How to research and verify trading ideas

[talontrading]

Good thinking but don't do anything aside from collecting the tickers and relevant dates yet. IMO you're overcomplicating slightly and I dont want you to waste time and energy.

I'll try to post more midday.

Quote from The Big D:

The other thing I forgot was sensitivity analysis on the data - I'd probably ask "how many of my best historic trades would I have to replace with my worst before this method tanks". If the number is 2, I need more data. If the number is 17, I'm probably good to go.

 

[The Big D]

Quote from talontrading:

Good thinking but don't do anything aside from collecting the tickers and relevant dates yet. IMO you're overcomplicating slightly and I dont want you to waste time and energy.

I'll try to post more midday.

I may not even get around to doing anything for a while - that's just what I would do if I were going to trade this. I've got something else that's chewing up most of my time right now that I think has a better return.

 

[mikesmithv]

Quote from talontrading:

it's not about hedging... it's about excess returns... so you have to offset with the market.

OK, I think I get it now. Essentially every trade is offset with an SPY order, it's just that when there is an opposing stock in the opposite direction the two SPY orders "cancel out" and are not required. This works because the equal-but-opposite market component of the opposing stock serves that purpose. Correct?

 

[bwolinsky]

Nothing about my posts were detractions. I asked questions that anybody wanting to <i>research and verify trading ideas</i> would want to know. He gave only two references, which really isn't sufficiently large enough to say anything.

What would instantly gratify me is a list of all the S&P 500 changes made going back probably decades. Till I have that, this thread is useless. Without real time calls, there's no point to the thread.

 

[talontrading]

<sigh> what an idiot.

i didn't give two references... i gave the two most well known academic papers... and i also gave a link from the s&p website that gives the index changes going back years.

so you should be instantly gratified now right? what more am i missing?

why is this guy such a prick? is he just like that?

Quote from bwolinsky:

Nothing about my posts were detractions. I asked questions that anybody wanting to <i>research and verify trading ideas</i> would want to know. He gave only two references, which really isn't sufficiently large enough to say anything.

What would instantly gratify me is a list of all the S&P 500 changes made going back probably decades. Till I have that, this thread is useless. Without real time calls, there's no point to the thread.

 

[bwolinsky]

Quote from talontrading:

<sigh> what an idiot.

i didn't give two references... i gave the two most well known academic papers... and i also gave a link from the s&p website that gives the index changes going back years.

so you should be instantly gratified now right? what more am i missing?

why is this guy such a prick? is he just like that?

All right, so how'd you do? A tradelist would be very helpful. I'm not asking for your statements, moron, I'm asking for you to produce your backtest yourself, because it's not on the forum to prove your theory, it's on you. I'd accept a date, an open price, your sell price, when you sold short, and when you covered.. That's normally how to prove a system, not with this random oh you look it up shit. Till I see specifically where it calls for a 25 day hold time and when to buy, sell, sell short, and cover, this isn't getting anywhere.

I have no idea why a nothing alias would come on trying to prove a system that has no real academic proof. The papers would only list correlations without hard line profit. That's assuming your system is articulated verbatim from your original post, which is very doubtful. Till then, it's on you, I'll be waiting.

If you are actually able to produce your backtest and I can verify two of the trades, that'd be good enough for me.

 

[dtrader98]

There really is no need to dispute the veracity of his claim. It is so well known, that it is even displayed in popular laymen books on trading, such as this older one, 'beyond the random walk.'

All of the information to replicate the results for one year (more recent) can be found at the s&p site he mentioned.
Someone can step up to the plate and replicate the graph results with the latest data at S&P site. How hard is that?

If you aren't satisfied with one year, they give 3 years, and you can likely pay for the entire history. My intuition is that the phenomena will not be as pronounced in the most recent year as the posted example shows, due to efficiencies I mentioned. I'm not going to do the work, but it sure isn't that hard.

Next thing you want to verify is that the net win from all the winning trades exceeds net losses (since the results in the attached chart are for average returns). Very easy to backtest, and very simple rules.

And by the way, the method is based upon large numbers, not one or two trades. You would need to make all or many of the announcement trades in a year in order for expectation to work in your favor.
That's how statistics works.

 

[talontrading]

Ok I guess I should cut you some slack since I did call you a prick... but you have to realize the point of this thread is how to evaluate a trading idea.

This is NOT how you do it: find someone who says they have traded it and demand verification from them that it's a good idea.

I chose this idea because it works and is easy to evaluate, but for some reason you don't see that the process is the point of this thread... instead we have wasted several pages on your bullshit... which maybe is my fault.

I dunno... I gave you a link that gave the dates for all the trades... you can assume I did the trades on those dates and figure some shit out for yourself.

As for your repeated claim that this has no academic proof... I looked at your other thread and your profile. I find it very unlikely that anyone who is a CFA candidate would not know the basic research on market inefficiencies. Isn't that still part of the CFA curriculum?

Don't you claim to have done some graduate work in finance... interested because anyone who has taken a graduate level course in fnance, or frankly anyone who has done any stock trading at all, should already know about this tendency. How is this news to you, a professional money manager who manages several million dollars?

I have interviewed 22 year old kids straight out of undergrad at second tier schools who could discuss this tendency and associated research in detail... I dunno am just kind of confused.

brutal trading day and i've had all the idiocy i can stand for one day... will post more constructive stuff tomorrow.

Quote from bwolinsky:

All right, so how'd you do? A tradelist would be very helpful. I'm not asking for your statements, moron, I'm asking for you to produce your backtest yourself, because it's not on the forum to prove your theory, it's on you. I'd accept a date, an open price, your sell price, when you sold short, and when you covered.. That's normally how to prove a system, not with this random oh you look it up shit. Till I see specifically where it calls for a 25 day hold time and when to buy, sell, sell short, and cover, this isn't getting anywhere.

I have no idea why a nothing alias would come on trying to prove a system that has no real academic proof. The papers would only list correlations without hard line profit. That's assuming your system is articulated verbatim from your original post, which is very doubtful. Till then, it's on you, I'll be waiting.

If you are actually able to produce your backtest and I can verify two of the trades, that'd be good enough for me.

 

[talontrading]

short post here... here is how I would evaluate this idea.

1. Get the entry dates. You need the announcement date and the effective date. (I am a huge believer in the 80/20 rule.. it is possible to expend 20% of the effort and get most of the results... so there's a shortcut in this case... you may want to start with only the second part of the trade, the actual effective date because it's harder to find the actual announcement date. eventually you will need to do that, but only do half at first to see if it's worth your time to go further.)

2. record the closing price of the stocks and the spy on those dates. in actuality, we hedge with futures here, but spy is fine.

3. calculate the percent change from the effective date to 25 days later for each stock and the spy. then subtract the spy change from the stock change to get the excess return.

4. some basic stats will suffice here: mean, median and standard dev of returns. i would plot a histogram and maybe boxplots (the point is to understand the outliers.) understand this analysis is only done on the excess returns.

that's a rough outline of the first step... and it's probably less than an hour's work... and that bwolinsky is the point of the thread... not me making "real time calls" (see how silly that sounds?!)

ask questions if you got 'em and i'll try to give coherent answers.

 

[kiwi_trader]

Here's an idea.

We'll all put bwolinsky on ignore. Then NO ONE reply to the sod. And we'll never see his posts again. OK. Hes on my ignore list - please do the same.


(Top of the page :: Your Account :: Edit Ignore List :: and just put the suckers name in there)