It has been gently suggested to me by some friends that this thread isn't a constructive use of my time.. nor am I doing anything actually that constructive as evidenced by my juvenile bwolinksy post last night. i have to admit, this is true.
rather than drag this on, let me bullet point where i was hoping to go with this and the really interested people can probably still get something out of it. i will post this massive post (which i'm writing off an on during the day.. so it may not be extremely cohesive) and then pretty much walk away.
To reiterate, the point of this thread was never to get anyone to follow me or make any trades I suggested. When I got started I had a lot of false starts and I actually owe a lot to the kindness and generosity of a few people who set me on the right path. I guess I was hoping to maybe pay some of that bill by hopefully helping someone else out.
The idea of the thread was to move from black and white simplicity to less-quantifiable "art". Here is what I had planned.
1. The index adds/drops. Doesn't get more simple than that.
2. The Earnings Drift idea. Also a quantifiable idea, but much more difficult in my opinion. For what it's worth, we do not trade this as a standalone system, though I'm absolutely confident you could do so. Rather, knowledge of this tendency informs our other trading--we make position sizing and even outright swing trading decisions knowing this tendency is probably in effect for a specific stock. FWIW, if I was going to trade this as a standalone I would look for a pattern like what RIMM has done here: a stock makes a large (say > 6 standard deviation) gap down on earnings (regardless of whether numbers were good or bad.. just look at the reaction), then trades in a quiet range for 1-3 weeks. Take a breakout of the range in the direction of the original gap, or even better a test of the "wrong side" of the range. Obviously lots of questions remain to be answered there, but that's a basic framework that should get you started.
I was also going to introduce one of our testing methods here which is basically to take many occurrences of an "event" and track returns for each of many trading days after. For instance, if a company reports premarket on 1/1/09 (i know, humor me.. and pretend there are no weekends either) that is day 1 and the return from previous day to that close is day 1. 1/2/09 is day 2 and the return from 12/31/08 to 1/2/09 is day 2 return. now imagine another company reports on 1/15/09 postmarket. day 1 for this company is return 1/15 to 1/16.. day 2 1/15 to 1/17 etc. (i'm sure bwolinksy could break out his study books and tell us what those returns are as decimals, percentages or basis points (sorry couldnt resist)) then you take all the return series and line them up so all the day 1's happen together. as a place to start you can look at this in two basic ways - you can take a "histogram slice" through any specific trading day or you can look at the whole series to see, for instance, if there's a tendency here that seems to roll off after day 5 or something like that. You can do this with any timeframe from monthly to 1 second (as we do for intraday testing). The programming is not that difficult (certainly easily doable in R) but the storage and processing requirements can become pretty extreme when you're working with higher frequency data.
Our goal here is pretty much to quantify everything as much as possible.. i think you got the wrong impression from the art statement.. but at the end of the day, many things remain unquantifiable.
3. An intraday S&P system. There are very very few intraday tendencies in the S&P that work. It is one of the most random and "noisy" markets out there, and I cant understand why 98% of ET posters claim this as their market of choice. There are a few easily quantifiable tendencies though and one we use is the tendency for breakouts midday to fail. There are a lot of qualifiers and the development work on this system (which is doable using 1 min OHLC bars) is a little bit more rigorous.
It is worth pointing out that this was an ATM machine for many years then, during the financial crisis, the market changed and we saw a lot more continuation than we had in the past midday. We went on a policy of "managing" this particular system by shutting it down on certain days.. a lot of quant desks do this with black boxes as a regular practice. Without that, it would have been marginally profitable through 2007-2009. With that policy in place it was wildly profitable. I bet not too many people can say they made money fading moves in the market during the meltdown.
The testing framework mentioned above is very useful for this kind of system as well.
4. Then we get to the art. I'm not being arrogant when I say this, but I'm one of the best 2 day - 2 week swing traders I have ever seen. I have returns of well over 100% for many years running.. without any significant drawdown. The slumps that I have had are most likely MY fault.. I had some personal issue or some bs going on in business or life that took energy from my trading.. rather than a problem of the markets or the methodology.
The real problem with this kind of trading of course is scalability. Can you generate those kinds of returns in a 500M fund? Absolutely not. 100M? Probably not, and you will be restricted enough getting in an out of positions that it's no fun.. however.. 25M? Yeah.. you can do that.. year after year. So it's not enough to attract institutional money to the fund, but it's certainly enough.
One of the curious things about my swing trading is that it has remained elusively unquantifiable. And believe me, we have tried. Computer science guys, physicists, mathematicians.. brilliant guys from schools like Chicago and MIT.. we tried this many different ways including the one that almost works of using some of the same algorithms used in facial recognition software.
In the end I know there are 3 things that contribute to the success I've had (knock on wood.) 1) superb risk management and control of position sizing (this is easily quantifiable) 2) an understanding of how markets react at inflection points and extremities (also rather quantifiable.. similar to #3 above) and 3) an intuitive sense of when momentums from multiple timeframes are about to come together in the same direction. This has proven teachable with a fairly long mentorship process, so I know it's transferable, but utterly unquantifiable.
Which is why, at the end of all of this, I come down to a certain amount of trading being art. What is actually going on is that the human mind is processing many different inputs in a complex web of ways we can't even really understand. The trader gets an impression or sense of the market and quantifies that into a risk / reward scenario and places the trade. It takes an incredible amount of experience and discipline to trade like this.. when I hear people on forums like this say they are a "discretionary trader" that usually means they are trading from their gut and I think they are absolutely doomed. Quantify everything.. don't be lazy (the label discretionary trader often is an excuse for that.. sorry but true).. push back the darkness as far as you can.. but maybe at the end you'll find at the end a little bit of mystery and wonder remain. i think i said early on in this thread that trading, for me, is as much an exercise in epistemology as making money! (And for the record, after some experience, I look much more for artistic backgrounds in the people I train than financial (ugh.. accountants make the worst traders), engineering (close seconds because they expect it to MAKE SENSE and it often doesnt) or mathematical backgrounds.
So.. that's it. I won't be posting here actively anymore. I will try to check in once a week for a while to answer questions if there are any. I think I just planned something that was too ambitious for a forum like this.. too much noise and not enough people really interested. If you are one of the interested people, then I think this should provide you with a direction. My way is absolutely not the only way.. nor is it necessarily the best way. It is only one way.
I also have been deluged with PM's from people wanting career advice. I tried to answer most of those (apologies if I missed yours). If you are serious about making it as a trader, I think you are MUCH more likely to succeed in an environment with other good traders. Sitting at home is not your best bet. Coming home from your day job, looking at markets, writing pretend systems (bwolinsky.. I'm looking at you kid) and posting on ET is a waste of your time. Do what you need to do to build some financial stability for yourself. If that's flipping burgers or waiting tables there's no shame in that.. get some stability, a small bankroll and then I think most traders would benefit from moving to Chicago or New York.
There are many good prop firms that will take newbies, train them and give them money to trade. Chicago will be more focused on futures and NYC more focused on stocks (which have a slightly easier learning curve imo) It can be a brutal interview process and you have to be careful because there are also many scams in the prop world.. but this is a new trader's best shot today imo. Do your research because I cannot recommend specific firms!
Your chances of making it with a good prop firm are way under 20%, but that's astronomical compared to the 1/100,000 chance you have trying to trade ES from your home. At least you will be around other people constantly talking about markets, and some of them are successful so you can learn constantly.
And that's it folks. The end of the line. Like I said I'll try to check in, but it won't be that often and I probably won't get to PM's (swamped with regular email) so I'm more likely to answer a question here in the forum.
Hope this was somewhat helpful to someone!
