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)
after extensive testing using Monte Carlo mathematically correct for Drawdown analysis I see only bad news.
Using random selection with replacement to create 200 trades long series of cumulative return (More than 10 000 of them) and from this extracting the distribution for drawdowns it is very hard to come below very big drawdown with any system!!
As we know from Howard Bandy, Drawdown increase over time. This increase follow a diffusion equation. Square root over period.
This is really bad news because any test over longer periods will have DD of over 50% for ANY system. The only way to feel good and get it lower is to test shorter periods. (Much like peeing if its cold, keeps you warm a minute but then...)
The law of large numbers is a strong force!
Is there anyone who have seriously looked into this big problem for any and all system and can recommend solution, literature or other sources of information?
Kind regards
Loop
Can you help answer these questions from other members on NexusFi?
You are tracking random numbers with Monte Carlo. Your found result is true when you play lotto
or other games with random data (as well as fix cost on every turn which deletes your account in
no time.
Systems for trading are different:
1) they should be optimized for letting winners run
but
2) should be optimized from market view to restrict losers to a minimum.
Therefore optimizing is the clue to prevent your account from dripping to zero.
This is a number crunching task.
To make it short: Trading does not show a random outcome like lotto et al.
Now: All my books on Monte Carlo from studies at University are not useful to predict trading results.
I am not tracking random numbers but draw from a given set of trades. This is best practice and accordance with, for example, the method used by Bandy in "Modeling Trading System Performance" or Tharp in his "definite guide to position sizing.." as well as many more.
The trades that are drawn are "optimized" for letting winners run. (Not in the sense of curve over fitting though..)
ATR trailing stop, percentage stop loss and more is used in all systems. Always.
I am afraid this might be a sad reality of trading systems.
And you might find those University books on Monte Carlo more useful than you wished for... There is no way to do a correct analysis of system performance and robustness without them!!
I was rather hoping someone have direct experience with this problem and have thoughts around it!!
Ok lets drill in a bit then:
Step ONE
To design and backtest a system that survives in the market(s) is not trivial. Many traders want to see the automated,
semi-automated or manual (upon upcoming signal) system work in ANY market at ANY time.
This is the crucial point - as one fellow poster here said "Know Thy Market" you need to specify a system on the specialty
of exactly this one.
Now coming to the time BEING in the market and the time STAYING at the sideline:
To avoid any big draw down you need to exactly define for a system WHEN to NOT enter.
Sounds easy? It isn't!
The most errors are made by entering the market under circumstances to do not so.
Step TWO
When you can define to squeeze points out of the market when several conditions in your model are true then you can
get results and note them for the Monte Carlo test. You may see that your tests still bring negative results. If I understand
your words right then you see still a dripping account to limes zero...
Step THREE
Even knowing your entries and exits you need to define your hard stop. To prevent losses or large losses in your account
or even wiping this out. That step seems to be the most important:
You need to filter out NOISE and to stay in the game at the "edge" of normal movements to get your positive points out
of the market.
That means backtesting and even more test under normal market conditions.
You may find the optimized hard stop for a given trade - even with some exceptions for special situations. I am speaking
from patterns that occur.
Once found that optimum one can trigger the system and see the outcome under market conditions in the bay of sharks.
To sum it up:
A trading system needs to be stable with positive results in the long run. This means to find the moments when NOT to
trade to avoid in shaky markets some draw down which could be eventually foreseen.
Once the system presses some points out when conditions are "good" then it survives.
Finally it does NOT matter the quantity of trades but the QUALITY. Think about!
Hope this helps
GFIs1
who recommends to read the "realtime" (system) journals on FIO that show trading results that are consistent.
PS: one last word on the economics of any given market from an economy point view:
If there are losers there must be winners - to balance it all...
I am afraid we are talking about different things. I only refer to drawdowns.
My point is valid even for a live system that brings you a profit.
If you take 500 trades from a live profit system and forecast future likely drawdowns from this they will, for a 200 trade period, have a high risk of being above 50%.
The drawdown risk increases over time. This increase is same as for a diffusion process.
This has to do with the law of large numbers in probability theory.
The only way to control this seems to be position sizing. Using Safe-f.
You note perfectly the zone of position sizing:
I voted in my upper post for position ZERO when uncertain area is in sight or upcoming.
My trigger is the Ichimoku cloud - of which you can easily read many things about - even in my journals.
SO - given the obvious - we CAN get out of a high risk momentum and let pass the ugly head of draw down
with just omitting the risks by taking a trade right there and then.
But that needs a certain preparation for the system at work.
Getting back to my second to last post: SOMEONE is always making the gain in the market...
THUS - we need to get into row ONE.
Good trades
GFIs1
If risk has been contained and position sizing controlled then surely the systems you refer to just have relatively low expectancies? (i.e. lacking any real edge)
edit: And/or, are highly susceptible to shifts in market conditions, in which case they are almost always going to come unstuck on one sequence sooner rather than later.
There is something I realized when running Monte Carlos...
For example, I'll put in my 100 trades that took place over a month's time, and let it run over 10,000 simulations. That's, effectively, 1,000,000 trades. *clucks*
*Thinking*
Can a Monte Carlo simulate my bias on mechanical trading? Can the Monte Carlo simulate fear that makes us break our rules? Can a Monte Carlo follow our rules? It cannot do any of these things, and cannot do many other things we would need it to do to give us an accurate emulation of what may happen if we kept doing what we did, based upon our previous trades.
I like the Monte Carlo, but only so far. It is simply not a realistic view of what WE would do in the long run, over 1,000,000 trades.
