What principle states that larger pools of similar exposure units lead to more accurate loss predictions?

The principle that larger pools of similar exposure units lead to more accurate loss predictions is known as the Law of Large Numbers. This statistical concept asserts that as the number of exposure units increases, the observed outcomes (such as losses) will converge to the expected values based on probability.

In practical terms, when an insurer groups together a large number of similar risks, it becomes easier to predict future claims more accurately because the random fluctuations in small datasets tend to even out. For example, if an insurance company insures a small number of homes, a couple of significant claims can skew the overall assessment of risk. However, with a much larger group of homes insured, the impact of a few large claims is diminished, leading to a more stable and predictable outcome based on the average loss per unit.

Other concepts mentioned, like the Law of Averages and the Central Limit Theorem, relate to statistical analysis but do not specifically address the relationship between increasing the number of observations and improving accuracy in predictions about risk. Frequency Theory, on the other hand, concerns itself more with the probability of events over repeated trials rather than the specific predictive power gained through pooling larger datasets.