Hahahahaha! Now I get it! Been there, done that, have you? An excellent precis! Gentle readers, I think I detect the bitterness of repeated failure. Suivez moi.
I use a proprietary BT platform which I built myself. It follows a distributed architecture which consists of Matlab, Excel, Bloomberg, SQL Server and C++. I had a CS background before I became a trader. I have thousands of different trading ideas. Before I will ever consider to trade any of them I need to prove the concept is valid, makes money (reasonable Sharpe) and is robust. I want an idea to make money for N pairs, securities, baskets etc... not just one. That is how I can scale it. My trading platform is 90% automated so all I have to do is monitor it intraday. Plus I look at everything from a portfolio POV. A new idea might make money but if it's highly correlated to my existing portfolio then I will not add it. If it's better than an existing idea I would probably allocate 50/50 between the two. Arthur and Mikasa are you married?
Marine you my friend are a start up bullshit I choose not to help you Marine because you my freund didn't suffer reality yet. As for Arthur Deco, he needs to write the following before I will help him. "all those work and fizzle out strategies I created prove market is random. I ask Mikasa for help" Simple, no ?
Thanks, Marine, you are way out of my league, as my old new friend MeCasa well knows. So your BT consists of multiple tools, not just a simple historical price platform? I assume some statistical analyses as well? Only thousands of ideas, haha? I try to invent and test one new idea every day. I have a BT shell in EasySignal that allows me to plug in a new idea in minutes, as long as the math isn't too higher. Apologies in advance that what I am about here is waaaaay below your skill set.
So we now unnaturally progress to the Second Law: It follows from the First Law that if you cannot write an algorithm for your trading idea, and therefore test it, you do not have a valid idea. In simple language, "You cannot backtest an illusion."
WOW what a classical attempt at intelligence maybe guys like me and guys at Goldman (your competition in futures) don't waste time with bullshit Swallow you pride (for God's sake its the internet) and admit that it took you years to realize markets are random and ask for my help I won't be here forever you know.
Soooory. I don't buy into the random walk theory. That trends are Brownian. Too much evidence at the tick level to the contrary. And I seriously doubt that anyone at Goldman is peculating using market orders. More likely they are too busy maintaining Jack's Pool Extraction Paradigm with limit orders. That is, of course, what makes it all possible. Later. Gotta get set up for the day.
Well, this AM was almost fun. Since this thread is generating more heat than light, I'll wrap it up quickly. Here is the Third Law: If your idea doesn't work with no optimization, it's a bad idea. Run your first test with WAGged parameters based on the conditions at the time you first had the idea. If it has negative expectancy, forget it. Don't waste your time with elaborate parametric variation, stops, profit targets, reversals, etc. This is the "pain" my friend Mikasa referred to. What do you do if the idea DOES work? That's covered by the Fourth Law. Bet you can't wait.
Have you finished your first year at your Big Bank, or are you winding up your 6 month stint just out of college? Did they give you a desk yet? Are they letting you press any buttons?