How to research and verify trading ideas
- Thread starter talontrading
- Start date
- Status
This is a bit overboard when refering to a subject as vast as monte carlo methods....
In the stragey being described here, the value of MC would be in seeing how your system compares to monkeys throwing darts at a board on the same data, throwing a number of darts similar to the number of trades the system makes and holding for a similar time.
Then you can figure out how much you beat the monkey by and how confident you are that beating the monkey wasn't random itself.
As dtrader mentioned earlier though, there is basically no point in having intelligent discussion on this board anymore....I have a massive ignore list built up over the last 2 years but still the morons are always going to chime in and ruin it. You would really be better off taking this discussion to basically any place but here.
No doubt that is how a lot of people use monte carlo, but I don't find it necessary or particularly useful. For most of our work we have the full sample set of returns and then a subset that conforms to our signal. We compare the two datasets (note that the 'control' group also includes the signal) and see if they are, statistically, different.
i'm a big believer in nonparametrics, but monte carlo is far overrated for financial applications in my experience. i have LONG been suspicious of risk management models built by monte carlo, and many of those models blew up in recent years. monte carlo is kind of a last resort when nothing else will work method... and i know the students and engineers love it, but the stats we do could mostly be done by hand... i mean we use R and Matlab, but a whole lot of our work gets done in Excel which is superUNsophisticated.
the concept of comparing returns from full sample and your signal is sufficient in almost all cases.
i'm a big believer in nonparametrics, but monte carlo is far overrated for financial applications in my experience. i have LONG been suspicious of risk management models built by monte carlo, and many of those models blew up in recent years. monte carlo is kind of a last resort when nothing else will work method... and i know the students and engineers love it, but the stats we do could mostly be done by hand... i mean we use R and Matlab, but a whole lot of our work gets done in Excel which is superUNsophisticated.
the concept of comparing returns from full sample and your signal is sufficient in almost all cases.
Interesting idea. There are people here who don't understand what you mean by "trade vega" so why don't you talk a little in very simple terms about how you would execute a strategy like this?
Talon,
Good thread. Thanks for the effort.
I would like for you to discuss a little bit the 25 day exit.
Can you explain how you came about this exit strategy and this specific number? I find it interesting that the entry has several different criteria to it but the exit is simply get out after x period of time. In fact if I were to sit down and try to come up with a written strategy I would probably try to do the opposite and have an extremely simple entry with a more convoluted exit.
Also please discuss the 25 day number and more importantly how you came up with this figure. Are you looking for the highest return to select this number or were your results the best for the range... 20, 25, 30, 35 so you chose a value somewhere in the middle. Did you take outliers into consideration when looking at this number.
Good thread. Thanks for the effort.
I would like for you to discuss a little bit the 25 day exit.
Can you explain how you came about this exit strategy and this specific number? I find it interesting that the entry has several different criteria to it but the exit is simply get out after x period of time. In fact if I were to sit down and try to come up with a written strategy I would probably try to do the opposite and have an extremely simple entry with a more convoluted exit.
Also please discuss the 25 day number and more importantly how you came up with this figure. Are you looking for the highest return to select this number or were your results the best for the range... 20, 25, 30, 35 so you chose a value somewhere in the middle. Did you take outliers into consideration when looking at this number.
I find the entry very simple also. Buy when a change is announced, short on the actual effective date. Don't trade bankruptcies or b.s. like that. I only start off with that because I have gotten several PM's that suggest people think this system is complicated. It's not... it's the essence of simplicity. If it seems complicated it's because I did an inadequate job of explaining it.
There was no serious research into the exit. "About a month" is the concept we're working with. There's too much noise, etc to want to optimize this exit... and we do trade the exit with some discretion.
Another note to anyone looking at this... you must be looking at the RELATIVE performance vs the SPY. Don't just look at the stock.... it's possible that you bought a stock that absolutely collapsed, but if the % loss is less than the SPY loss, that's a winning trade if you're long the stock vs the SPY.
Are there more thoughts / questions on this or should we move on to another concept?
There was no serious research into the exit. "About a month" is the concept we're working with. There's too much noise, etc to want to optimize this exit... and we do trade the exit with some discretion.
Another note to anyone looking at this... you must be looking at the RELATIVE performance vs the SPY. Don't just look at the stock.... it's possible that you bought a stock that absolutely collapsed, but if the % loss is less than the SPY loss, that's a winning trade if you're long the stock vs the SPY.
Are there more thoughts / questions on this or should we move on to another concept?
i think this is a great thread, and hope it evolves for newer guys into how to test properly using excel, thx Talon.
Two questions i have, first off is how do i make sure i notice all announcements the day they get announced, i.e. is there a site dedicated to announcing these? Currently i have my news set up and filtered only for stocks on my trading system.
Secondly from what i am seeing from the graph off the site you first mentioned it seems like the day these things get announced they go down about 9%, well if they go down 9% that day and only end up at 11% on the effective day there is really only 2% in there to capture as i am assuming that when it is announced the stock is gapping down for most of the original 9% how do you get in before the 9% initial move happens?
A common mistake i see alot of new people make is coming up with an idea which is rock solid but very hard to trade because sometimes it can be hard to set up a screen for exactly what you are looking for. I have went through this problem countless times where i come up with an idea and originally and it makes sense but am unable to trade it because ti is just to damn hard to screen for it or create a screen where it will be shown to me before it happens. I have used both trade ideas and i use ALOT of excel, but i think when creating a system another thing you have to keep in mind is that whatever your idea is it has to be something which you will be able to FIND in advance of the trade. This is obviously a bigger problem the smaller the timeframe you trade on.
Anyways thx again Talon, looking foreword to reading the rest of this thread as it is nice to hear from other traders on here who actually know what they are doing. Dont let guys like bwolinsky get to you as there are far to many on elite chasing good people like you off.
Two questions i have, first off is how do i make sure i notice all announcements the day they get announced, i.e. is there a site dedicated to announcing these? Currently i have my news set up and filtered only for stocks on my trading system.
Secondly from what i am seeing from the graph off the site you first mentioned it seems like the day these things get announced they go down about 9%, well if they go down 9% that day and only end up at 11% on the effective day there is really only 2% in there to capture as i am assuming that when it is announced the stock is gapping down for most of the original 9% how do you get in before the 9% initial move happens?
A common mistake i see alot of new people make is coming up with an idea which is rock solid but very hard to trade because sometimes it can be hard to set up a screen for exactly what you are looking for. I have went through this problem countless times where i come up with an idea and originally and it makes sense but am unable to trade it because ti is just to damn hard to screen for it or create a screen where it will be shown to me before it happens. I have used both trade ideas and i use ALOT of excel, but i think when creating a system another thing you have to keep in mind is that whatever your idea is it has to be something which you will be able to FIND in advance of the trade. This is obviously a bigger problem the smaller the timeframe you trade on.
Anyways thx again Talon, looking foreword to reading the rest of this thread as it is nice to hear from other traders on here who actually know what they are doing. Dont let guys like bwolinsky get to you as there are far to many on elite chasing good people like you off.
- Status