I have a basic question regarding Monte Carlo Simulations. From a pure statistics point of view Monte Carlo methods are known to be effective only when you want to find the value of Y in the equation Y = f(X) and the X or Xs is known.
From a market perspective, Monte Carlo Simulations are done to figure out the value of Y in the future based on the value of Y in the past! The most important point that is left out is that the value of X is never known! What is the point of adding randomness (albeit statistically computed) to existing Ys to derive future Ys?
Is the application of Monte Carlo Simulation in the stock markets really effective or is it purely esoteric?
My familiarity with Monte Carlo (slight variations of related random walk models) is limited to non-financial computational simulations, but here's my rather modest understanding of it in the financial context.
When a model is trying to forecast an outcome in the future, there are going to be a lot of preliminary assumptions and estimates involved. Minor changes in these estimates could very well produce wildly varying outcomes. Stochastic models like Monte Carlo help "tune" these estimates by simulating large numbers of possible outcomes (in terms of markets, it could be scenarios of -m% to +n% possible returns).
However, the effectiveness of the simulation is purely dependent on your preliminary estimates (like you said, input X). Thus, Monte Carlo is not a black box for predicting the future in any sense. It's only a tool that'll help you optimize the input X you estimated yourself. The wilder the X, the wilder the possible outcomes.
So, really, Monte Carlo's effectiveness is only as good as your model. Whether it's esoteric or not is subjective :)âmany swear by it, many don't.
I guess you are unable to define the function Y = F(x) in the context of markets (or so I have assumed). Let me attempt explaining this in a market context by defining Y as a function of X.
Consider a simple regression equation where you are trying to predict the value of letâs say Reliance (dependent variable Y), based on the movement of Nifty, which in this case would be your independent variable X.
Assuming a linear relationship exists between the two, you can equate the two by a simple linear equation â
Y = α + ÎČX + Δ â> This is pretty much takes the form of Y = F(x)
Where α = alpha, B = Beta coefficient, and Δ is the standard error element.
Relatively both α & ÎČ are constants, as they donât change values on a daily basis. X is a variable anyway. This leaves us with Δ which is the standard error. This also means that if I have to predict the value of Y (on a short term basis) it is largely going to depend on my estimate of Δ.
To get a fair estimate of Δ, I can run a Monte Carlo simulation to understand various possible outcomes. I would agree with what Kailash says â the effectiveness is as good as your model, which in this particular example requires you to define a lower and upper bound for Δ. Bases on how you calibrate these âboundsâ, you will get a more realistic estimate of Δ.
Besides do remember, once you simulate various possible outcomes, taking the average of these estimate always helps!
Monte Carlo simulation is one of the most important steps in Trading system development and optimization. It is often overlooked by beginners considering the mathematical complexity it contains. Also, there are hardly any articles available at Internet which explains it in layman terms.Â
What is Monte Carlo Simulation?
Monte Carlo simulation is a process which performs repeated execution of pre-defined set of steps by adding randomness to the input parameters at each iteration. The results are noted down at the end of each iteration which forms the basis of probabilistic analysis of the desired result. In Trading terms, Monte Carlo simulation is performed to forecast the success of a backtested trading system. In order to ensure that your trading system is robust,  backtesting should be performed multiple times by adding variations to your trading rules or data. If it gives consistent result each time, then it has a higher probability of generating profit.
Monte Carlo simulation in real world
Monte Carlo simulation is very popular in the field of statistical and scientific experiments. For ex: Consider a scientist who wants to estimate the trajectory of his space shuttle. Since the trajectory is highly dependent on atmospherical condition which is random, he has to perform Monte Carlo simulation in order to arrive at the most probable trajectory.He will repeatedly simulate the trajectory by adding randomness to the atmospheric parameters after each repetition.
Advantages of Monte Carlo simulation in Trading
It is a well-known fact that âMarkets are Randomâ, so Monte Carlo simulation is a method to absorb this randomness in your Trading system. If your system performs well in random market conditions, then it has a huge probability of success. At the end, everything boils down to probability and that is actually the basis of all profitable Trading systems. Itâs really not enough to believe in the Trading system just based on profitable backtest reports. Monte Carlo analysis weighs equally while designing a system.
Read the below articles to understand step by step process to perform Monte Carlo simulation: