Hello I have a question regarding the calculation of the Sortino Ratio in a Monte-Carlo (MC) simulation environment. A quick explanation of my current simulation setup: I am conducting a MC simulation of 250 days over 1000 different paths. I am testing a specific strategy, that is investing between a risky asset and a riskless one depending on the strategy parameter and the current state of the market. I am conducting up to 20 different MC runs with increasing values for a trading filter and a multiplier value which increasing the percentage of investment in the risky asset. I want to create a graphical surface that shows how the values of the Sortino ratio changes along increasing trading filter and increasing multiplier. The graphic would show the sortino ratios on the z axis, the trading filter values on the y axis and the multiplier on the x axis. Now, I wanted to calculate a Sortino ratio for each of the 1000 paths for , then cumulate these Sortino ratios to one average for that specific trading filter and multiplier setup. Is the averaging of the sortino ratios along the simulation paths legit? Best regards
IMO it depends what you are trying to determine. Obviously you are not going to trade 1000 paths so the number is meaningless for that part. But you can use the average of the 1000 paths and the standard deviation to determine how you specific strategy ranks in that distribution, which is hopefully normal, using something like a t-test.