I think the math seems logical. But, probability is generally...

I think the math seems logical. But, probability is generally more valid on large sample size.
E.g. With success probability of 50%, 
For 4 products, since the sample size is too small, it's hard to say 2 will be approval.
But, if it has 40 products, then it's actually more likely to have closer to 20 to be approved.

I think mesoblast initially approach to develop multiple tier 1 or tier 2 products is right from de-risking perspective, but it's too costly.
So, now we only had about 3-4 products. (better than 1 or 2 products).

The high failure rate and the grueling process of clinical trials with large cost is one reason why barrier to entry in pharmaceutical sector is high.
Typically, existing dominant players have huge advantages, given they usually have the financing in place for multiple product pipelines (on large sample size, the success rate works out, quite close to average rate, it's more predictable and has lower risk).
For smaller firms with only a few product pipelines, their success rate is highly unpredictable, they might have 100% or 0% success rates overall (e.g. when they might only have 1 or two products), thus being very risky.

I think having  3-4 product pipelines might still look risky, but Meso is already in better position than firms with 1 or 2 products, especially when having the first product approval would be crucial, leading to financing for more products and significantly reducing the risk.