url

Machine Learning with algoTraderJo

If you have really no clue what to do with your money I may PM my IBAN

What is the fastest method to blow the account?

I know there are rules of thumb. But they are just that: rules of thumb. You can keep your NN manageable but I needn't a manageable NN I need a NN that learns and generalizes in the most efficient manner. You know Einstein's "as simple as possible ...

Machine Learning with algoTraderJo

This is why I dislike NNs. How many inputs? How many hidden layers? How many neurons in each layer? Which activation function to use (You may have a different one in each layer and even in each neuron)? How are the neurons linked to each others ...

Machine Learning with algoTraderJo

My guess is that the two NN kill each other. Say we have an up trend in place. - Both Algo1 and Algo2 say LONG. You open a LONG position. - Let's say the next two days they keep agreeing on LONG. The position is kept open. - The day after Algo2 says ...

Machine Learning with algoTraderJo

A trailing SL has no effect until it is hit. It can be seen as a regular SL. Partially closing is the same (up to the spread) as closing and re-opening with a smaller lot. Or it is the same (again up to the spread) as adding an opposite position ...

Assessing the Profitability of All Possible Trading Strategies

Shall we understand that B+C means open both a long and a short at the same time and same price? Do you realize that opening X1 lots long and X2 lots short is the same as opening a single order of size |X1-X2| plus paying additional spread? If X1>X2 ...

Assessing the Profitability of All Possible Trading Strategies

Additional order types are really useless and confusing. SL/TP/{buy,sell}{stop,limit} are no more than net exposure variation. Look at your graph (1) thinking exposure instead of orders. In the case n=1 your sell limit is simply cancelling the ...

Assessing the Profitability of All Possible Trading Strategies

OK. But why do you require the target to be (1,1,1,1,1) ? You only need the values to be non negative For example using (1, 1/4, 1/8, 1/16, 1/32) your solution becomes (1, 2.25, 3.625, 7.0625, 17.65625) -still a martingale-

Assessing the Profitability of All Possible Trading Strategies

Sorry but I don't get it. You say the rows of your matrix encode a scenario. The first row is (1,0,0,0,0,1) but the scenario n=0 is made of only one single buy. Why are there two 1's in this row? Also the columns encode orders. The first column ...

Assessing the Profitability of All Possible Trading Strategies

Can you increase the PF threshold to say PF>1.2? If this indeed improves the results you can make the threshold value adaptative. Something simple like PF>SMA100(PF)

Machine Learning with algoTraderJo

A RW is bad at visiting its state space. In clear text: "in some time the price isn't going to be really far from where it is now". It implies that the best target is the current price. It also means that if the price moves significantly it won't ...

Assessing the Profitability of All Possible Trading Strategies

Your idea boils down to the question "can a system generate a positive expectancy from a pure RW". Then to find such a system. If you assume the pure RW, no hedging can exist with any other instrument as the hedging exists only if there is ...

Assessing the Profitability of All Possible Trading Strategies

I'have downloaded the data from FXCM and Dukascopy throught their java api.

How can I get 99% of profit trades? this is my idea...

Hello and Happy New Year It may be good or not so good. It depends on your RR and your winrate. The RR must be greater than 2 otherwise you risk less and less... The bigger the RR the better. The greatest boost occur when you get a sequence of ...

Risking a n% of profit from previous winning trade on next

Thanks Patrick. I understand this scaling issue. It is important when mixing JPY and EUR. But in the lin. reg. of the price of E/U, all the prices and the returns are lagged values of the same time series. There are trivially the same scale, aren't ...

Machine Learning with algoTraderJo

Can you explain this sentence much better please? Preferably with mathematical proof. Predicting the next price is exactly the same thing as predicting the next return since the return is the difference between two consecutive prices. Based on the ...

Machine Learning with algoTraderJo

getSymbols("EUR/USD", src="oanda") pacf(EURUSD) # <--- (partial ACF,I don't know the name in R)

Machine Learning with algoTraderJo

I don't expect to find anything different from image EDIT: BTW I've never been able to make some use of this fact. It only makes really harmful problems with Kalman filters. I call that "flip a coin paradox" as I don't know if the phenomenon has a ...

Machine Learning with algoTraderJo

Where have you seen this heavy auto-correlation? I never found any. ARMA models' coefficients change almost completely at every bar. Redlion opened a thread because up to 160 bars he couldn't find any significant auto-correlation. The few articles I ...

Machine Learning with algoTraderJo