One of the things hardest to accept is that the future is uncertain. In Fortune's Formula, one percept that was included that as # of sample increase, the results should APPROACH what the long-term odds should be. For example, flip a coin 10 times, 100 times, 1m times, 10 trillion times. While we know the odds are 50/50 in the real world we may get very different results. So, for example, you're "edge" might work in the first 1000 samples, but over the long-term it might be negative. This is one of the eternal questions for edge development (ie is your sample so small that its NOT significant), and how will you quanitfy in the future that you're edge isn't an edge at all (or maybe its BETTER than you thought). Interestingly, this brings the point to position sizing, once again referring to Fortune's Formula when Thorp realized that that the player had a statistical edge in blackjack (versus what was believed, it was very small). Yet people still consistently loss. I also have research demonstrating this with a positive expectancy game where most players lost (they're only option was to adjust their bet size). Why? Because even with coin flips 5 losers in a row ppl with lower their bet size, and 5 winners in a row vice versa. (Actually in big Drawdowns many traders oversize to "get it back"). Consistent position sizing is a HUGE HUGE edge in my opinion. I'll need to find the links to some of this research. So, at this point I believe that a) random entry/exits (say on the daily S&P with buy/sell) since it has long-side bias. b) consistent position sizing. c) Skewed Risk Reward (ie 3:1, etc). ... will actually unbelievably be a huge edge. Why? Consistent sizing even in drawdowns or equity highs, not panicking out at the wrong times, and skewed R/R meaning even if the odds were 50/50% theoretically it would still be positive expectancy. This would be versus variable sizing (ie smaller or extremely large during DD, oversizing at equity highs typically), stopping out because of professionals fading common patterns (actually giving WORSE expectancy than random), and interpretation of results affects future performs (overconfidence/underconfidence, patterns working not working, the world out to get you whatever...). William O'Neal stated (read somewhere) that one of his great disappointments was that none of his investors got the results that he posted. WHY? Because they added money at equity highs and took money out during drawdowns (This results in inferior performance is the long-term equity curve is up). This is HUGE HUGE stuff, imho and really attacks most of what traders believe. Unbelievable, this means the less you know the better you might do. This really attacks the psych of the trader. (I trade ATS btw). I'll be travelling so write and post some of my stuff when I can or when I get back. Cheers.
so mike what you are basically saying is that a ''edge'' is whatever trading behavior or technique that you have that is profitable to you. Therefore, every trader that makes money has a edge. Therefore, waking up in the morning, being on time at your trading desk, clicking the mouse when its time, trading a strategy that ''statistically significant'' probably meaning it shows profits in backtesting, are all considered like so, because they lead to your profit. If you are referring to a edge as the thing that good traders have that failing traders don't, then that would be called talent, experience, willpower (for discipline issues), etc etc i always laugh when i see traders talk about their edge like if it was the shit. you trade profitably, or you dont, period. my butter knife as an edge too you know.
I have done many tests on the S&P going back to 1996 with random entry and exit (long only). Here were the rules: A) On a given day, randomize a number 1 through 10, if its 5 buy. If not a buy, go to next day. We invest 100% of equity in each trade. B) If in a position (from A), randomize a number 1 through 10. If its a 5 sell. (Only do this if we're in already in a position from A). I ran 1000 versions of this test. About 1/2 of 1% (.5%) resulted in a negative outcome (I ran several times and got similar results). (No commission or slippage included). This on on the DAILY S&P. I used no STOPS (since we are randomly entering and exiting we should get some fraction of the return of the overall index). How does this simple system beat (meaning MAKES money 99.5% of the samples) versus most people who ACTIVELY , EFFORTFULLY, and PURPOSELY are trying to make money? (Since most ppl lose). Could it be that ACTIVELY trying to beat the market actually has a NEGATIVE EXPECTANCY? (Or vary position size at the wrong time makes positive expectancy negative?) You could argue that I am curve-fitting since I knew the market went up since 1996, and you'd be right; but one of the truism (though I don't agree with) is that long-term the market always goes up. If a system had a long-term flat expectancy, but went thru periods of winning and losing, how long would a trader trade this "edge"? (One of the things I find so entertaining is when traders are so certain that THEIR edge is truly an EDGE, in reality no one knows FOR SURE what the FUTURE results will be). Such a simple system outperforming most partipcants, unbelievable. Fascinating stuff!
Now lets attach a name to this system. We will refer to Lunar Waves, or Q Cycles, or some Astrology, or whatever. Now its a real system! Unbelievably, following our "Q Cycles" system, the average retail trader would actually DO BETTER than on his/her own. He (or she) would justify the monthly fees for our "holy grail" system because they would be HIGHLY LIKELY to be making money (tho it should lag buy and hold) with periods of out performance! Of course, we would tout all kinds of stuff like the developers being rocket scientists and the advanced particle labs we had to use to design this system. Our followers would believe us to be "gurus" of the market. Doesn't this pretty much illustrate everything that is wrong when it comes to accepted trading beliefs? What does this say about the industry?
Good replies. My comments were directed at the poster I quoted. Not necessarily towards your statements. I believe silvermotion's comment was sarcastic and rude (possibly wrongly so), hence, I replied with my own sarcasm. I'm all for an open discussion: a) "Statistically significant" likely means different things to different people. This is likely based on their mathematical background. Most important, IMO, is the way in which you test a an initial concept for a proposed strategy. One must be careful not to introduce bias (curve fit) during the in and out of sample testing. There are hundreds of things one can do wrong both before and during the actual strategy implementation... that said it is not my intention to start a thread about sound design principles. I am fortunate in that I have a strong financial mathematics background and put bluntly, this does make a difference. The system you mention is likely a system designed with combinations of curve-fitted data - the use of too many adjustable inputs without a solid fundamental base. b) Charts - as the original poster referred to - as in a 4.5 min to frontrun a 5min - do not in a such case produce a edge. We are all looking at price at the same time, specifically looking for price patterns, as you mention. Whether that be on a 30 min or 5 min time scale *seems* arbitrary to me (if some one can prove me wrong then please do so). A key distinction - this is definitely NOT to say that price does not show repeatable patterns - which it definetly does and can be used to define a systemic edge. My issue was with the use of time scale as the poster referred to, not in the usefulness of charts. c) Simple case example. Definetly not the only way. Doesn't mean anyone else's view is invalid, its just the way I look at markets. They are either testing price zones and reverting, or they are moving directionally. I did not mean to profess any more than that. My apologies if this came across as invalidating anything to the contrary. d) Ok. e) I know you said neither - I am venting a bit of frustration with that statement as it has been explored sooo many times. The point being twofold, 1) Everyone would be making loads of cash if random entry worked and 2) The selection of entry and exit *times* makes it inherently non-random - you would have to vary, randomly, all aspects of the trade to prove this theorem in any true mathematical sense. This involves the solution of some tricky PDE's where you have to set finite boundary conditions-- again another non-random selection opportunity... been there. If I have come across a close minded, well, then be it. Maybe its the case that after doing this for sometime now I have become convinced of what does and doesn't work. I will make an effort to be open to what you have to say about "random" beating out "pattern", it is always interesting to hear another thought process on the matter. Let's be clear though - are you making a case for or against T/A? I have no bias for/against it, its a tool that will provide me with *discretionary* information... but, systemically, TA has been "proven" to fail... right? Regards, Mike
Well sure... by definition you have an edge if you are profitable, right?? So why would you claim that an edge is an "illusion"? I guess that I still don't get it... Talent and experience come from many years of practice (that's no illusion). Say that between 700am and 800am you know (from years of watching) that fading any spikes in the ES will work 70% of the time (given the right opening)... Maybe one can't program this but it is definetly an edge that is quantifiable simply because someone is profitably executing it over time.... Can your butterknife trade? Does is trade spreads? What's the fee structure? Where's the website?
If you cant statistically prove your edge beyond any reasonable doubt, then you probably dont have one. If you dont know how to accomplish this, you are not a real trader.
Not wishing to disagree with everything you say, in fact you might assume agreement with anything not included below. You should also not read this as advocating the 4.5 minute bar. On the topic question: I believe that most successful traders have some combination of entry, exit, management, and state control that comprises an edge.
There is no such mystical thing some call an edge. This is a business of mathematical probabilities and human psychology