Real big thank you Denny!
No, I think you are right about a lot of the notes that you have on opportunities to improve the sim. I see some of you guys here with a lot of experience and definitely much to learn from.
I spend quite a lot of time optimizing but will probably only make public things that I consider to be of some value. I don’t want to clog the community search with numerous experiments.
I, since I joined, probably spent more time looking at and experimenting with the p123 and the community systems than my own ones.
This one just showed not bad results. And again one thing that was special about this one is that it did not give in on a lot of optimization. The system and the sim are very close to what I started with.
Let me give some reasoning to things that you (Danny) comment on. You are welcome to let me know what you think. I think we have to keep these forums as much as possible open to any ideas and insights, since this an excellent way for us to improve.
Ranking system performance - I too like the high average gain of the >99 bucket. I would love to see p123 put a distribution bar, as we talked, to see how consistent are those results. I find the difference between the 99 and 98 bucket an advantage because this demonstrates a good discriminating power of the ranking system. I have other systems developed on the strategies of very famous successful investors. One thing I dislike about them is the avg return across buckets 30 to 99 are almost one and the same. I know that the difference between 99-98 can be a problem if the rules for diversification (either because one wants to hold more stocks or wants to cover more industries) force you to buy in lower buckets. Again the distribution graph will help because it will tell us whether there are better performers in the lower buckets and whether it is practically possible to re-shuffle them better by changing the ranking system.
Trading costs parameters in the sim - 1 cent per share is not uncommon. I trade with a broker that charges 1c per share for up to 500 share regardless of the price of the share, (min $1 - that‘s hard to program but you rarely trade less than 100 shares in the sim) and 0.5c per share for above 500 shares - really less than 1c on average. They charge maximum 0.2% of the trade value. So really my sim is set with higher than real commission costs.
I can think of at lest two other brokers who change such or lower commissions including ones that change $0.00 (under certain conditions). I have been trading for the last 12 years and have been extremely sensitive to how much I pay the brokers. Among else I currently have accounts that I trade stocks for $0.00 commission, mutual funds (-all of the types!) for $0.00 commission, options for $1 a contract no minimums (- this was real hard to find), forex - $0.00, futures and futures options - $2.5.
For the slippage I think you are right I have to increase it a bit to be more realistic. My reasoning though was - I want to see whether the system catches well the general movements. Percent here, percent there may eventually average out. I see is, for a system that already shows some promising results I will try to put it under additional stress (reality check). But again the average % return/loss is about 30%/9% so a shift of about 1% should not affect the quality of the system. I agree that it would be important for comparison with other systems. May be, and I though about this before, we should, informally, create a standard “benchmark” condition set so we can compare systems more fairly.
The buy rule for price > 0.1- I have not put that much reason in especially how it will interact with the other rules and whether it is not needed. It came from testing the ranking system. And I want to have it there especially if for some reason decide to remove or change the others that currently replace it. Why I have been happy testing my ranking system with 0.1 price is because it simply does not degrade the results and until we have a tool to see the distribution of the results in the buckets I am not convinced that I should simply avoid these stocks. Well I agree that in our case we have other rules filtering those out.
Stop Loss. I love stop loss orders and I use them exclusively in my real trading. Why I do not believe in them in P123 sims is because first they degraded the performance of this particular sim. Second I do not think (and I need to experiment and discuss with Marco and everyone more knowledgeable here about this) the stock gets any special status after it is kicked out because of a stop loss. I see a window for a stock that has been dropped by a stop-loss rule to be re-purchased again (even if you skip a week) because the ranking may still be high but the downturn may not have ended. I have ways to control this manually and it is fantastic thing one can do to avoid losses. In p123 what can automatically help you achieve that without a “stop loss” per-se is a good technical factor in the ranking system. If the price is hading South, the ranking will automatically kick the stock out and will not allow it to reenter until improves. Or technically based buy rule that will filter plunging stocks.
I like the idea about the laggard sell rule. I was not able to improve the sims performance with it. I am also thinking about a technical factor that eighth help with that.
I hope everyone does not hate these long posts here. But I find this a great opportunity for us to learn.
Thanks again Denny! Great input.
Vlad
