How to research and verify trading ideas

Status
Not open for further replies.
I have found your posts very useful and you've done a great job trying to shepherd this thread along a constructive path. However, the current topic is just wrapping up this failed thread... nowhere left to go with it.

I thought it would be useful to show people how to develop systems from backtest to execution (rather than simply stopping at models). I originally planned on using some bullshit like buying MA crosses which would have been easy to test. The fact that the system had no edge didn't matter because the point was the test process.

Instead, I chose to use the first system idea I use with trainees. Whether or not the system works was the point of the thread... I was under no obligation to prove it (though your backtest has done so). The point was the process but it never took off because of BoWo's insistence on me proving myself to be the guru with a great system.

Not the point, respectfully.

Quote from whoisjohngalt:


1. bwolinsky was only asking for a backtesting result. For some reason, we haven't seen one until I posted it. I can understand how he can be skeptical about the whole idea. No one was asking for the source code or some secret. Only thing was a backtested result for the Add/Remove.

Talon, why didn't you post a little to shut him up? I wouldn't be complete off if I said that it does make people skeptical... seriously... why didn't you shut bwolinsky by posting your results?

That said... what's the current topic?
More...
 
Quote from talontrading:



Not the point, respectfully.
More...

It would suck to be your trainee. Here's an idea, go test it, guys! This thread went nowhere because there is no proof, nor do I take the current backtest as proof, either. You said 10% was too large a position size, so if we're talking 1%, it is absolutely impossible to get the results reported in the backtest. Even at 5% it's not, and we're left at 10%, so how big should the size be?
 
Quote from talontrading:



3. Trading on the open. Does your data source reference 930 print or primary market open as the opening print? Answer that and we'll move to the next step. If you don't know... well... hmmmm....
More...

Yes, it's designed to do that.
 
Quote from talontrading:



1. It's cool that the best pairs trader / systems trader in the world can take 20 mins out of his morning on the open to post to ET. Most traders would be trading or focusing on the market during that time so I guess that shows his dedication to ET. Normally I wouldn't expect a trader to have that kind of time on the open, so thank you.
More...

Pairs trading is a quantitative automated process requiring no intervention. Trades do not come daily.
 
Quote from talontrading:

It tells you something about our work culture here that someone who, 6 months ago didn't know what a bid / ask spread was (who is now daytrading profitably 19 days out of 20 btw) can walk in the door this morning and having read the ET thread on the train can tell me I was "acting like a 10 year old."
More...

Well, at least your trainees aren't total dumbasses.
 
Quote from talontrading:


Do you find that answer "acceptable"?

5. You're so focused on models and backtests because you aren't a trader and it's what you have. A model is only one small piece of the development process for us, and a model with position sizing is several steps down the road. If you had not derailed this thread with your bullshit, we would have covered all of that. There is a tremendous gulf between model and execution... almost nowhere more so than pairs.

You sum it up when you said "Mainly, I'm interested in the backtests." Mainly.. I'm interested in trading.
More...

You're so full of shit if you think I don't trade. I don't trade till I have edge. My models are my edge, and it has benefited me greatly for years to be able to program algorithms. My current system is 50 pages long in word, and you think I don't trade?

And, yes, the answer was acceptable.
 
Quote from talontrading:



C. We run some multidimensional arbs. Think a program that looks at SPY, ES, SSO, options on all of the above, with the possibility of also actually physically executing the whole basket of S&P 500 stocks... so you're actually looking at 7-8 instruments and have to work off bid / ask spreads because of asynchronous quotes. (Do you know what that means Beau? It's not in the CFA manual.) This one is transaction intensive and works at the tick level. (We are colocated NYSE and Chicago for this one.)
More...

Yes, I know what that means, you moron. For my internship I had to study equities market makers, and it is why I know market internals to this day.
 
Quote from talontrading:


2. Why is it SO FUCKING HARD for you to understand that the point of this thread was to create the backtest? I don't think you're actually stupid but why do you refuse to grasp that? I could have picked any system... buy 50 day highs, buy moving average crossovers, buy stochastic crossovers, but I chose something simple that I had good P&L history on to show it had an edge. For me to produce a backtest would have eliminated the whole point of the thread. I don't expect anyone to follow me. WHY CAN'T YOU GRASP THIS?
More...

You're naivete is not understandable. You come on here, with a system, tell everyone they need to follow it for a year, and <i>then wonder why we need proof before following it?</i>

<b>You're a dumbass if you think there is no need for a backtest in that situation.</b>

Do you seriously believe you don't need a backtest?

<i>We already know the answer. We don't according to you, but for every other trader out there, would need a backtest and complete tradeslist before considering putting any money in a brand new idea out of nowhere.
 
any chance we can drop the fight here and just let talon continue to post if he so chooses? i guess just agree to disagree and move on with the thread
 
Status
Not open for further replies.
Back
Top