How to research and verify trading ideas

Discussion in 'Strategy Development' started by talontrading, Nov 2, 2009.

Thread Status:
Not open for further replies.
  1. Every few years I post a bit on some of these boards. I don't usually find it very productive and it can be very frustrating, but I'm hoping maybe I can help someone along the way. There were people who were very generous with their time when I was starting out, so maybe I can pay something back here.

    I replied to a few the "market is random" (it's not) and "it's impossible to trade" type of threads but my message it a bit at odds with those threads. I have built a career finding places the market is not random and finding trading edges. So I thought maybe it would make sense to start a new thread with the correct intent from the beginning.

    Here's what I intend to do in this thread:
    1) Discuss how to find, prove and apply edges in the market. That may seem ambitious but I can tell from some of the PM's I've gotten (and honestly from my experience of 15 years in this business) that people don't really know how to do this.

    2) Open people's minds a little bit. I think 99% of the people on the internet are looking for a system that involves buying when a 9 day RSI goes under 30 and the stochastics are crossing and the 11 period SMA is crossing under the 6 period XMA... ENOUGH! lol. I think those kinds of systems don't really work very well for the most part and I haven't been able to verify a statistic edge to those indicators... so I don't use them in my trading at all. I want to get people away from looking for an ES daytrading system using OHLC data and thinking about some other possibilities

    3) I hope to give people some tools and ways to understand that market. Maybe even clear up some misconceptions along the way.

    What I will NOT do here:
    1) Provide any personal information or verification of performance, etc. It is rude and completely in appropriate for anyone to ask because... (read on)
    2) Sell anything at all.
    3) Conduct any private tutoring.
    4) Promise to post on any sort of timely schedule. I may even vanish completely one day and not come back. Sorry but that's just the way it is going to be.
    5) If this thread is hijacked by trolls and fills up with BS... I don't need that... see #4.

    Ok... let me kick it off with this... One of the "no one can trade so you should all just index" threads I posted a system that I have used with good results. I got a few PM's but it was completely ignored by the thread. Interesting... because you don't see too many profitable systems posted in their entirety for free. Here it is:

    Index Adds / Drops from S&P 500
    On the days following an announcement that a stock will be added to the S&P 500, work a limit order to buy that stock at the settlement of the announcement day. If not filled, don't chase the market. If not filled by a few days before the effective date, cancel the order.

    On days following an announcement of a drop from the S&P 500, work a limit order to short that stock at the settlement price of the announcement day.

    On the close of the day that the change is actually effective, flip the position. If you were unable to fill on the limit order then get short the addition and get long the deleted stock.

    If there is no offsetting position (for instance, an addition announced without a deletion) then "hedge" the position with equal dollars of SPY or futures (though you will be creating a taxation issue with futures.)

    The portfolio is hence always long/short equal dollars. If you buy $100K of ABC and sell $100K of XKY on the announcement day the values might be +125K ABC and -112K XYZ on the effective date. Doesn't matter, on that day you then short 100K ABC and buy 100K XYZ. The idea is you initiate each position equal dollars. It may be possible to be beta neutral or something, but betas change... so keep it simple.

    Exit the trade 25 days after the effective change date.

    Do not trade deletions due to corporate actions, takeovers, bankruptcies, etc.

    Think long and hard about position sizing. There is no stop so theoretically the entire investment is at risk (and... um... you're shorting so think about what that could mean.) Normally, I like to risk no more than 2% of my account on any trade, but that's really not possible here. Maybe allocate 15% of capital to each trade. Be aware that you can go many weeks with no positions or have on 4 positions at once.

    That's it... do the work... it's hard to backtest because it's not like Buy when the 3 period XMA crosses over the 28 period SMA and the RSI is > 70 and ADX is < 15 or bullshit like that... those kinds of things don't tend to have a great edge, but here I have just given you a system that is supported by a vast body of academic research and works in real time.

    Not one person here will do this for a year. Hard to believe but I would be wililng to bet.

    I think my explanation is clear, but I think that because I wrote it lol. If it's not ask and I will attempt to clarify.

    [I have gotten some PM questions and requested clarifications... will get to those here sometime soon.]
  2. I just ask you to post percentages of your recommendations, something like short 20% XYZ, buy 20%ABC, that's all. This thread is pointless without actual calls on the system you've described here. It's intriguing, and impossible to backtest in any platform I'm aware of.

    Also, will you please provide a bibliography of the research and academic literature you claim this is based on. I'd like to read it. No, really, I would.

    I already know the system will have very low degrees of freedom, as S&P500 changes have happend and will continue to happen irregularly, and it's debateable how many actual trades you have based your assumptions on. Hopefully more than 30.

    Not sure this thread will last with a 65 total post member since 2003. Anyway, Good Luck!

    ex post edit: PS:I'm not sure why call this how to research and verify trading ideas when you don't appear to attend to verify this system. I also hope you can actually produce credible sources for the basis of your system.

    Last question, when you say 25 days, do you mean 25 trading days or bars? or calendar days?
  3. First of all... I'm not making any calls or recommendations. Respectfully, you're missing the point I hope to make which is to teach people how to fish... how to think for themselves.

    I'll post a bibliography of sorts but you can also google I'm sure. Someone in the other thread found a little SeekingAlpha blurb they posted so I think 30 mins of google research on "Index Additions Deletions" or something like that will reward you.

    It's not debatable how many trades I've based it on. :) More than 30. Way more.

    Now... about that backtesting... how could you do it? that's a good question... and VERY close to the point of this thread.

  4. ex post edit:
    It probably won't last lol.

    I have verified it already. Not my job to sell it to you but you're asking the right questions.

    25 = trading days. The exact number is fairly unimportant. 20, 30, 15 probably... a calendar month... doesn't matter a lot.


  5. No, your premise is how to <i> verify</i> trading ideas, and until I see an actual simulation, it's not something I pay too much attention to. People only care about walk forward results, and your task is to strategize based on past data the best you can to trade tomorrow.

    I will really only accept journal literature as being academic content. I do not consider sources of information like personal blogs, or sites like SeekingAlpha, academic literature. Sorry.

    Well, I just would like to see your symbol list. You appear to type good, and I'm sure looking at your entire account history would produce the list quite readily. Note I don't care about position size or anything like that. I just want to see the symbols, then I can backtest myself, since we are learning how to research and verify, you know?

    I don't recall very many in the last two years, perhaps you could enlighten us? GM is about the only one that comes to mind immediately.

    I see, a good starting point would be your historical trades list, for verification purposes only.

    Sorry about the bold, I didn't mean to do that.
  6. Talon: With due respect, I find your example of systems too complicated. Profitable systems can be written on a lunch napkin, and should be based on mathematical results.

    In addition people in this forum do not have a lot of trading cash. So all systems that require liquidity, not moving markets, etc, are really not relevant for the majority of this forum.

    The majority of guys here want to make some money to earn an easy living, and reach a point where they can give money to charities. I believe they also want to get out at the end of the day, so no holding overnight. They can always get in in the morning. They are flexible people.
  7. They are right not to do it for one year. They want to see the expected value soon, not dream about a return in a year from now, that is not even sure.

    People understand or should understand that they have to run multiple trials from a random variable, but they want to do it so that the result is known say in a week's time or so by playing a strategy that applies to multiple vehicles.

    If all you will say is wait 25 days, and do it for a year to see the result, I am sorry but just stop this thread. It is useless.
  8. Ok I'm going to take a minute to answer this in detail... and I'm not going to be careful to be incredibly polite so apologies in advance if I hurt your feelings

    Look. I trade full time and I do a lot of different kinds of trading. Stocks, futures, options... no forex or fixed income to speak of... intermediate term down to daytrading. I chose a system that I have used for a while with good results and posted the complete rule set here. I chose something that is already well known that I have no ownership of and honestly didn't even do that much research on. This is what we call in the business a "lay up". (Though do watch your position sizing if you're going to trade it...)

    I don't care if you pay attention or not. If you don't... someone smarter paying more attention to details will... and this thread is for them, not you. I am not going to do the work for you and give you an "actual simulation" (I dont even know what the f*ck that is btw). My premise is exactly how to verify trading ideas... so my advice is sit down and get out a pencil and paper and get to work.

    Ok... I don't mean to be a prick but just so you know I also have an academic background... the real deal... frankly, anyone who has done any research or even casual reading of academic research already knows about this body of research. for you to say with arrogance that you will "only accept journal literature as being academic content" tells me that you probably couldn't understand the literature. if you could, you wouldn't be asking the question. but since you asked... start with these...

    Chen, Honghui, Singal, Vijay and Noronha, Gregory , The Price Response to S&P 500 Index Additions and Deletions: Evidence of Asymmetry and a New Explanation (June 2003). Available at SSRN: or DOI: 10.2139/ssrn.427001

    "An Anatomy Of The 'S&P 500 Game': The Effects. Of Changing The Rules," Journal of Finance, 1996, v51(5,Dec)

    Do those meet your rigorous criteria? Those are two of about 20 well known papers out of several hundred on the topic. Anyone who knows the literature should already know this.

    You know? Yeah I know. You know something? With that last question you have just proven yourself to be out of touch with markets because there have been a handful more S&P 500 adds / drops than GM lol.... in fact a few very recent ones come to mind. Pay attention to the news.

    I'm not spoonfeeding people... And I find your tone annoying and out of place here.
  9. spinn


    this method sounds like gambling at best.

    I would not touch this "method" no matter how much research the op thinks he has done....

    also an endorsement by blowinksy is the kiss of death.
  10. How do you so completely miss the point. Get a f*cking pencil and paper. Go back 5 years and find the adds and drops from the index... calculate the returns and see what the strategy would return. This system is only a starting point for the discussion... somehow you are missing the point.

    Given a trading idea... let's start with this one since it works... given a trading idea... how do you verify it and research it BEFORE you risk your money? That is the question of the day here.

    When I first decided to do this trading idea it took an intern less than a week to crank out the numbers and stats on this. It's not complicated.

    What do you find complicated? This is by far the most simple trading system I use so I guess I must be really confused about what a profitable system is. I didn't realize they could all be written on a napkin.... Crap... I'm going to have a find a new job.

    And please define a "mathematical result"? What do you mean?
    #10     Nov 2, 2009
Thread Status:
Not open for further replies.