- What is so dangerous about over-optimization?
• That the more complex your strategy is, the less likely it is to perform well under real market conditions. - How long should a testing period be if you are serious about building a profitable trading strategy?
• 9 to 11 year so you make sure it works well under different market conditions. - Why should you avoid asymmetric trading signals?
• Because the market can be affected by economic variables that change due to economic cycles (interest rates, inflation, etc). Adding separate criteria for longs and shorts automatically increases the strategy’s degrees of freedom and makes it excessively prone to curve-fitted solutions.
1 Like
@filip
1.Over-optimization could end up in curve-fitting, afew factors are involved to avoid it.
2.Minimum 5 years up to 20 years depending how many trades and their market history. Simulations are a most.
3.Because we want to increases the strategy’s degrees of freedom and makes it excessively prone to curve-fitted solutions.
Mar '19
In this reading assignment we will dive deeper into the most important issue when it comes to building a profitable trading strategy - Backtesting. We will look at some of the common mistakes, over-optimization and curve-fitting and how you can avoid them. Read through this blog post and answer the following questions in this forum thread. Use the knowledge you have learned so far as well.
What is so dangerous about over-optimization?
How long should a testing period be if you are serious about building a profitable trading strategy?
Why should you avoid asymmetric trading signals?
1 Like
- What is so dangerous about over-optimization?
If you doing over-optimization it is possible that you instead of optimize you are actually curve-fitting. So you do your best to get as good results as possible that just is an optimization of that time of the history you are studying. So when other history data is executed the idea doesn’t work as it should. It doesn’t work following the strategy, instead it works in the special condition that was in the situation you previously had when you were over-optimizing. - How long should a testing period be if you are serious about building a profitable trading strategy?
Ideally 9-11 years in order to ensure that a large amount of market condition become available. - Why should you avoid asymmetric trading signals?
Adding separate criteria for longs and shorts automatically increases the strategy’s degrees of freedom and makes it excessively prone to curve-fitted solutions.
1 Like
- Focusing too much on the optimization might lead to curve fitting.
- Preferably around 10 years or at least some period filled with multiple different economical and social factors.
- Adding separate criteria for longs and shorts trades automatically increases the strategy’s degrees of freedom and makes it excessively prone to curve-fitted solutions.
1 Like
*What is so dangerous about over-optimization?
Curve-fitting: this is because the function may not produce the same result when put in practice/trading. The function could have no predictive capability even though it was able to predict all previous items successfully.
*How long should a testing period be if you are serious about building a profitable trading strategy?
A period of 9 - 11 years of data. The longer the trading period, the more statistically significant the data set is and the less likely it is to allow the curve fitting of your system.
*Why should you avoid asymmetric trading signals?
Symmetric trading signals are excessively prone to curve-fitted solutions.
1 Like
- By over optimizing we are might miss opportunities because our trade is too detailed and specific.
- 9-11 year according to this article.
- Economic factors might have changed so you will get rekt.
1 Like
- Over-optimization can lead to curve-fitting.
- 9-11 years of testing are required for serious attempts at profitable trading strategies.
- Up and down trends cannot be guaranteed to continue.
1 Like
-
It can easily come to curve-fitting, which means a perfect strategy that fits the old specific conditions, that in future doesn’t going to be the same.
-
9 to 11 years of data should be used with timeframes over 30minutes
-
prone to curve-fitting through more complexity, keep it better simple.
1 Like
[quote=“filip, post:1, topic:7688”]
-
What is so dangerous about over-optimization?
OO can lead to curve fitting, which can limit the effectiveness of a strategy when applied to different date sets, time frames etc. -
How long should a testing period be if you are serious about building a profitable trading strategy?
This article suggests upwards of a decade.
- Why should you avoid asymmetric trading signals?
It may be prove to curve fitting as the asymmetry may work for particular market cycle with which it is applied yet not as effective when applied on different market cycle or asset.
1 Like
- Curve fitting
2.min 30 min ideally 9-11 years - Can lead to curve fitting.
1 Like
-
What is so dangerous about over-optimization?
It can lead to curve fitting. Meaning you have tweaked the strategy to work on historical data really well… but it is now too specific to work on either another historical data set or when implemented on live data. -
How long should a testing period be if you are serious about building a profitable trading strategy?
Testing should be carried out on 9-11 years of data and on time frames of 30 minutes or greater. This amount of data is not available yet for BTC, ETH and alt coins, you just need to work with the largest possible dataset you can. -
Why should you avoid asymmetric trading signals?
Asymmetric trading signals (where the entry and exit criteria for longs and shorts are different) makes the strategy excessively prone to curve fitting. It is increasing the strategy’s degrees of freedom because there are considerations on the historical data at will be different due to the economic cycles (such as interest rates differentials).
1 Like
-
There is a danger of over-optimisation is to produce a trading strategy with best results that will not be achievable going forward.
-
Time frames should be greater than 30 minutes and around 10 years of data should be used for the process in order to ensure that a large amount of market conditions become available.
-
More probability of making a mistake. And more space for over-optimizing due to “more parameters to adjust”.
1 Like
1. What is so dangerous about over-optimization?
Can result to curve-fitting strategies.
2. How long should a testing period be if you are serious about building a profitable trading strategy?
To build a profitable trading strategy long testing periods are required example: 9-11 years of data should be used, time frame longer than 30min , and profit and stop loss targets should be set above 10 times the spread.
3. Why should you avoid asymmetric trading signals?
Asymmetric trading signals increases the strategy’s degrees of freedom to artificially “fit” different market conditions and makes it excessively prone to curve-fitted solutions.
1 Like
- What is so dangerous about over-optimisation?
Over-optimisation is dangerous as it runs into curve-fitting, which conforms to past data but forgoing reliable predictive capabilities.
- How long should a testing period be if you are serious about building a profitable trading strategy?
A testing period of nine, ten or eleven years is reasonable to build a profitable trading strategy because a longer time horizon reduces the probability of curve-fitting.
- Why should you avoid asymmetric trading signals?
Asymmetric trading signals increase complexity, degrees of freedom and lead to curve-fitting.
1 Like
- over-optimization can lead to curve-fitting.
- Around 10 years of data for building a profitable trading strategy.
- They can increase complexity in the trading strategy and can lead into curve-fitting.
1 Like
curve fitting
About 10 years
Adding separate criteria for longs and shorts automatically increases the strategy’s degrees of freedom and makes it excessively prone to curve-fitted solutions.
1 Like
- Over optimization can lead to “curve-fitting”, which is the unwanted tuning of the strategy to fit specific past data.
- According to the article, approx 10yrs on higher time frames for accuracy.
- Asymmetric signals provide more complexity which tends to lean towards tuning the strategy for specific datasets… curve-fitting.
1 Like
The most dangerous thing about over-optimization is curve fitting.
9 to 11 years testing data.
Asymmetric signals should be avoided because it allows for cure fitting. If you adjust variables too much in the past other variables in the future will be missed.
1 Like
- What is so dangerous about over-optimization?
If you over-optimze your model, you are in danger of curve-fitting, and when that happens it is like you try to adapt your model perfectly to the past price data so much, that it will be impossible that the exact same thing happens in the future , and so the model wont be valid. - How long should a testing period be if you are serious about building a profitable trading strategy?
in the article the guy mentions 9 to 11 years of data, but I guess that is for the oldest instruments, because how do you build trading strategies when a company is new and you want to trade their stock? or how about ethereum thatś only been around like for 7 years or so? I guess the more time you have for back testing, that will make the model more reliable for backtesting. I guess that for newer instruments, the fundamental analysis will take more relevance. - Why should you avoid asymmetric trading signals?
dding separate criteria for longs and shorts automatically increases the strategy’s degrees of freedom and makes it excessively prone to curve-fitted solutions.
by the way … I think the article is a little outdated 
1 Like
-
You can end up with a trade strategy that is curve fitted to past trends.
-
A period of 10 years is to be used to avoid ending up with an overoptimized trade strategy with an additional year or 2 of out-sample testing.
-
Overcomplexity in trade strategy can lead to curve fitted solutions.
1 Like