Home › Forums › Trading System Mentor Course Community › Progress Journal › Nick Radge – Daily Journal
- This topic has 543 replies, 32 voices, and was last updated 3 weeks, 6 days ago by Nick Radge.
-
AuthorPosts
-
March 24, 2020 at 12:27 am #111161Nick RadgeKeymaster
Just posted this on my Twitter feed:
Great to see people taking the time in this environment to get up to speed on things and do some extra learning.
Here’s a few things I’m doing:
– finishing up a new book. Don’t ask about the topic. It will be a hard cover and available in the coming month.
– continuing research on a market neutral trend following strategy. I’ve never seen this style anywhere, so I’m either barking up the wrong tree, missing something major, or very smart. I highly doubt it’s the latter.
– studying the impact of ‘signal luck’ within my momo strategies and researching how to lower that. I’ve always considered diversifying across strategies, but I think some signal diversification is required.
– reallocating some funds across my various strategies. I’ve mentioned before I was weighted 70% to trend and 30% to mean reversion and wanted to balance that out. The day trade strategies are proving their value in this environment so more to be allocated there.
– keeping clients informed on my thoughts and potential opportunities
March 24, 2020 at 5:07 am #111168JulianCohenParticipantBy Market Neutral do you mean not relying on an Index Filter to switch it on and off?
March 24, 2020 at 6:55 am #111169MichaelRodwellMemberPairs? Long/Short?
Thanks Nick.
Agree with your suggestions. Still finding I need some distance from the market with everything unfolding. I guess with time and experience this will get easier emotionally. Even know I am in cash Im still finding the whole situation a bit distressing.
March 24, 2020 at 7:30 am #111170Nick RadgeKeymasterMarket neutral is not worrying about broader market direction, rather the ‘spread’ between two stocks.
Very early days with this one.
Mike,
I’ll be doing an article on ‘signal luck’ this Friday in the newsletter. Something you may wish to consider – as I am.March 24, 2020 at 9:40 pm #111171AnonymousInactiveHey Nick anything much to report on how your DMK9 strategy is going through recent conditions?
March 25, 2020 at 12:26 am #111172Nick RadgeKeymasterMatt,
I switched that off as the trade frequency really wasn’t high enough and I wanted less o/night exposure. My focus has been on a day trade system that’s not correlated too much with what I had been using.Needless to say that unless you changed down the exposure and leverage the system would’ve been smacked down these last two months.
March 25, 2020 at 4:53 am #111173AnonymousInactiveThat’s a shame to hear.
Yes, anything seemingly over exposed/leveraged in last month or so has been monumentally horrible.
March 25, 2020 at 11:29 pm #111174Nick RadgeKeymasterI’m not fussed about it. It wasn’t a big CAGR. I just started using it thinking it was adding some return without much effort…I think that was a lazy thought process
March 27, 2020 at 2:24 am #111175Nick RadgeKeymasterFor those of you that read my newsletter article today, “Diluting Signal Luck”, here is the reply I rec’d from Cory…
Forwarded Message
Subject: Re: Diluting Signal Luck
Date: Fri, 27 Mar 2020 00:39:00 +0000
From: Corey Hoffstein
To: [email protected]Thanks for citing my work here!
I’d suggest you missed an easier solution, though.
You can continue to do monthly (4-week) signals; just divide your portfolio into 4 equal sub-portfolios and trade each of those sub-portfolios at the end of a different week!
Best,
CMarch 27, 2020 at 2:56 am #111179MichaelRodwellMemberThanks Nick. When I went back and looked at my regime filter I noticed a few near misses. Most of the time in my favour but as you know the most recent and worse one went against me.
For testing, would this approach work…
– Assume 2 rotations per month rather than weekly or one per month
– Run 2 individual back tests, 1 rotating on the first of the month and the second rotating on the 15th of the month
– Export data to excel to plot and analyse resultsThanks!
March 27, 2020 at 3:29 am #111181ScottMcNabParticipantBe interested to hear results from others (slow day at work here as work continues to dry up)….my coding always suspect (used barssince in code which may be issue) but when I divide into the 4 tranches and rotate on 1st day of week (for each of the 4 weeks) I don’t see any improvement (reduction in maxDD) compared with the single EOM rotation (tested SPX 1960 – current)
Edit: realised I need to look closer…max DD may be same over test period but if occur at different times then overall maxDD will be improved of course
March 27, 2020 at 4:17 am #111182Nick RadgeKeymasterScott,
It’s not supposed to make it better. It’s supposed to reduce the risk being unlucky with a single rotation date. Consider the black swan that was never in a backtest which is always a possibility. Even with 20+ years of data, the sample of signals is still quite small.March 27, 2020 at 5:23 am #111184Stephen JamesMemberScott,
Its not that your coding is necessarily suspect, sometimes functions do not behave in a way we might expect, or are not designed for a particular use.
If there is a potential issue with a function, or using one for the first time in a particular context, a good way to check is to code the same rule in a different manner, if at all possible.
I used both BarsSince and Sum for the same thing, yet got different different results so started digging. It was used in similar context to yours but I had the advantage of chart signals to look at (i.e. not rotational).
I’ve used BarsSince in other contexts without an issue so it may be just a case as not suitable for a particular scenario.
However, if you test with both functions and the results are the same there is likely no problem.
March 28, 2020 at 3:21 am #111185TimothyStricklandMemberNick,
That was how I understood it. Overall the monthly momo would essentially get a bit higher CAGR over the longer term but splitting them up in weekly momo reduces some of the signal luck offering a smoother equity curve, at the expense a few % points in CAGR? In times like these, that sounds like a fair trade-off to me.
So what’s suggested here is to run 4 seperate monthly rotational systems essentially ending on different parts of the month, I wonder how hard that would be to code in.
March 28, 2020 at 3:49 am #111186Nick RadgeKeymasterI wouldn’t bother coding it in. The result will more or less be the same.
What it WILL reduce is if there was an extreme drawdown caused by a single rotation, so for example, Oct 2018 may have been reduced somewhat.
What’s he’s suggesting is trade the same system, just do it in 4 tranches; week 1, week 2, week 3 and week 4. It will mean you slide into positions and slide out of positions.
-
AuthorPosts
- You must be logged in to reply to this topic.