Non-fitting system - main features - page 23
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
The main proof of non-fitting is statistical validity. The more parameters we optimize, the more trades are needed to confirm non-fitting and the higher is the FF of obtained results. Some signs of non-fitting increase statistical validity - the total number of deals that confirm the correctness of the system. It's working on different symbols, the width of the optimal zone etc. Another good sign when the maximum profit is attained at the maximal FP according to the optimization results. In general, the optimization results provide the basis for the analysis of a trading idea as a whole. A lot of runs - a lot of statistical data that can confirm or deny this or that hypothesis. In short, optimization adds the risk of adjustment to the system cost, but at correct analysis of results it can vice versa confirm the robustness of TS using statistics.
Other methods check robustness by equity - basically the uniformity of obtained results. The basis is that statistical characteristics (MO, PF) must be as similar as possible in different equity areas. The minimum value of the plots depends on the system and its characteristics (MO, PF).
What's dark about it. Whoever has more money in the market is right. Any prediction system is only a help, nothing more. And you have to trade with your hands, otherwise you'll get the most out of the market. This is my imho.
2. a TS that follows the market. I.e. the context is traded - movement, sideways and their different types. In this case not a future price is forecasted, but continuation of the detected market condition. Entries and exits are not based on the forecasted price, but on the change of context. Not quite so - the order, of course, at some closely predicted price, but the context rules. ))) // To repeat the hackneyed phrase of some super-duper trader (Williams, I think, but can't remember which one from)))): "I don't try to predict the market - I just follow it.">> I have no authority, but that just accurately captures the essence of my trading.
Afternoon .
The information is secret for sure. Can you give me a direction where to dig?
Good day ...
The information must be classified. >> can you give me a direction where to dig?
Good day...
God be with you! What secret? It's more like top secret number one. The problem is that the people there are very dark - they won't say a word in simplicity. Of course, I don't have statistics, but most of successful trading systems I know from my trading history are based on the principle of following the market. So I don't even know what to answer you...
Other methods check robustness by equity - basically the uniformity of the results obtained. The basis is that the statistical characteristics (MO, PF) should be as similar as possible in different equity plots. The minimum value of the plots depends on the system and its characteristics (MO, FF).
For example, the results of the last N trades of some M trading system as arguments, the function returns a number, if it is higher or equal to 5, such equity is similar to the sample and worthy of working on the real account, if higher, it is cool. If it is less than 5, the system continues trading on paper, waiting for its market phase.If it is possible to use a formula, for example, as an argument, the results of the last N trades in some M trading system, the function returns a number, if it is more or equal to 5 for example, then such equity is very similar to the example and is worthy of working in the real market, if more - then it is cool. If it is less than 5, the system continues trading on paper, waiting for its market phase.
imha, PF is quite suitable as such a formula :)
imha, PF is quite suitable as such a formula :)
Ok, but as an example, we have a number of trades +25, -10, +8, +5, +0, +4, +3, +1. Here PF greater than 1 seems good, but the angle of slope of regression results minus, imho is not good .
And one more question, minimum number of trades on which we give the conclusion.
Good...
God be with you! What secret? It's more like top secret number one. They are very dark there - they won't say a word in simplicity. Of course, I don't have statistics, but most of successful trading systems I know from my trading history are based on the principle of following the market. I don't know what to answer...
If the market is inert, then there is a chance to make a profit within the range of ATR pips. I'm not being sarcastic, very seriously, I need a set of trading systems with their settings, trading principles that can cover the entire history in profit, the only limitation - they work with one TF. The hypothesis is that if history repeats itself the phases of the market alternate, then with a high probability we have TS that makes a profit on this phase. It's like a lucky monkey, if two analysts, one says + and one says -, we have reliable information, add another 10 analysts who give forecasts up to 10 pips, you have all the information on the next figure.Ok, just as an example, we have a number of trades +25, -10, +8, +5, +0, +4, +3, +1. Here PF greater than 1 seems good, but the angle of regression results minus, imho is not good .
Why is it bad? we are interested in the dynamics of equity. build LR to the cumulative amount of this series and it will be up.
And another question, the minimum number of trades for which we give an opinion.
The main condition is again statistical validity. But the system itself may make adjustments to the minimum requirement of trades. For example, ratio of SL to TP. The more this ratio differs from 1, the more trades are necessary. The extreme case is when we trade without stops - any number of trades will not be enough. Although not quite so, because there is a stop loss - a margin call.) In short, the number of deals should be enough to contain both profitable and losing trades for a reliable estimation of their probabilities. If SL/TP=100:1 or 1:100, then 100 trades are obviously not enough to estimate probability tp in first case and sl in second, and it means that other statistical measures of system are unreliable
why is it bad? we are interested in the dynamics of equity. build LR to the cumulative sum of this series and it will be up.
Avals with all due respect, but we are getting close to the formula I am using now, I would really like to hear another solution.
PF is one .
P.S. If you are interested I can send it to you in person but I will send it later.