Anybody-of-you-doing-meta-analysis-outside-of-Wealth-Lab

QUOTE:
is it possible to realize such an analysis also within Wealth Lab?

To clarify, C# is an NP-complete language. That's computer science talk that it can handle any complexity, but you have to code for it.

I think what you're trying to ask whether there's a built-in WL Position sizer that can scale down the position size based on the most recent trade outcomes? I would take a look at all the existing WL PosSizers to see if one fits this bill. I think there is an Advanced PosSizer that monitors the equity momentum, which "might" interest you, that's part of the WL Power Pack extension. But I haven't used it.

For your first question, "yes" WL can export a trade history into an external program such as Excel. Go to the New Strategy menu and select "Trade History" to get started.

Thank you so much for your answer! I guess you already often looked for criteria outside of the trade signals: External symbols, other markets, certain seasons, all time highs?

Do you think it is worthwhile looking into that to improve the strategy?

Werner

QUOTE:
I guess you already often looked for criteria outside of the trade signals: External symbols, other markets, certain seasons, all time highs? Do you think it is worthwhile looking into that to improve the strategy?

I think it's beneficial to apply as many orthogonal metrics (what you're calling "external ... metrics") to the strategy as possible. For example, we know stocks move with the market (S&P 500), so I decorrelate all my stock prices against that index. That leaves me with a smoother result to analyze further.

If it were a perfect world, you would include every orthogonal metric into an empirical regression model--many terms--and "attempt" to decorrelate as much as possible to get a distilled residual without market noise.

The other point is the regression coefficients (What we call the particular solution.) have high precision (minimal stochastic behavior) because they are based over many bars. Moreover, they are unique to a particular stock. So how they change is insightful.

---
Back to your PosSizer problem, I would search the forum for topics that discuss using advance PosSizers that are based on equity behavior.

Just understand, such PosSizers must make trades anytime the strategy calls for them (even when equity is falling) in order to properly model equity behavior. So you're placing trades with very small position quantity sometimes just to get a reading on the equity behavior--even when the equity behavior isn't ideal. That's how it goes.

Thank you very much for the message, very interesting! I tried to use a "fear" index to decide if trading should be stopped depending on that index. The result was: It is never profitable to stop trading. (At least with the strategy I tested.)

Then, I analyzed if it would be profitable to invest more money (e.g. buying additional CFDs), if this index reaches a certain threshold. And then - of course - the profit was higher because more money got invested.

So, I tried to eek that out by calculating the average invested capital, which resulted in 123% (so, in 23% of cases the invested money get doubled using CFDs), and then in the "normal" version increased the capital for every trade by 23% (every trade 100% money + 23% CFD). So that both alternatives have the same amount of money invested overall.

Using this method I found a threshold which seems to increase the Total profit. But I am not sure, if this is a legitimate calculation, or if it is just cherry picking & curve fitting and will not work in reality. (I analyzed 10 years of data.)