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)
I can't really justify any of this scientifically (I'm not much of a mathametician), but I prefer my SQN performance function for choosing parameters. I then use expectancy for ranking my many system attempts to decide which ones to deploy my limited capital on. Actually I rank by sharpe ratio x expectancy. I figure by doing so I'm preferring both high expectancy and consistency.
Can you help answer these questions from other members on NexusFi?
This is an interesting thread. If I understand these posts, what was coded can been seen when analyzing the chart after the optimization, or walk forward, or back test for that matter.
If the trade looks to good to be true on the chart, then it likely would not occur in a live environment. After an optimization or back test or walk forward, I am always studying the chart to see if the numbers are real or just the way the bars plot outside of the live environment.
RE Win/Loss, what about a trailing stop? You could have a winner but not a winner according to your parameters, and with a health risk reward anything over 50% is pretty darn good.
What is particular interesting to me is the max consecutive losers and biggest loosing trade.... How much pain can you take before you're tempted to switch things up in a live environment.
I am looking forward to examining you're optimization code, thanks for sharing.
I tried once to code, it. but not had enough experience tweak ninja. It should be possible to code. I found all values.
I try to find my code, if somebody else want try to code it.
Funny you should ask. I went through the nexusfi.com (formerly BMT) threads and did not find anyone who had comleted the work identified in your link. I think there is another nice link here in nexusfi.com (formerly BMT) that goes along with this discussion that is a bit older, but still relevant to this discussion:
Here is my rendition of the Expectancy Score from Trade Your Way to Financial Freedom (http://www.amazon.com/gp/product/007147871X?ie=UTF8&tag=unicoreseacor-20&linkCode=as2&camp=1789&creative=9325&creativeASIN=007147871X)http://www.assoc-amazon.com/e …
I borrowed a lot of the work from @caprica and then took the formulas from Unicorn Research to build the optimizer. Here are the formulas from Unicorn:
Expectancy is how much you expect to earn from each trade for every dollar you risk. Opportunity is how often your strategy trades. You want to maximize the product of both. Expectancy = (AW × PW + AL × PL) ⁄ |AL| (expected profit per dollar risked) Expectancy score = Expectancy × Opportunity where AW = average winning trade (excluding maximum win) PW = probability of winning (PW = <wins> ⁄ NST
where <wins> is total wins excluding maximum win) AL = average losing trade (negative, excluding scratch losses) |AL| = absolute value of AL PL = probability of losing (PL = <non-scratch losses> ⁄ NST) Opportunity = NST × 365 ⁄ studydays (opportunities to trade in a year) where
NST = <total trades> − <scratch trades> − 1 In other words, NST = non-scratch trades during the period under test (a scratch trade loses commission+slippage or less) minus 1 (to exclude the maximum win).
studydays = calendar days of history being tested
Hope you guys enjoy and I am looking for feedback and improvement ideas.
So I built this Expectancy Score optimization (prior post) and I get some values (see attached). Now I ask, which is best? You can sort each column and come up with a different answer for that question depending on which column you pick. But how do you combine all the columns together (becasuse they are all important) and come up with a 'BEST'?
Multiply them together? Calculate a Bayesian inference? Least squares? Pythogrium therom? Pick a number between 1 and 10?
Remember, inside an optimization, you can only look at one row at a time and give that row a value. Then the optimizer picks the top values for your consideration. I am not a mathmetician, but it just seems to me it can be done mathmatically. Anybody got any ideas?
Tell you what. You come up with the idea and I will put it in an optimization and share. Deal?
