Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
Hi,
I have troubles understanding Monte Carlo simulation value in the context of a position sized list of backtested trades but maybe my understanding is wrong.
On a single contract sequence list of trades with no position sizing, I can see how Monte Carlo shuffling of the trades helps determine system boundaries (max loss / wins etc etc...).
But on a position sized sample of trades how can this be accurate at all?
Let's say first trade of a position sized sequence is 1 contract with $10 gain, and last trade 100 contract with $1000 loss.
If Monte Carlo, as practiced by most plateforms shuffles trades and happens to put last trade of $1000 first it will creates an unrealistic, huge drawdown on that simulated sequence (because $1000 loss is based on 100 contracts which never happen at start of a backtest sequence).
Am I missing something? How to deal with this?
Can you help answer these questions from other members on NexusFi?
I am not a math guy at all but I have learnt enough to put it to practical use so I will answer from that point of view - not from a strict math point of view.
Also want to point out that I do not use a commercial platform, I built my own. I have used commercial platforms but that was a few years ago so I cannot relate much to any platform specific things you refer to.
My understanding is that the purpose of applying monte carlo simulation to your backtest results is to make your backtest more realistic by random shuffling, exclusions and doubling. It tries to mimic your sick days, when you could not trade or days when you make a mistake and buy 10 instead of 1 etc etc.
From that point of view, even with a backtest that does not have a consistent trade size, it can be considered to do what it is supposed to do - make the backtest more realistic.
Also, I have kept position sizing a separate calculation. Not part of the backtest. Backtest only tests the legitimacy of a strategy. Various factors such as volatility, liquidity (time of day) etc determine position size. And especially if you are trading a basket of securities.
Position size is a function of volatility, and account size = $risk per trade. If you are use Excel to create a monte carlo use the raw input of stop loss and target, and then formulate it so position size is based on the current simulated account size for that trial.
So that last trade of 100 contracts may end up as the first in one of the simulations with a position size of 1, and a loss of $10. It may have had a SL of 5 points in the backtest, and so continues to have a SL of 5 points, but will have all sorts of position size associated with it, based on the sequence it's selected. If you don't have the SL associated with each trade, then you should have a risk per trade, so you'll have to go through and figure out how many points that equated to in the back test results.
Once you have the points risk for each trade you can size up each sequential trade in the simulation based on the current account size.
You make a great point.
I think the beauty of MonteCarlo is that running it without position size, will tell you how much profit you need before you can safely increase your position size. For example, if your strategy needs 25K to trade with the risk of ruin you desire, then youre gonna need 25K of profit before you add that second contract, and wont need to run another MC. but if you do run it, then it would be with 2 contracts and 50K initial balance. Given the way MC works I dont think its accurate if you include your position size.
