EA test misconceptions

  • 2023/2/24 18:47:49
  • read: 5
  • forextradingsessiontimes

 Recently seen many peoples so-called profitable cashback forex to achieve hundreds of times the effect, a careful look at the l forex trad forextradingmarketg session timest of transactions, you can find the following problems:  1, the choice of test chart, some people use very small chart, such as a minute, the accuracy of the chart is certainly different from the actual, will affect the actual operation effect if you want to dial the scalp, the operators spread should also be considered Within, especially each time to earn 10-20 points of the system is basically unreliable, stop loss so little is basically also knocked off with EA test, need to earn and lose in 50 points towards the upper system, the larger the more and the actual running difference is forextradingtime much personal feeling 15 minutes - 1 hour graph accuracy is higher, do medium and long term, the history of the test and the actual test results are basically close to the graph period and then larger, there will be unable to quickly follow the The trend of the problem (I only use the previous data, not the current data, so buying and selling decisions always have to lag behind a chart column, if you use the 4-hour line, the fast time may have to lose a few hundred points)   2, the data must be used historical, can not use the forecast of the current chart column calculation is all based on previous data, not according to the current data, not to mention the later data, otherwise your test has ahead of time, ahead of time, the result is that your success rate will be very, very high if you use the data of the current column, there will be and the actual results do not match the situation, because the changes in the current column and the simulation of the interpolation is completely different, the results of the error that is reasonable 3, with a large time period, you can not do multiple operations within the same time column, a good way is that a time This can solve the problem, but also reduce the CPU usage, is a more stable system often see someone with an hour chart, but all the buying and selling in the hour, such operations and the actual must be different, because EA is based on the interpolation of the estimated price direction, the actual walk is completely different will appear grape bunches of effect 4, the number of hands can not be too many. If too many, it will be affected by the actual spread, and the interest differential is different every year and every time, it is impossible to be well simulated, if the number of operating lots is very low, such as my system 10 years more than 700 lots, 70 lots in one year, the simulation results and the actual results will be similar, I have been right for each, after all, only 700 lots, they have checked the number of operating lots with the graphics have a great impact, do not believe you can put the loss of EA reverse, your results are still losses, the reason lies in the spread 5, data optimization, EA must have parameters, there is the problem of data optimization, the principle of optimization does not lie in whether the profitability to be high, but in the profit zone and will not change dramatically because of changes in parameters to find some parameters, so that in the parameters around the change is that the profit does not change too much, that is, to find a large profit zone parameters, otherwise your system is very fragile, once the market is not quite the same change as before, the system can not profit I am not talented, research 5 years, get the following trend system, is not very satisfied because the maximum drawndown is still very high, there are more than 30%, of course, you can get control by reducing the position, but the profit ratio will also decline, if the maximum drawndown in about 20%, then the system profitability ratio will be in about 1500%, all the other are still passable, the proportion of profitable transactions about 36%, but earnings when earning more than losses, but also very much in line with the Wall Street masters described the system, is ready to use it to earn money, huh   but the system have weaknesses, the trend system in the oscillating city should not earn money, oscillating The system in the trend will also capsize, want to find two kinds of money system, that is basically impossible, because if you can well distinguish what trend is now in, you already do not need EA system life is inherently flawed, not always perfect Remember, do not predict the future, regardless of how tomorrow, just follow the market on the line, the inflection point I can not catch, but I can in the trend I cant catch the inflection point, but I can keep up when the trend comes, when the trend reverses turn around