Annualizing Volatility in Monte Carlo Price Simulations

# My Montecarlo Simulation is not working?


# My Montecarlo Simulation is not working?












My aim is to predict 1 year ahead and daily, the price of a stock under certain scenario. These scenarios are the ones that this year the stock will have a similar year, in terms of standard deviation and return, to the 2008 and to the 2017.

So what I did is to compute the mean of the DAILY returns and the mean of the daily standard deviation.

However, even though in the 2008 the return were of -40% in a year (mean of daily returns: -0.003074428479940944, mean of daily std: 0.028883647401261883) and for 2017 the return were of +30% (mean of daily returns: 0.0010560777183462407, mean of daily std: 0.011807274319995538), by plugging into the model the parameters, the MC simulation is giving me really similar results, with the VaR that is suggesting me very small possible losses (400$ from the starting price of 8200) which is roughly the 5%, but in a year in which the stock made -40%!

Can someone explain me where I made a mistake? Is it wrong how did I define "sigma" or the dt, which maybe should be 1 and not 1/252 ?

```
def mc_asset(S0, r, sigma, T, Nsteps, Nrep):
    SPATH = np.zeros((Nrep, 1 + Nsteps))
    SPATH[:, 0] = S0
    dt = T / Nsteps
    nudt = (r - 0.5 * sigma **2) *dt
    
    sidt = sigma * np.sqrt(dt)
    
    for i in range(0,Nrep):
        for j in range(0,Nsteps):
            SPATH[i,j+1] = SPATH[i,j] * np.exp(nudt + sidt * np.random.normal())
    return SPATH
S0 = datiRame['Copper Cash'].iloc[-1]
sigma = #MEAN OF THE DAILY STANDARD DEVIATION
Nsteps = 252
T = 1
r = #MEAN OF THE DAILY RETURNS 
Nrep = 2000
SPATH = mc_asset(S0, r, sigma, T, Nsteps, Nrep)

plt.figure(figsize = (10,8))
for i in range(len(SPATH)):
    plt.plot(SPATH[i])
plt.xlabel('Numbers of steps')
plt.ylabel('Stock price')
plt.title('Monte Carlo Simulation for Stock Price')
plt.show()
print(S0,sigma_selected_year, mu_selected_year)

# Define VaR and CVaR functions
def mcVaR(returns, alpha = 5):
    return np.percentile(returns, alpha)

def mcCVaR(returns, alpha = 5): #Return CVaR or ES
    belowVaR = returns <= mcVaR(returns, alpha=alpha)
    return returns[belowVaR].mean()

# Compute VaR and CVaR
VaR = S0 - mcVaR(final_prices[:-1],alpha =5)
CVaR = S0 - mcCVaR(final_prices[:-1],alpha =5)
VaR, CVaR
```

## Answer by KaiSqDist (score 2)

https://quant.stackexchange.com/a/78238

Your sigma seems to be incorrect, the sigma used should be annualized. Therefore, if you use your mean of standard deviation, you need to multiply it with $\sqrt{252}$.

I do not think there is anything wrong with your T, if you are simulating for a year ahead. Each day $dt$ should be $1/252$.

Can you update me with the results using these amended parameters?

## Answer by AlRacoon (score 2)

https://quant.stackexchange.com/a/78241

Even with 2000 simulations, you may not get the expected return and volatility as your estimated parameters.

Try running your monte carlo with more simulation paths and this should close the difference between your estimated parameters and your simulated parameters.

Also, you may try some variance reduction techniques such as antithetic sampling to get your expected results with fewer simulations.