wrote: »

Fair comment then. So a question. What makes me more likely to have a crash tomorrow, when you assess risk for the years premium?
Assume I've had say, 10,000 days with no incidents so far, so I have no probability of an accident, or statistically am I somewhere on the bottom of the curve?

Dunno Mikey, I'm not a motor insurance underwriter, as you know.

Spiderham

starrystarry wrote: »

Ok. If you were to roll a dice, the odds of you rolling a 3 are 1 in 6. Agreed? Let's say you roll a 2, what are the odds of you rolling a 3 next time? Are they less than, the same as, or greater than 1 in 6?

Depends if it's an equally weighted fair dice. Car crashes are not an uniformly distributed variable.

Let suppose we have an unfair die, we don't know how the weighting works between the faces. I roll the die 100 times, we get
1 - 20 times
2 - 8 times
3 - 17 times
4 - 15 times
5 - 28 times
6 - 12 times

What is the probability of rolling a 6 next time?

We don't know

The more data we have the closer our modelling of the situation is to reality via the law of large numbers.

starrystarry

Spiderham wrote: »

Depends if it's an equally weighted fair dice. Car crashes are not an uniformly distributed variable.

Let suppose we have an unfair die, we don't know how the weighting works between the faces. I roll the die 100 times, we get
1 - 20 times
2 - 8 times
3 - 17 times
4 - 15 times
5 - 28 times
6 - 12 times

What is the probability of rolling a 6 next time?

We don't know

The more data we have the closer our modelling of the situation is to reality via the law of large numbers.

You're overcomplicating Mikey's scenario now. Remember, he said...

"I suppose it could make sense, if they haven't been hit yet, statistically they must be more at risk as the odds increase on them being hit eventually."

So Mikey said that if they haven't been hit yet they're more likely to get hit in the future. No weighting, no adverse road conditions, no faulty brakes, no driver nodding off at the wheel. No weighting. What are the odds of rolling a 3 second time around?

Spiderham

starrystarry wrote: »

What are the odds of rolling a 3 second time around?

I don't know, I don't know if the die is fair. If it is a fair die it is 1 in 6, however we don't know the distribution that the die in question leads to from the information given.

I know I seem to be evading the question but that is because you are hugely oversimplifying the question. I'm guessing Mikey was being facetious with that comment.

Whether someone has been hit or not does not affect their likelihood of making a claim, in exactly the same way that having a fault claim and causing an accident don't make you any more or less likely in having claims in future. Both incidents are useful at indicating what the statistical distribution is.

starrystarry

Spiderham wrote: »

I don't know, I don't know if the die is fair. If it is a fair die it is 1 in 6, however we don't know the distribution that the die in question leads to from the information given.

I know I seem to be evading the question but that is because you are hugely oversimplifying the question. I'm guessing Mikey was being facetious with that comment.

Whether someone has been hit or not does not affect their likelihood of making a claim, in exactly the same way that having a fault claim and causing an accident don't make you any more or less likely in having claims in future. Both incidents are useful at indicating what the statistical distribution is.

Ha ha! Mikey being facetious? Surely not.

I'm not getting dragged into a debate about this. Mikey's original statement that if someone hasn't been hit they're more likely to get hit in the future is incorrect. Fact. You can drag any number of variables into the argument to cloud the issue, but remember Mikey didn't.

1 in 6

Spiderham

starrystarry wrote: »

1 in 6

But you never stated the distribution of outcomes of the die so I was unable to accurately answer. (Can you tell I'm a mathematician?)

starrystarry

Spiderham wrote: »

But you never stated the distribution of outcomes of the die so I was unable to accurately answer. (Can you tell I'm a mathematician?)

You did answer! 1 in 6.

I'm going to cry in a minute. Somebody help me. Are you Mikey in disguise?

Spiderham

starrystarry wrote: »

You did answer! 1 in 6.

I'm going to cry in a minute. Somebody help me. Are you Mikey in disguise?

No, my point was that you made an assumption about the die without stating it. If you read through what I've said it does make sense.

Ok. If you were to roll a dice, the odds of you rolling a 3 are 1 in 6. Agreed? Let's say you roll a 2, what are the odds of you rolling a 3 next time? Are they less than, the same as, or greater than 1 in 6?

The problem was I didn't agree with the premise of your question, that the odds of rolling a 3 apropos nothing were 1 in 6. From there it's all semantics.

Mikey's situation was arguing about the link between the cost of claims per policy and various factors, my "unknown distribution" die example was meant to show (very basically) how the rates and implied probabilities come about.

Let me ask you a question. If you were handed a die, rolled it 100 times and didn't roll a single 6, what do you think your probability of rolling a 6 next time would be.

My answer:

Very close to zero, no mention of it being fair and experience tells me the 6 is very unlikely