Please sign me up, i will PM you my credit card info.
Backtesting: My trading strategy works too well
- Thread starter Victory5
- Start date
Please sign me up, i will PM you my credit card info.
In "Algorithimc Trading" you can find simple rules which prevent curve fitting.
1. Share your data series on two halves. Search for parameters/formations based on first half and test your parameters on the second half.
2. There is a formula which tells you how many data you need to have to make your test significant. I don't remember it but t is based on the number of parameters.
1. Share your data series on two halves. Search for parameters/formations based on first half and test your parameters on the second half.
2. There is a formula which tells you how many data you need to have to make your test significant. I don't remember it but t is based on the number of parameters.
The problem can also be the weakness of the system you built to trade. You should test your system to see if it is valid.
This is how the test works: in each system there are parameters. You should be able to change the values of these parameters (within certain limits) without influencing the performance of the system. If the performance stays more or less equal, you have a good and stable system. If your performance changes too much it means you performance was depending on accidental circumstances that might not occur in real or future trading.
It points then in the direction of an optimalization for a specific situation or trend in the market. You should avoid that.
Simpe example is: trade on an optimal MA. Because today it can be 12 and tomorrow it can be 17.
Last edited:
I know that not everything that is posted here is true and real. But I can confirm that making 300$ a day on a 3K account is possible. I don't know the OP, so I don't know if he can, but it can be done CONSISTENTLY.
The only bad news fo you is that people who can do this don't need your creditcard. In fact they don't need money from anybody. They take it in the market.
LOL, 10% a day and can be done consistently.
See my reply to NoBody previously:
"
You must be smarter than anyone that I ever see in any institutions (Goldman, Citi) or any famous prob firms. No one (or even any machine / algo) can have this kind of accuracy to predict market direction even with all the best data and market info, it seems you get the direction "right" almost every time (the small losing trades are seemed more due to market chop)?
I recalled previously we have one young guy claimed his has similar trading result (technically his P/L is almost a linear line to north as written in his CV). The management decided to give this guy an exception for the interview for a trader position (This guy is only a science graduate from one of the community college. The company policy guideline is we will never interview any traders without any "proven" trade record or a solid education background- e.g. he has to been recommended by others in industry or good education background, e.g. PhD or equivalent). The only request from the firm is he has to prove this type of trade records in real time, the result is - he lost more than 70% in the paper account in less than a week, obviously he never get hired.
As usual, Nobody never make any real time call (with sufficient number of trades to prove it is statistical significant) to prove what he claim in his trade records.
Highly suspicious"
He may either want to sell something here, or try to trick some new trader (mostly young, creative and non experience) to share their idea by "pretending" he is a expert in trading, or worst, he is alike of some insane guy that claimed who can make x3 time the ATR a day using some fictitious imagination.
Again, Highly suspicious.
Indeed, since the data is available from Feb 09, there is a clear short vol bias in play. There are so many people shorting volatility out there and they make their assumptions on the same data the OP is using. If I was the OP, I'd just watch the strategy for 6 months (maybe paper trade) without playing around with the parameters, if the equity goes flat then he knows it was a curve fit.
Show me a posting where I am selling anything. You will not find it. So your suggestion is slander. On the contrary, there are colleagues bashers from you who do sell subscriptions here. They have to because they have to make money in some way. I don't need to sell anything because I make money in trading.
Again slander, you never proofed any confirmation of your statement. On top of that it proofs that you are really a newbie in this business. Somebody who is young, creative and non experienced will need years before he can have anything meaningful to tell. And even then 90% will never make real money consistently. Only desperate people (like you?) would maybe do this. If that's the only option left for you....
Maybe that's why you put so much energy in following my postings. Hoping I will prove my right by giving away me secrets.
Again slander, no proof at all, only wishful thinking with an unlimited amount of fantasy. The ultimate proof how weak your statement is and how desperate you are. The only person with fictitious imagination is YOU. You confirmed this with your suggestive posting.
If your firm wants to give an opportunity like you described, it confirms that this firm is of the same level as you are. If a firm gives such an opportunity to such a guy (technically his P/L is almost a linear line to north as written in his CV), it only confirms that they are in a desperate position too. They are sceptical and don't believe the guy but apparently are in such a desperate situation that they hope this boy might bring the miracle they need. But it is logical, losers and losing firms go together. They attract each other. Or trying to steal his knowledge by pretending you are an expert in trading? But that would really make a joke from you and your firm.
Last edited:
This is exactly I expect this "NoBody" will response, he will try to twist the divert away from the main questions:
"As usual, Nobody never make any real time call (with sufficient number of trades to prove it is statistical significant) to prove what he claim in his trade records."
Highly suspicious
Forward test it in simulator mode. If successful, trade it real money. Harvest and enjoy. 