Brute Force

[W]hen I write here, I usually have in mind an investor, as a reader. Averagely, a professional with a certain wealth, that uses markets to park money with an interesting rate of return. There are a lot of available techniques that let you shape an investment idea and then you have to trigger decisions and operate. The return tends to differ wildly from your expectations. How many parameters does your analysis involve?

If you are a trader, not an investor, you have a quote monitor alive and watch the market on very short time frames, and your chart is sided with technical analysis indicators. Once I read somewhere that 95% of traders exhaust their capital in 12-18 months. How many parameters does your analysis involve?

The A model that powers this site, on daily base, has 81 inputs, at the moment, that range from stock markets around the world to commodities, from rates to forex and more. The weekly base model has 199 inputs, because some econometric data is added.

Well, it’s quite a large matrix. Dozens of columns and thousands of lines. Yes, the A model builds a very large set of past experiences that drive the forecasting process and model future market behavior.  All the sets of experiences (daily, weekly and monthly) are either code ordered and manually  configured, to obtain a mix of automatic and empirical output. Every single input is evaluated by the neural networks and contributes to the building of the output.


Posted by Luca in educational, free, model insights, 0 comments

Apocalypse Now!

The market has undoubtly taken the upside, and a strong and vigorous sprout is blooming. It will take time, but it will flower. When I say the market here I refer to the S&P 500 index and the index is booming. It’s like looking at a rocket launch: few seconds after the start it is at few dozens meters from the ground, pushing madly. Can you imagine in that moment where the satellite will be in a few minutes, hundreds of kilometres away?

Where will the market be in 4, 10 weeks? and in 12 months? Let me think… higher?

What I can say now is that the growth will be incredible! We are looking at a large scale common interest in parking money in dollars and this shifts part of this money into the market. A self breeding device has been activated: more value, more people interested, more money, index rises and next Christmas you will be eating your hat because you did not buy in July, afraid of any news and ticks. As the dollar evaluates, more money will be attracted into the US stock market, seen as more solid than government bonds. If FED hikes, more money into the dollar and toward US markets. If FED lowers, well, you know, market booms.

Uncertainty in Europe will push money into dollar. as well. Market will get mad. We will be stoned , I promise. Imagine the S&P 500 at 2400:  all the financial world would enter climax; then after few months we will arrive at 2600 and then 2700. Time for  Apocalypse, a world wide progression will flood the market with money: rising dollar + rising market = 2 gains with one shot. With the index at 3000 madness will rule. A vision from Bosch:1024px-Mad_meg

Someone will say: “This time is different” and this will induce us to open the most fantastic short of our lives.

The appreciation of the dollar will be crucial. Probably the dollar may sustain a 20% evaluation against euro or yen, but beyond that level an internal recession may be critical. A dollar too strong will push US economy in recession, with the rest of the world in a very fragile configuration.



Posted by Luca in forecasts, free, 1 comment

Enhancement to indicators

All indicators have been enhanced to a new adaptive behaviour, so that the indicators, when calculated, are never aware of any bars of the current move.

This was already partly true, but now this has been stated peremptorily.

The charts are much more interesting now, because Targets and Stops, for example, tend to form clouds of values that are self confirming.

The A model outputs at the moment five AI indicators:

  • Top / Bottom recognizer is a market excess locator, marks turning points
  • Position works on a deep learning base, provides the basic set of signals to enter and exit positions
  • Target estimates the goal of the acting wave
  • Stop puts a protective value : if a bars closes behind the Stop, well, something is going wrong and better get out soon.
  • 24 Bars ahead gives you a visual feedback

When indicators confirms each other a turn is promptly recognized.
The system always considers the execution of the orders at the opening of the next bar. This makes it perfect for part-time distributed investors.

I never thought, when working on the first experiments about S&P 500 forecasting, to create a trading system. It generated itself, maybe six months ago, once I completed the Stop and Target indicators: all the number crunching and training were ready and data could be extracted easily.

The Position trading system can be used as a swing/ position/ investing/ trading tool relative to how you set your entry/exit strategy and your preferred time frame.

Position trading system works on daily, weekly and monthly time frame, to provide a complete framework  of evalutions.


Posted by Luca in free, indicators, model insights

Enhancement to Signal

SPX-da-am-20160707-225903-914E63B3-ECBB-4429-AD03-03EE136D4636-3500-24Please note, this is just a part of a larger chart and it shows three indicators calculated by the A model. As the set has been trained, results are perfect. Last 8 bars are excluded from training and are guesses of the model bots. The image was produced few minutes ago.

We have the Target, the Stop and the Signal: Target points to trend exhaustion (cyan dots), Stop places the stoploss (yellow circles).

I introduced a modification to the Signal indicator. The original Signal marked the extremes with a triangle, either red (tops) or green (bottoms). Just the extreme. (And this, I believe, was the origin of many false signals).

Now, for any day marked as a top or a bottom, the day before and the day after are marked as well, with different values.

The behaviour of the indicator is as follow: the appearence of a first triangle must be considered as a warning. A second triangle is to be taken in consideration for position opening. I always consider that execution will be placed the next day, at opening, following the market and serching for a favorable entry point.  (Personally, I often look at hourly CFD rates of S&P500 to trigger easy exit or entry point).

For sure, the triangles may now be in overnumber and I will monitor that they do not invade the screen, but I feel interesting pushing the limit of the system to be as much confortable as possible. Now we should have a possible warning 2 days before the event and 1 day before the top or the bottom, plus a recurring confirmation of the flowing move, and I think it will make easier trading decisions.

For prudence, I always invite to check reciprocal confirmation from the various indicators before operating.

Posted by Luca in free, indicators, model insights

Evaluating position

Just a few words about the evaluation of the signals from the Amodel and their execution: I always consider the execution of the position opening or close at the opening price of the following day, so that any position can be easily managed, even unattended. My own attitude is that of an investor, not a trader, and I imagine that you have other important things to do rather than stay in front of the computer screen following the dance of prices. 


Posted by Luca in free, model insights

Weekly checkout

Today the S&P 500 index closed the week at 2057.14 and it was forecasted six days ago (last Saturday) at 2058.903.

Posted by Luca in checkouts, free

Big Bear Chase still open

The area 2060/70 is a strong resistance that will produce some turbulence, but probabilities are still on the side of a rising index, supported by all the barking bears around, by all the uncertainty and by the  weakness of the euro. A short phase of consolidation is not rare at this stage of the rise, something that may last one or two weekly bars, where some more bears traps will be set to work.

Acquisizione a schermo intero 23032016 115455The Elliott Wave analysis (AdvGet) confirms that the upward move is unfolding as forecasted, even if targets are slightly different from those previewed by the model. What we are looking at is – with much probabilities – the first wave of a new cycle: it will go for new highs in the 2180 area.

Posted by Luca in elliott's soup, forecasts, free

The Big Bear Chase

If any of my readers is keen in battlefield strategy (I’m not as much), she/he will recognize the beauty of the struggle we are looking at in the arena of the market. The hunters are the bulls. They will lay their net.  A long flat laying and the following action will be very fast, rapidity and suddenly. Even if you are long, now you may expect a correction even below 1900 and be tightening the stops.  As bulls are crazy, I’m led to believe that the move to come will be largely unusual. The Big Bear Chase is open for a couple of weeks.

Posted by Luca in free

The Amodel features

As this is an experimental service, that since the launch has been improved, I recently focused that the missing Stop indicator has to be developed seriously. While developing the Stop indicator, I modified the training model quite a bit, and as a result I found a Trading System already embedded in the model. More development. The Big Mess starts. I’ve learned that when The Big Mess starts, it is a problem in my undersized hardware. Days of testing, no exit.
Divide et Impera said Julius Caesar. Break the code in more and more units and then I have the control back. A real time testing is under way. The complete features of the Amodel are:
  • Bars ahead  – the price is forecasted for the next 24 bars ahead
  • Signal – a top and bottom detector
  • Target – a measure of the available distance to the next turn
  • Stop – a value that confirm the trend
  • Position – a simple, but detailed swing trading system

Of the five, only Bars looks into the future. The other four are readings of the present.

We have to test it in the near future and verify.
Posted by Luca in free, model insights

Under development


the screen of the training code, where positions will be ranked, with stops, targets and signals

The new trainer has taken shape: it should provide a better preprocessing of the model and has just been tested. Now it will be applied to the basic SPX data to have each bar analized and ranked. More testing ahead, it is the basic process for a neural network model. This new learning model is much more rich than the previous, and I hope it will enhance the performance.

It’s a long time I’m thinking about a stop indicator. The stop is the key factor for every position and is the value at which your position will be closed. I hate stops. The stops you usually find on trading platforms have big  problems of responsiveness to volatility. The stop must be a sort of safety net to protect the position profit.  What I’m searching for as an agent that guesses a reasonable stop for every bar close.  Now I’ve tested the new stop for the model and seems to work fine. A larger testing ahead, but as an effect of activating a stop, I noticed that a trading system is taking shape. At the moment, I’m more concentrated on having all the bits in place, but a sort of trading system may take shape soon. We will have a signal system, targets and stops, plus a view into the future: all the building blocks for an investment project.


Posted by Luca in free, model insights, r.Virgeel

About targets

[su_column size=”1/1″]target_sample

[T]he target is an instant reading of the price potential energy. When the price serie is trained, the best possible trades are marked with a start and an end, that generates the arrows  that signal market extremes. Then taking the average of the last day of the position and the successive day, the difference for every day of the position  from this value is calculated: for every day there is a percentage af value that still has to be gained until the position has to be terminated. The chart is quite simple, isn’t it? The main idea is that whenever the price is above (or below) or even very near to the target, the movement has finished it’s energy. This is for the ex-post and the training phase.


[su_column size=”1/1″]

Now, what about realtime readings? Here is the sample from today. Undoubtly the cyan dots (the targets) are quite confused, but it’s just what the model reads from the never seen before data. Generally speaking, we may note that:


  1. usually the target stands on the side of the movement, over the price during the uptrends and below the price during downtrends
  2. targets tend to “converge” toward the reversal areas
  3. by construction the scattered dots marks the most sensible support/resistance areas

In conclusion, the target is a confirmation tool that add an interesting information that can be read from price action inside the model.





Posted by Luca in free, indicators, model insights

Traditional analysis is out of touch

lm_3-4 [I]’ve found and corrected a couple of minor bugs in the Signal and in the target models (minor logical errors that under some circumstances did crash the whole code!) and this has made me think that the Signal model acts similarly to the bifurcation model under fractal theory.

The market is a chaotic self-adapting structure, it is asymmetric, but regular and continuously passes through levels of bifurcation: the side image is quite near to what I’m suggesting.

I’m not a mathematician, so I cannot explain the theory behind, but I can recognize that a similar structure lies inside the market, as the result of the interactions of all subjects (traders, investors, scalpers, big and small money managers, etc).

In some way, I think that this complexity is what nowadays has made the market so “technical”, or difficult to read and play. And why traditional analysis (fundamental or technical) is getting out of touch.


Posted by Luca in a.i., free

In memoriam of Jim Slater

2015-12-21 16.45.55[I]’ve just discovered that on 18th of November, Jim Slater has died, at 86.
Here I want to remember him, whose book “The Zulu Principle” back in 1992 opened my eyes on how to approach share investing. If you are interested in investing in stock shares and are new to the field, this book is a great reading. Again, thank you Jim.


Posted by Luca in free, generics