it is possible to create a strategy without a ranking input,
the ranking is a waste of time, empirically
Every strategy has a fixed number of positions. It’s very hard to get a fixed number of positions without a ranking system. Buy and sell rules aren’t precise enough.
You could create a very simple ranking system for this purpose with only one node, e.g. market cap.
1 Like
Euro, could you explain what you mean by this?
Are you attempting to test factors via screening rules only?
I’m very curious how other people conduct tests and experiments.
Tony
I use this for a No-Rank system for ETFs:
1Yes, it's possible. I've had a live system running for six years with minimal ranking, using 16 buy rules and 3 sell rules. At times, it has held only 2–3 stocks and requires some ramp-up time to reach its full capacity of 15 stocks.
I also developed a similar system that incorporates ranking and consistently holds 30 stocks. While it has slightly lower total returns, it also experiences reduced volatility. For that reason, I believe it is more robust.
Right now i run the systems with 50/50 funding to get a mix of both worlds (higher total return, lower vola)
1 Like
Very interesting. And the results?
That's great. I was interested as I use quite a few Buy Rules in one port which some have said not to do. My belief is that if they are well thought out and not just datamined then it is okay. You seem to have confirmed this.
Nice results. Massive DD in 01/23 tho
Thanks, the drawdowns were the reason to rethink the system and create a ranking based system with more stocks.
This is based on the same principles but often picks other stocks or with a different weight.
And interestingly both combined in a book delivered better returns than each system alone, probably a result of shannon's demon:
The two pics are of course just from sims.
1 Like
Is this one graph live or sim?
the one that i posted 15 days ago is live and unchanged since 2018, the two pics in the last post are from sims.