Right now, all strategies published on WealthLab.com are ranked using performance results from a single internal dataset (“WealthLab 100”). This approach ensures a consistent benchmark across strategies, but it doesn’t always reflect how a strategy is meant to perform in its intended market.
I’m considering a change that would allow each published strategy to specify its own dataset for ranking purposes. For example, a crypto strategy could use a crypto dataset, an ETF strategy could use an ETF dataset, and so on.
The benefit would be more accurate and relevant rankings within each strategy’s domain.
The downside is that it could open the door to curve fitting if authors choose datasets that flatter their results. So, we’d likely enforce some manual moderation around dataset selection.
I’d love to get your feedback before moving forward.
- Do you prefer that all strategies continue to use a common benchmark dataset for comparability?
- Or would you rather see rankings reflect each strategy’s own target market and data universe?
- Any ideas on how we could balance fairness and flexibility?
I’m considering a change that would allow each published strategy to specify its own dataset for ranking purposes. For example, a crypto strategy could use a crypto dataset, an ETF strategy could use an ETF dataset, and so on.
The benefit would be more accurate and relevant rankings within each strategy’s domain.
The downside is that it could open the door to curve fitting if authors choose datasets that flatter their results. So, we’d likely enforce some manual moderation around dataset selection.
I’d love to get your feedback before moving forward.
- Do you prefer that all strategies continue to use a common benchmark dataset for comparability?
- Or would you rather see rankings reflect each strategy’s own target market and data universe?
- Any ideas on how we could balance fairness and flexibility?
Rename
QUOTE:
- Do you prefer that all strategies continue to use a common benchmark dataset for comparability?
- Or would you rather see rankings reflect each strategy’s own target market and data universe?
I would do both. You can't compare apples to oranges, but a cherry-picked dataset is a good thing, not a bad thing.
All my datasets are cherry-picked and reevaluated each weekend. Yes, that's cheating, but I'm here to make money, not to play fair. But playing fair for comparison purposes is still useful.
I’m afraid both is not an option, it’s too much maintenance. We want one or the other. Which would you prefer more for the web site strategy rankings?
QUOTE:
I’m afraid both is not an option, it’s too much maintenance.
Naturally, the most powerful solution is intractable from a logistics perspective.
Let's simplify the problem to a compromise to make it more tractable. Rather than having one dataset to rank the strategies by, let's have two instead. One dataset will be for large cap stocks and the other for small cap stocks. Ranking would work exactly as they are now, but there would be two independent parallel rankings for both datasets. Simple is beautiful.
I thought about having three datasets instead, but let's see how the two dataset solution is received first.
Spoiler alert: I think we will find the small cap and large cap datasets will rank about the same. It's not a powerful strategy that determines success, but rather the datasets and market sentiment that sets performance. In some environments, the small cap stocks will outperform and vice versa. (Perhaps someone can prove me wrong.)
So you’re not interested in the idea of the dataset being specific to the strategy? What if someone creates a crypto strategy for example, why should that strategy be run in stocks?
I'd like to see it with a custom list of symbols.
QUOTE:
I'd like to see it with a custom list of symbols.
Then there's no way to compare your results to all the other strategies. The whole point of the rankings is to compare on a level playing field.
QUOTE:
What if someone creates a crypto strategy for example, why should that strategy be run in stocks?
Well, then have two datasets. One for stocks and one for crypto and rank those separately.
Another example is a Strategy like "One Percent a Week." It's built to run on TQQQ and TQQQ alone. It seems disingenuous to throw this into a ranking using a 100 symbol DataSet.
QUOTE:
Another example is a Strategy like "One Percent a Week." It's built to run on TQQQ and TQQQ alone.
Paired strategies require a special ranking system. One size doesn't fit all.
Your Response
Post
Edit Post
Login is required