Not quite, because of interference from money management. If you had a constant trade size and a constant stop size equal to your target size going in (1:1 risk-reward, constant risk), then perhaps that might get you closer down to that 50% in accordance with @marketsurfer . No? In what form do you believe day trading to be possible? Why this obsession with automating what constitutes a craft (evaluating supply vs demand in real time)? I think this may be the root of this incessant argument: people on each side aren't talking about the same thing to begin with! By definition, a craft takes into account more variables than its textbook description, which is why it needs to be learned by experience and not books alone. Therefore, if you try to implement an algorithm for it, unless you can reach a point where you actually duplicate a successful practitioner's calls in real time, for a significant number of samples (say 1000 trades in a row), you cannot claim to have a full implementation of what's really going on inside his head, and thus cannot expect meaningful backtesting results from that incomplete implementation. Have you ever gone through such trouble successfully, developing an algorithm that managed to produce exactly the same trades as a price action advocate? If not, then on what other verifiable facts are you basing your preaching about randomness? If your claims weren't evidence-based, then you'd be a form of unsubstantiated guru yourself, like all those you so cheerfully enjoy pointing out here. :eek: Sorry, English isn't my native language and in the context of trading this can be confusing. You mean false prophets, right? Or do you mean that too many "gurus" out there claim profits which don't exist?
Each trade I decide what to do, always with the same stop. That's the only information you need to test for randomness. I have no target for profit taking. When my system says to get out, I get out, no matter how big or small my profit is. I get out because the trend will reverse, so stay in will only cost me money. If I have more than 50% winners it means that from all decisions I made the majority was correct. If I manage to have 60-70% winners it proofs that my decisions are NOT random. It is essential to keep the stop at the same level, if not you can manipulate the possibility to have profitable trades. Putting the stops further away gives you more possibility to stay in a trade till you make profit. If you make substantially more than 50% profitable trades it means that you don't enter randomly. 60% or higher will make the evidence more convincing. My only money management is my stop, because I never needed a more complex system to protect my capital. Never have a drawdown of 30% since I am consistently profitable, although I use high leverage.
(FYI I'm only trying to reconcile @marketsurfer 's claims with your own, I for one believe that price isn't random, just complex, but that's beside the point here which is demonstrating [non-]randomness.) Unfortunately it's not that simple at all, because the placement of your profit target directly affects the probability of your stop-loss being hit. As a gross over-simplification, if you were to always target +10% of the underlying, an overwhelming majority of your trades would end up at your stop-loss first. On the other hand, if your stop-loss is of say -2.5% and your profit target is +1 tick, the odds of having most of your trades turn a profit become very favorable even in randomly-placed entries. (Though, one should point out, probably not profitable overall because of how disproportionate the risk-reward is in that scenario.) Worse yet (from a standpoint of being able to reproduce the same results), your target is discretionary. That's profitable for you, but horrible for @marketsurfer to backtest. Lastly, let's not forget that a "win rate" is meaningless by itself: a high risk-reward system can perform very well with a 35% win rate where a 1:1 system might require 80% wins to achieve the same result (albeit with different drawdown patterns). Point being, given that a 35% win strategy can be consistently profitable, clearly (I hope!) we see that there is no correlation between win rate and randomness of the underlying instrument's price. 50% wins isn't any more or less an indication of randomness than 35% or 65% (each being part of an overall profitable system, that is).
I see it in a very simple way, I ask myself for each trade: can I make money in this trade? If I make money in 60-70% of the trades it means that I know when to trade and when not to trade. So it is not random. For me that is proof that my entries are not random. I agree that the total picture is more complex, so to make the picture more clear: in all my losing trades I never give 10% back of the profit I make in my winning trades, so even if I triple my losses I will still be profitable. My average profit per trade is much bigger than my stop (which is almost never touched) too. So these figures are favorable for non randomness. I take much more profit than the risk I take and still have more winning than losing trades.
I think we're close to an agreement, except that I maintain that a win rate isn't directly related to randomness. The fact that you look for specific situations for entering a trade, that you have sound money management, that, I see as showing discretion and non-randomness on your part. Your P&L, win rate, etc. just shows how skilled you've become, but it was already not random to begin with even when you had negative and eventually break-even performance early on. I wonder if @marketsurfer claims that price or our entries are random, but I contend that neither are. We know that price is complex, not random, since it is the sum of the opinions of a finite number of participants, algos included. Given that price isn't random, then by transition entries based on price cannot be random. If price were random, then sure, recognizing patterns in meaningless price action would yield meaningless trade entries, and luck and emotion/money management would be the sole differentiating factors between winning and losing traders. Technical analysis would then be, in essence, superstition at the same level as "lucky socks". Fortunately, AAPL going up for the last year and a half comes from demand overpowering supply, not coincidence.
@ VPhantom: Very well thought out questions and comments. Before I address some of your comments and questions, I want to once again apologize to OP for hijacking his/her thread. I earlier said I wouldn't, but this discussion is way too interesting for me to let it pass without comments. Let me address the point about win-rate first before addressing the other issues you raised. There are two metrics available to a trader (does not matter if you are automated or manual): (a) the validity of every signal within a context; and, (b) the expectancy of each of those signals. As you already know, one provides a measure of signal consistency, and the other its profitability. Since each signal is evaluated within a context, and such signal has a unique signature (or pattern or whatever you want to call it), the context provides you two things: (a) a level to place stops; and, (b) a level where one can confirm that the signal is working. For the purposes of this discussion, let us call this level a "marker". This is the reason a context is very important to a signal. Also, as you would guessed, the "marker" level is not a profit target, but just a conformation indicator. Profit targets are usually, but not always, based on subsequent Price Action. Hence, based on these two levels, one can evaluate the validity of a signal without needing a profit target. Many (I didn't say all) systems programmers/developers are not aware of what I said above. I used to be one, and was definitely amongst the naive. It was not until I started looking into how to develop a proper trading plan that I realized the importance of context. But enough about me -- way too boring! Anyway, so for people that are not used to context thinking, expectancy is an easy, and in my opinion, an incorrect alternative. The reason I think expectancy is a bad methodology is not only because we do not know how large or small profits will be, but also (and more importantly) because profit is a variable belonging to context and is independent of signal, where as stop-loss level is dependent on signal. So, back-tested expectancy of a signal is a useless measure. To make matters worse, I know of system developers who lump a signal that occurs in multiple contexts into one monolith, and calculate expectancy on it -- waste of time, effort, and the paper it was written on! As you can see, I made two assumptions: (a) back-tested win-rate of a signal will continue to exhibit similar win-rates in the future; and, (b) expectancy of back-tested signals will not continue to be similar in the future. On the surface the above claim could seem contradictory. However if one views the above statement through the lenses of a pattern-generating complex system then the assumptions can be reconciled. Signals (within a context) are patterns -- one cannot predict the nature of its occurrence, or its timing, but one can be sure it will occur within the said context. However, since the context (which is the environment in which the pattern occurs) itself is not exact to the back-tested context, profits cannot be predicted (remember, profits are variables of context and not of signals). My above response should have answered this error in thought. Be careful with the word discretionary. A manual trader is not necessarily, but could be, a discretionary trader. And, as to @marketsurfer: He is neither a systems programmer/developer nor a intra-day trader but likes to play one. "He who knows not, and know not he knows not, he is a fool -- shun him". All the best. Regards, Monoid.