How to avoid over-optimization of EA

Whatisforexrebate 6Browse 0Comments Collection often mentioned MT4 h Whatisforexrebatetory bestforexrebate useless theory, before there is indeed a history test What is forex rebate the real test does cashback forex match the situation with mt many times upgrade, compared to the real test and history test, now can be said to be more accurate, history test for optimization encountered more is the problem of over-optimization, this and mt history test The problem of over-optimization is not related to the precise test will also encounter this problem, then again, you can not spend a lot of time simulating the test for each ea, say too short test may not be able to respond to the full picture, historical testing is an essential tool, depending on how you use, how to avoid forexrebatenetwork optimization over it? The first two stories here: ◎ someone practice archery, but always shot inaccurate, see other people arrow red heart, can not help but eye red, then he thought of a good way, is to shoot the arrow first, and then run to the arrow into the target to draw the arrow target, so it seems, also arrow red heart; ◎ also said that the British football culture is prevalent, a certain gentleman also want to through the football to get a hand, but he always guess not, so he thought of an idea. The weekend before the game, he selected 3,000 people through the Yellow Pages as the target, respectively, letter, claiming that they can accurately predict the outcome of the weekend A team against the B team, of which 1,000 people, he predicted the A team to win, another 1,000 people, he predicted the B pair to win, the remaining 1,000 people, it is predicted that the two teams battle and the next weekend, he predicted the accuracy of the 1,000 people last week and divided into three wholesale letters, respectively, to make three predictions. He made three predictions for the next week, and the next week... By the time 33 people were left, he wrote to them separately, stating that if you want to get a prediction of this weekends results, you need to pay him £1,000 first, as a result, most of these people have long been attracted to this unique expert, and a certain gentleman finally made a fortune The so-called optimization of excessive, is the use of historical data matching system, for a period of historical quotes and indicators, the relationship between data to write EA, in order to make EA seems to have a good performance, and constantly make adjustments to the parameters, optimization, set filter conditions so that EA and historical information data between the exact match, the results can be sure that this set of EA in the historical data testing good performance, will be in time to buy before the big rise, before the big fall in time to sell, but when the next big rise and fall before the EA will be timely signal? Im afraid not, because the EA is written for past conditions, it does not necessarily apply to the future from their own experience, the more conditions set by the EA, the more complex the structure, the more serious the case of over-optimization Then again, the preparation of the EA can not leave the historical data, or we become a blind man feeling the elephant, no way to start how to avoid excessive over-optimization? I do this, first prepare a long enough history data, will be divided into two, the first one for observation, the second one try not to look at it, EA out first with the first data for testing, adjustment, until the EA to achieve the target requirements and then use the second history to test it, if the EA performance and the previous data is about the same, then I think the EA basically qualified outside_Hub_Bang

Articles related to this article