Profiting from an unpredictable market?

I think understanding the market (core of the market) is understanding the lower tf. Most people say the lower tf has to much "noise" which makes it stressful to trade. Yet within that noise there are patterns where if traded you could enter inside of the breakout "before" everyone else has jumped in. The trick is to understand what is an extension and what is a reversal on the lower tf.

Hi Spaceduck

Does http://www.forexfactory.com/showthre...83#post8198883 summarize your point for this thread?

Chandan

Not really, fx cannot be traded the institutional way (as it's an otc market), and just "going with the momentum/trend" as i've mentioned in that post is just another form of gambling that obviously isn't going to be profitable in the longrun.

In my opinion there is only one proven way to extract profits from the markets on a consistent basis, i'm sure many of you already know such a way (perhaps you don't have it as a clear picture in your mind now, but if you meditate on the whole trading thing for some time, you will realize the irrefutable truth behind it all).

To share it here, sure it WAS the original purpose of this thread, i haven't done yet because i already know that most of readers would simply not understand it, the only ones who understand it are the ones who have already found it by themselves, so the whole sharing idea is but a nonsense.

Now that it WAS the original purpose of this thread, what's the new purpose ? Cause i'd love to know this proven way to extract profits from the markets on a consistent basis.

check out his post carefully, he never state to always got profit in the long run. most people always seek of 'perfection' from anything, including trading system. and math sequence combine with sequence of probability (scenario may occur) seem become most popular method as final approach. I do mean final approach when people just get too tired (maybe confuse) while using indicator as tools in seek for any market indication.
I keep spaceduck post related NFP scenario, his result are inspiring, and force me to print out those post (shame on me ). and to had a proper understanding, one should give a try on real practice, as I do. I saw many traders do some news entry check with their current platform, some use ctrader, mostly mt4. as for now I do simple approach by putting stop order right before the release, hope i've got the point he said after doing these several times.
in term of research we may do gamble with our entries, that's why i keep doing with several demo account. even with the real account one, I only risk my small ECN accounted with Tickmill, use their deposit bonus as initial risk percentage. with their zero stop level/order found more space to modify order and put entries. since during news release it's not the correct entry that's define our gain, but the exit, as for now I think the moving stop loss are the best deal for exit during that time.

Define predictability. Are we talking about predicting the mean/average of the next period returns? If that's the definition then a way to evaluate predictability is by root mean squared error (RMSE). BUT, is that necessary for profitability? Maybe not.

"To illustrate the difference between the approaches of the typical academic and practitioner, consider the following simple example. Suppose the investor has three currencies in his tradeable universe: dollar, euro, and yen. EURUSD today is trading at 1.50 and USDJPY is trading at 100. He forecasts the Yen to depreciate by 10% (leaving USDJPY at 111) and the Euro to appreciate by 10% (leaving EURUSD at 1.65). He therefore takes a long position of +50% Euro and an offsetting short position of -50% Yen in his portfolio, while remaining neutral on the dollar. Subsequently, EURUSD ends the period at 1.4850 (a depreciation of 1%) and USDJPY ends the period at 105 (a depreciation of 5%). The RMSE of the investors forecasts is 12.5%, much larger than the RMSE of the random walk model (4.8%). The investor has made two serious mistakes: the magnitude of his forecasts was totally wrong, expecting moves of 10% in either direction when only moves of 1% and 5% were realised, and he did not even predict one of the two directions of change correctly. Nonetheless, despite large errors in his forecasts in regards to both direction and magnitude, his portfolio made a 2% positive return. This is simply because his rank ordering for returns for the currencies in which he took active positions (i.e. EUR > JPY) was correct." (Melvin M., 2013)

Or are we talking about predicting the distribution of returns in the next period? For instance, the mean/average could be zero, but we can predict a high probability of a small move down and a low probability of a large move up, thus creating a skewed distribution. This kind of predictability is present to some extent in markets and have been a topic of study in recent finance literature.

Finally, most studies have only looked at linear predictability. But even then the results show that technical indicators DO HELP to predict future return. This table is from a dissertation I submitted recently. These are out of sample results, test period from 2009-2015. It shows that even on statistical fronts, we can find predictability in short-term returns of slightly less than 1% R^2 for the ELASTIC NET Price model, but guess what, like I said earlier, no predictability does not equate no economic value. This translates into very significant economic value over a model of no predictability. A typical risk-averse investor is willing to pay 636 basis points p.a. to trade the Elastic Net strategy over accepting a model of no-predictability and doing carry trade. These returns are actually pretty good because they are unleveraged returns. Of course any decent trader who knows what he is doing can do a lot better than these arbitrary models that are just dumb. But it proves a point, that markets are not completely unpredictable & predictability is not necessary for profit.

So if we go into the world of non-linear predictability, that's where all the technical analysis, fundamentals, and trading systems come in. We can't operate without context. Context helps shape predictability. Ofcourse I can't predict whether price will go up or down the next period, no one can. BUT I can predict a slight edge in mean/direction (i.e. price is more likely to go up) and predict a skewed distribution (i.e. if price goes up, it'll go up by A LOT, but not so much if it goes down) and these two are sufficient. And if I keep acting on these predictions using a good risk management plan eventually I'll make money in the long term. See this post where I talked about my trading philosophy and how I apply these ideas.

Great post ... Thanks.
This is something that i'm trying to do, by using the fundamentals...

Hi Thx Thx can check out this paper. http://www.rinfinance.com/agenda/2013/talk/JiahanLi.pdf

They use fundamentals to predict and find elastic net works as well. I used historical returns + historical technicals in my dissertation; adding fundamentals worsen the predictions and profits. I follow their methodology closely but used a weekly timeframe instead of monthly and incorporated more recent data.

So there is definitely some form of linear predictability, although not much... around ~1% of price movements. But such an edge is immense when translated into profits as you can see from the studies.

The takeaway here imo, is not that we should use these models to trade, but rather that it's possible to gain an edge in the market and the importance of money management to allow such edges to play out. If linear models using simple predictors can achieve such results, what about humans looking at contextual information and doing all sorts of non-linear thinking based on visual information? Rule based/mechanical systems certainly can work, it's just not easy to define everything and program something that works.

Also see: https://ideas.repec.org/p/ces/ceswps/_4238.html for an excellent article on Forecasting exchange rates from Investor's perspective. Written by fund managers from BlackRock.

Words of wisdom

Thanks for the reference, .. greatly appreciated ...
Completely agree ... this is why my current day to day trading activities is basically listening to ransquawk and trust my brain to come up with a plan on a minute chart ..

only few people willing to put more attention with these 'noise'. most of them may thing why bother with lower low, higher high at low tf while we got bigger picture. but in the end they (me instead) close their entry prematurly even they base analysis are comes from higher time frame. understanding the swing back, jump in, at core 'beat' upon low tf price range are tiring process. but it's all worthy when we finally got those market behavior masked at extension.