The monte carlo trading system assumes it's random but in trading it's not random if the market is manipulated so that is the flaw in this system. This applies only where the outcome is 'random' and constant R/R. Investing and trading is not random but manipulated. Monte Carlo trading stategy assumes the probability is random on each trade and is a simulator. Lots other professions use monte carlo simulator using random numbers for their analysis. It assumes your reward/risk is the 'same on each trade' for the system to work. or be a positive trader. Its a very powerful statistical tool. free downloads. trading log spreadsheet to show your trading expectancy http://www.marketcentral.ca/expectancy/traderecord.xlsx monte carlo simulator https://www.daytradinglife.com/wp-c.../DTL-Monte-Carlo-Trading-Simulator-v1.01.xlsm this website is probably the best website on trading. teachers of trading should download this and give it to students. It also has the holistic part of trading which is emotional mental management.
MC doesn’t assume randomness at all. Just get a sample of data or a distribution, Then generate different forward paths from it.
with monte carlo, it assumes your probabilty percentag is random but the market is NOT random but manipulated it doesn't apply to trading it applies to gambling in the casino which this simulator or formula was based on . it has to be random and you still make money only if you are right 55% of the time after 500 trades. You are still profitable even if you are only 55% probability of being right which is almost the same as flipping a coin on each trade and minimum trades is 100 trades preferable 500 trades in practice, most traders cannot make 100 trades with small accounts. before being wiped out. you have to do like 100 trades min. for backtesting this monte carlo system.
Well this isn't true. Monte carlo methods are popular because they rely on randomness. The underlying assumption is given enough samples from the underlying probability distribution you can calculate some summary statistics on the population. They can be used to solve nearly any problem who's solution space is probabilistic. For example, you can approximate pi by throwing darts with a uniform "hit distribution" at a square board, counting each dart that hits the circle and each dart that hits the square not covered by the circle and taking the ratio of these (knowing the properties of a circle). Ideally these samples are independent and identically distributed which is why applying it to stock prices generally leaves you with less than desirable statistical power. Your reasoning is flawed. You need to learn more about expected value and games of chance. Expected value is in the limit. 100 to 500 trades is not enough to calculate any form of expectancy with any statistical power. It's unfortunate monte carlo methods are so easy to implement. People apply them to everything without understanding the underlying math. I'm assuming the sheets posted above suffer from the same misunderstanding. Here is a simple example: Code: import random def test(trials): heads = 0 tails = 0 for i in range(trials): flip = random.randint(0, 1) if flip == 0: heads = heads + 1 else: tails = tails + 1 print(f'Out of {trials}: heads: {heads/trials} | tails: {tails/trials}') Run this function using 100 and 500 trials. More often than not you'll find that each realization (aka each run) returns an "unfair" looking distribution (52% heads 48% tails for example) when it can be calculated to be exactly 50/50 in the limit (randint returns an integer from a uniform distribution). If you ran this function 10,000 times you'd approach an expected value of 50%. But to use your example your 100-500 trades is simply a single realization sampled from the population of possible trades and therefore possess no predictive power even if we discount the fact trades are not independent nor identically distributed. Though, if you resample carefully and accept that trades are not identically distributed you can achieve a form of ruin analysis by running monte carlo simulations with proper re-sampling (block bootstrapping, etc). See Kevin Davey's book on algo trading.
monte carlo is best for high frequency traders. Based on the simulation even 100 trades you can be unprofitable. at 500 trades the system generates a profit but it has to random and commissions are low. commissions is like management fees in investing, in the long run fees can reduce returns or even negative returns if the fees are too high as a percentage of profit in each trade. it been around since the 1950s developed by some scientist this is the system gamblers in casinos .as casinos are random. like shuffling cards or slot machines. it has to be more 100 trades preferable 500 trades.
So, what is random, what not? Are the markets now random, or "manipulated" as h-p suggests?! I think he means: since markets get manipulated, there is no real randomness?! Hehe, am I right?
lotteries are random slot machines in casinos believe it or not is random Same with investing in companies. the theory of randomly buying stocks in an exchange where 99% of the stocks listed are worthless or will be worthless is foolish and waste of capital. so stock picking via research not randomly buying stocks is the edge or key to successful investing. your an customer, employee of the industry etc. and know the company. that is the flaw of monte carlo siumulation---it's not random in some cases. mone carlo is used by scientist and other industries. it was invented by some nuclear scientist and wall street applied it to finance.