How to research and verify trading ideas

[dtrader98]

Talon, you do realize that the 2nd system you mention is also in the same book? I was just wondering, since when you were asking about why the add/drop phenomena exists, there is a very detailed discussion about arbitragers front-running the announcement and knowing that funds must buy on announcement date, not so much to beat the s&p, but because they are rated based on tracking error each day.

If you are interested in very detailed (and laymen, not heavy math) discussions from an academic about postulations regarding why these inefficiencies exist, go get that book.
It's very cheap on amazon (paperback).

Also, I was hoping we'd get to see someone reproduce at least the latest 08 results. C'mon seekers, you are given a perfectly good chance to have some review right here for free. I don't want to detract from Talon's thread, but I'm sure he will guide you on the backtesting (I'll be glad to help also), if you want to learn how to do it. The system discussion really isn't complete, IMO, until someone posts some results
(ok, there were a few, but not finished).


For starters, you can download the csv file at the site mentioned with 9 years worth of data. 2nd, just grab the data from 08 in excel, and then if you are not that familiar with programming, you can go to each stock on yahoo, and download roughly 60 days worth of data from Effective date to ED + 60, then calculate the running return for each set (as he mentioned AD is not as easy to track down, but not neccessary to run the pair reversion system).

Here's a tease: I took these straight out of the book from 2001. Notice the average on the plot shown earlier was 2.2%/trade AD+2. There are 20 trades from additions. The net gain in the book for the year was 45%, however, that was mostly attributable to 1 and only 1 trade which took place in the beginning of 01(and would have to be executed in late 00!). Notice this is a very slight variation on viewing Talon's system, but hopefully it gives you some insight and motivation to carry out the later years.

Talon, if I'm detracting or taking away from your thread, let me know and I'll zip it. It's refreshing to see you come along and share ideas.

 

[MarketOwl]

Quote from talontrading:

I think trading is a lot harder than most of these forums make it out to be. I think your chances of success are a lot smaller than most people would have you believe... but it IS possible.... and it's possible if you find good ideas, have adequate capital, and develop the behavior that allows you to implement these ideas successfully. Figure it will cost you $50,000 and 2-4 years to figure it all out out, on average.

Trading is only hard if you think like the average investor or trader. If you come in to trading and just observe the markets for a year, you start noticing patterns. Pattern recognition skills is probably the most important trait for a trader. After that, it comes to pulling the trigger and determining what your target exit point, win or lose. Target can be time or price based, or for the better traders, determined by feel.

 

[dtrader98]

Too late to edit earlier post, but I'm a bit sick today.

Funds must buy on effective date (not announcement) to minimize tracking error.
He also mentions that the phenomena was relatively recent (since announcement delay lag didn't happen till sometime in the 90s). It is the anticipation delay time and the tracking error mandate that allow arbitrageurs to benefit. Without the pre-announcement, it would likely disappear.

 

[talontrading]

Is that professor Singal's book? I read an advance copy of that a few years back and remember it being a good collection... in fact I probably still have it around somewhere. There are only so many easily tradable inefficiencies and his book, if I remember correctly, summarizes the important ones pretty well. I do recall he had some data in there on stuff like the January effect which, as far as we have been able to determine, is not tradable and may never have been tradable. Nonetheless, that's a book I would recommend strongly to anyone interested in doing this kind of work.

Detracting? lol... not hardly. Good posts and I think you're adding value to the discussion.

Quote from dtrader98:

Talon, you do realize that the 2nd system you mention is also in the same book? I was just wondering, since when you were asking about why the add/drop phenomena exists, there is a very detailed discussion about arbitragers front-running the announcement and knowing that funds must buy on announcement date, not so much to beat the s&p, but because they are rated based on tracking error each day.

If you are interested in very detailed (and laymen, not heavy math) discussions from an academic about postulations regarding why these inefficiencies exist, go get that book.
It's very cheap on amazon (paperback).


Talon, if I'm detracting or taking away from your thread, let me know and I'll zip it. It's refreshing to see you come along and share ideas.

 

[dtrader98]

Quote from talontrading:

Is that professor Singal's book?

Detracting? lol... not hardly. Good posts and I think you're adding value to the discussion.

Yes... and thank you.

 

[mikesmithv]

Quote from talontrading:

Let's open the floor for discussion. What are the issues with testing this system? How would you set up a testing framework? Where do we start?

So here is the big question. Are you going for relative returns or absolute returns. If you are going for absolute returns then I assume you would want to balance each long (good earnings stock) with a short (bad earnings stock) to cancel the effect of market direction and that would add a timing constraint. Is absolute returns your goal or are you comfortable just beating the market? I personally am only interested in systems with positive absolute returns. Do you feel that is realistic?

 

[talontrading]

You could set this up either way, but I would not be too concerned about being long/short. Rather, look at this as a system that gives a directional bias for individual stocks.

I'm not too excited about relative returns. I think it's one of the great scams of the industry. "Well the bad news is we lost 18% of your money this year but the good news is we outperformed the S&P 500 by 25 basis points."

Having said that, if you have a good relative performance system, then you could theoretically short the system against the S&P 500 and convert it to absolute returns. This is basically the concept of "portable alpha", but that's not really what we're interested in here.

Quote from mikesmithv:

So here is the big question. Are you going for relative returns or absolute returns. If you are going for absolute returns then I assume you would want to balance each long (good earnings stock) with a short (bad earnings stock) to cancel the effect of market direction and that would add a timing constraint. Is absolute returns your goal or are you comfortable just beating the market? I personally am only interested in systems with positive absolute returns. Do you feel that is realistic?

 

[talontrading]

You're still ignoring the huge behavioral / psychological component. I really think you absolutely must have 3 things to trade.
1. a real edge in the market
2. the right psychological makeup and learned behaviors to allow yourself to benefit from that edge and
3. a big enough bankroll.


the "3 B's" = brains, balls and bankroll.

Quote from MarketOwl:

Trading is only hard if you think like the average investor or trader. If you come in to trading and just observe the markets for a year, you start noticing patterns. Pattern recognition skills is probably the most important trait for a trader. After that, it comes to pulling the trigger and determining what your target exit point, win or lose. Target can be time or price based, or for the better traders, determined by feel.

 

[halligan]

Quote from whoisjohngalt:

It's really rare to see a good thread go on well without being a flame / troll fest. (Well... you don't see guru or some dumb vendor posting some dumb reply attracting the trolls) Anyways... I'll add a bit more hoping that it will continue as a great thread...

Being in the institutional side for some time... I've done a few more tests under the following premises...

1. Sharpe Optimization vs Add / Remove List. The reason why I did so is how the Long Fund Managers manage risk. 90% of the Long-only portfolio managers use BARRA or a feature similar to it using other packages to track their holdings and against tradable products.

Another is to programmatically tracking the top 500 market cap. list indepedent from the S&P 500 to anticipate the next stock that maybe added to index.

In another words... I have the independent list of top 500 Market Cap stocks and use Aegis-Optimizer like feature to track what a typical alpha seeker would be trading. Considering this being a mock S&P 500 alpha product, the model would only trade the stocks listed in the S&P 500 index.

.

Good insight. This thread is getting really interesting.

 

[talontrading]

Thanks for your insights and contributions here. We trade a lot of options, but mostly simple volatility spreads... so we hadn't looked at the correlation shift with the stocks vs index. a lot of options traders trade those kinds of strategy but so far that isn't in our setup list. maybe it should be... you've given me something to think about.

thanks

Quote from whoisjohngalt:

It's really rare to see a good thread go on well without being a flame / troll fest. (Well... you don't see guru or some dumb vendor posting some dumb reply attracting the trolls) Anyways... I'll add a bit more hoping that it will continue as a great thread...

Anways... good thread.