why have you accepted the price-predictability assumption?

Quoting Wolf_Wicked

Disliked

{quote} The coin was flipped ... This was indeed non random, the result may be unpredictable but one thing is for certain, the coin will land showing either a head or a tail. Can anyone prove the maket is NOT random? Maybe for a day, a week, a month? Maybe half the time ? Maybe sometimes the market is random and sometimes it is not at all random... Why should it be one or the other? Perhaps random to me is completely different to random to you... Maybe from some perspectous everything in life is random but from some perspectous out of reach to a...

Ignored

Quoting Turveyd

Disliked

Yes on more random than the others, even if the random number dataset you've used is for some reason giving a uptrend from the look of it.

Ignored

Quote

Disliked

You have to give it an initial value, which you can easily pick from any real data chart for example i used an initial value of:
CLOSE: 1.2345 , OPEN: 1.2344 , LOW: 1.2342 , HIGH: 1.2347

Quoting Magix

Disliked

{quote} You have to manually assign open and close data... {quote} If you choose random when close is higher than open, this is a bullish trend and thus the chart should mimic this, even in a random generator.

Ignored

Quoting Magix

Disliked

Close[2] on most cases does not equal Open[1].

Ignored

Quoting Magix

Disliked

{quote} You have to manually assign open and close data... {quote} If you choose random when close is higher than open, this is a bullish trend and thus the chart should mimic this, even in a random generator.

Ignored

Quoting Magix

Disliked

this is a bullish trend and thus the chart should mimic this, even in a random generator.

Ignored

Quoting Turveyd

Disliked

Yes on more random than the others, even if the random number dataset you've used is for some reason giving a uptrend from the look of it. I'm thinking Pipups, simulated ones aren't pure random, there is a system there, trying to produce market patterns of a sort to practice with.

Ignored

public class RandomChart extends TestCase
{
    private static final double POINT_SIZE = 100000; // 1 / 0.00001 to get "pipettes"

    // Pseudo-random number generator
    private MersenneTwister rnd = new MersenneTwister();

    // counters
    private int up;
    private int down;

    @Test
    public void testMakeChart() throws Exception
    {
        // My dataset for E/U
        Instrument instrument = Instrument.fromSymbol("EUR/USD");

        // Create a new instrument
        final Instrument randInstrument = Instrument.fromSymbol("Random/EUR");
        randInstrument.setName("Randomly generated from EUR/USD");
        randInstrument.setContractMultiplier(instrument.getContractMultiplier());
        randInstrument.setContractSize(instrument.getContractSize());
        randInstrument.setCurrency(instrument.getCurrency());
        randInstrument.setForex(false);
        randInstrument.setPointSize(instrument.getPointSize());
        randInstrument.setMarginRequirement(instrument.getMarginRequirement());

        // Save it into my database
        Database.get().save(randInstrument);

        // Prepare an histogram
        final Histogram histogram = new Histogram(1);

        up = 0;
        down = 0;

        // Ask my tick feeder to send the real ticks from E/U
        TickFeeder.get(instrument).feed(new FeedListener<Tick>() {
            private Tick previousTick;

            @Override
            public void update(Tick tick)
            {
                if (previousTick != null) {
                    // Compute this return by tick and convert in pipettes
                    int diff = (int) Math.round(POINT_SIZE * (tick.getBid() - previousTick.getBid()));
                    if (diff > 0) {
                        up += diff;
                    }
                    else {
                        down += diff;
                    }

                    // Feed the histogram
                    histogram.push(diff);
                }

                previousTick = tick;
            }
        }, 0, Time.INFINITY, new ConsoleProgress());

        // Just some stats
        final int count = histogram.getCount();
        System.out.println("UP    " + up);
        System.out.println("DOWN  " + down);
        System.out.println("COUNT " + count);
        System.out.println("MIN   = " + histogram.getMin());
        System.out.println("MAX   = " + histogram.getMax());
        System.out.println("---------------------------");

        // Build the cumulative distribution function as a tree
        up = 0;
        down = 0;
        final TreeMap<Integer, Integer> tree = new TreeMap<>();
        histogram.feed(new IntegerPointListener() {
            private int sum = 0;

            @Override
            public void onPoint(int x, int y)
            {
                System.out.println(x + ";" + y);
                if (y > 0) {
                    tree.put(sum, x);
                    sum += y;

                    int v = x * y;
                    if (v > 0) {
                        up += v;
                    }
                    else {
                        down += v;
                    }
                }
            }
        });

        // Feed again the real ticks to get the timing
        TickFeeder.get(instrument).feed(new FeedListener<Tick>() {
            // Start from a dummy price
            private double price = 1.0;
            private long next = 0;

            // A bucket to store my ticks
            private TickBucket bucket;

            @Override
            public void update(Tick tick)
            {
                // Generate a random number from 0 to the size of the histogram
                int nextInt = rnd.nextInt(count);

                // Get the corresponding tick move according to the empirical distribution
                Entry<Integer, Integer> entry = tree.floorEntry(nextInt);
                int move = entry.getValue();

                // Turn it from pipette to a price
                price += move / POINT_SIZE;

                // Build a tick using the timestamp and the spread of the original tick
                long timestamp = tick.getTimestamp();
                tick = new Tick(timestamp, price, price + tick.getSpread());

                if (timestamp >= next) {
                    if (bucket != null) {
                        try {
                            // Save the ticks to disk
                            TickStorage.get().save(randInstrument, bucket, next - Time.WEEK, next);
                        }
                        catch (IOException e) {
                            e.printStackTrace();
                        }
                    }

                    next = Time.getSunday(timestamp) + Time.WEEK;
                    bucket = new TickBucket();
                }

                bucket.add(tick);
            }
        }, 0, Time.INFINITY, new ConsoleProgress());

        // Now I request a chart, my library will generate all the charts from M1 to W1 from the ticks
        ChartSetFactory.get().getChart(randInstrument, Timeframe.D1);
    }
}

Quoting PipMeUp

Disliked

{quote} Nope mine are random as well. Samples are independent and identically distributed. The difference is the probability distribution used. ...

Ignored

Quoting Proximus

Disliked

{quote} so absolutely no trend is present here.

Ignored

Quoting Turveyd

Disliked

{quote} Looks like there is to me, check the file on your random numbers. Equal + and - numbers I presume ?? Also, check the value of the +'s and -'s add up to the same, I'm betting you have more value in the +'s which is giving you an uptrend. Or just a fluke coincidence, never to be repeated.

Ignored

Quoting Proximus

Disliked

{quote} Its a fluke, because its random, if you look at the entire data you can see downtrends aswell randomly forming.But there is no trend in reality its only an illusion. But i made other mistake,check my previous post, by using only 1 random distance parameter, so i will correct that and then it will look perfect ok?

Ignored

Quoting Proximus

Disliked

Because it doesnt have to be, not all new opens start at the previous close, there is always some minor delay or gap which disrupts the continuity.

Ignored

Quoting Proximus

Disliked

so absolutely no trend is present here.

Ignored

Quoting Magix

Disliked

{quote} What would be the point of a chart that had as many holes in it as the one your generator shows? Nobody would trade off of this data, it sucks. HARD.

Ignored

Quoting Magix

Disliked

Absolutely no trend is present? Uh, look again man. No wonder you are so confused. If you cannot see a definite bullish pattern here, you'd best remove you programming cap and step back into trading for beginners. Again, I still fail to see the exercise in creating something that ticks 24/7 and 24/5 for ALL retail traders. This isn't realistic at all. What could possible be gained from this worthless data? Proving that you could create a random pattern? Kind of a fruitless quest if you ask me.

Ignored

            OPEN[0] = OPST.Text;
            CLOSE[0] = CLST.Text;

            for (int GO = 1; GO < size - 1; GO++)
            {
                if (TRADESOPEN[GO] == "1")
                    OPEN[GO] = (Convert.ToDouble(CLOSE[GO - 1]) + Convert.ToDouble(TICKSIZECLTOOP[GO]) * 0.0001).ToString();
                if (TRADESOPEN[GO] == "0")
                    OPEN[GO] = (Convert.ToDouble(CLOSE[GO - 1]) - Convert.ToDouble(TICKSIZECLTOOP[GO]) * 0.0001).ToString();

                if (TRADESCLOSE[GO] == "1")
                    CLOSE[GO] = (Convert.ToDouble(OPEN[GO]) + Convert.ToDouble(TICKSIZEOPTOCL[GO]) * 0.0001).ToString();
                if (TRADESCLOSE[GO] == "0")
                    CLOSE[GO] = (Convert.ToDouble(OPEN[GO]) - Convert.ToDouble(TICKSIZEOPTOCL[GO]) * 0.0001).ToString();

                HIGH[GO] = (Math.Max(Convert.ToDouble(OPEN[GO]), Convert.ToDouble(CLOSE[GO])) + (Convert.ToDouble(TICKSIZEHIGH[GO]) * 0.0001)).ToString();
                LOW[GO] = (Math.Min(Convert.ToDouble(OPEN[GO]), Convert.ToDouble(CLOSE[GO])) - (Convert.ToDouble(TICKSIZELOW[GO]) * 0.0001)).ToString();

            }

Quoting Proximus

Disliked

How can any trend be present in a random data, its a uniform random distribution , the fact that the price moves up is only caused by consecutive 1's after one another, it has no hidden pattern or whatever.Its just random 0 and 1.You are clearly failing to understand how random data works. But this also proves my point how people can be fooled to search for a pattern where there is no one, the human brain has this function to find patterns everywhere, but looking at a random data, there is just none.So if you see patterns in a random data that proves...

Ignored

Quoting Turveyd

Disliked

{quote} But your using a random number, then a lookup table to get the how big it moves each tick ?? correct ?? So if your Lookup table, has a positive expectancy then the chart will follow the same expectancy. There is DEFINATELY a upward bias in that chart, which means you have an issue with your random numbers. True 100% random, would get you a straight line, with noise either side pretty much. Off to DL your file and check your data files!!

Ignored

Quoting Proximus

Disliked

{quote} The data has a 0 expectancy because its random, no bias no nothing, due to the small sample size it may look like trend, and its very confusing, thats why look at a larger sample ... Anyway the new version is out maybe it was caused because of the previous bugs,which is now corrected and tweaked, so check out the new version and see of there is any "trend" you find in 100% true randomly generated data

Ignored

Quoting Turveyd

Disliked

{quote} Looking at it, still doesn't 100% look random, I fear your random number generator is suffering from being only Pseudo. Here is a basic, Excel Random chart, most random generation goes straight back to 20,000 the start number, the trends you see are because of the Pseudo generation effect sadly. {image}

Ignored

Quoting Proximus

Disliked

{quote} Its no random number generator... Man the data was taken from here: http://www.random.org/integers/ True random number generator from atmospheric noise.If you think thats not random, then what else can be. I can ensure you that all my data which i use is 1000000000% true random, not pseudo. If you dont believe in my programming abilities then check the code what i used, you can see how i count, add and substract the length of the ticks. By adding or substracting random uniformly distributed data. If you doubt the data, then generate...

Ignored