1.What is so dangerous about over-optimization?
Curve fitting, unreliable simulations (under 30 mins time frames).
2.How long should a testing period be if you are serious about building a profitable trading strategy?
Ideally 9-11 years. But 2 years at least.
3.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.
What is so dangerous about over-optimization?
The strategy might look like it is working perfectly when you are testing on historic data, but might in fact be curve fitted and not work well at all in the future.
How long should a testing period be if you are serious about building a profitable trading strategy?
9-11 years of data, in order to ensure that a large enough amount of market conditions become available.
Why should you avoid asymmetric trading signals?
It increases the strategy’s degrees of freedom and makes it excessively prone to curve-fitted solutions.
Obviously avoiding curve fitting should be a very important part of any system developer’s efforts as we don’t want to generate trading strategies with absolutely astonishing results that will not be achievable going forward. Since our goal is to produce systems that achieve good performance in the past with the highest possible guarantee that that performance will be repeated in the future it becomes vital to take steps in order to ensure that optimization does not deliver curve-fitted strategies.
Optimizations should be carried out for long periods of time, ideally 9-11 years of data should be used for the process in order to ensure that a large amount of market conditions become available.
Asymmetric trading signals works pretty similar to curve fitting as it might usually does not work over a long time past trading data.
What is so dangerous about over-optimization?
Its dangerous because we can end up curve fitting our system. tighter= we optimized our system the more curve fitting.
How long should a testing period be if you are serious about building a profitable trading strategy?
The longer the trading period the better. Which mean 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?
Tends to tuning the data for more specific strategy which can lead to curve fitting.
What is so dangerous about over-optimization?
It can cause curve-fitting, over complex solutiuons or useless solutions that cannot really be applied.
How long should a testing period be if you are serious about building a profitable trading strategy?
9-10 years and for charts that depict longer periods than 30mins.
Why should you avoid asymmetric trading signals?
It can cause curve-fitting biased towards the upwards or downward trends, therefore not showing objective results.
Common Back-testing Mistakes – Questions
1. What is so dangerous about over-optimization?
Over-optimization can lead to curve-fitting, which may result in overly-complex solutions that may be useless when applied to current data. When trying to perfectly fit past data while selectively picking exceptions, coders nullify a trading strategy’s ability to successfully trade any future data points.
2. How long should a testing period be if you are serious about building a profitable trading strategy?
At least over 9-11 years with timeframes no shorter than 30 mins.
3. 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. » --> reason why it’s best to rely on symmetric trading signals.
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?
if you over-optimize your algo, then it’s prone to failure under any other (sufficiently --dependent on HOW over-fitted-- disparate) conditions
~10 years worht of data (HOW FINE-GRAINED?? …idk…)
to reduce complexity but also because ‘stonks only go up’ depends on so many macroeconomic variables that they’re basically outside our time-frame of interest (which is more short- to mid-term, not 30 years…right?)
Over-optimization makes it more likely that you are building a strategy that will work splendidly for a scenario that may never play out again, or less often than you expecting.
Ideally 9-11 years.
Asymmetry is a sign that you may be giving place to over-optimization, instead of treating longs and shorts as inverse opportunities.
By over optimizing, we might be generating trading strategies that have great results on past data but would not apply on future cases.
Ideally 9 to 11 years of data
Because it might lead to over-optimization of past data
Because you don’t want to generate trading strategies with fantastic results that will not be achievable going forward
9-11 years if possible.
Because this will increase the number of degrees of freedom in the strategy and make it excessively prone to curve fitting.
Over-optimization means curve fitting after curve fitting is introduced. This increases the possibility of fully correlated optimization. It is dangerous.
Meaningful or profitable optimization should be equal or more than 5 years because smaller periods of time are not statistically relevant to the long term changes in the market.
Adding asymmetric criteria for longs and shorts automatically increases the strategy’s degrees of freedom and makes it excessively prone to curve-fitted solutions.
Over-optimization optimises for nuances that only exist in the nuances of price action within given periods, and these nuances are rarely based on anything more than happenstance.
The nuances have no meaning, and are therefore very unlikely to happen again moving forward.
You could consider these intricasies as noise.
A profitable algorithm will be optimised around the signal, with disregard to noise as much as possible.
While all trading patterns contain both signal and noise, it’s useful to back-test to optimise for a strategy that works over time, without too much focus on the curves that the trader is back-testing their algorithm on.
It is wise to take into account large data-sets with cycles, and multiple economic states. Therefore it’s useful to back-test on 9 - 11 years of data.
Asymmetric trading signals means that there has been some amount of curve-fitting, as the trader has adjusted their algorithm to account for the nuances that occur based on economic sentiment and noise in the market, or on a broader economic scale.
Trading strategies should be symmetrical in both regards, as this broad-brush approach will work to your favour in any state of the market, whereas if you’re working with asymmetric trading signals, you’re optimising for one type of market. And if the market moving forward is different to what you’ve curve-fitted to, you’re gonna get rekt. 
It could result in Curve Fitting
9-11years
Because the market conditions could change compared to your testing period.