Group Collaboration System Ideas

[SaidBitar]

is this the one you are talking about ?

[ScottMcNab]

Darryl Vink wrote:

Julian Cohen wrote:

What’s your problem with position sizing Darryl?

just experimenting with something like this:

Code:

formula = BuyPrice/ATR(10);
SetPositionSize(formula,spsPercentOfEquity);

I’ve been wondering about changing position size too Darryl so that I would place larger trades if I were able to identify times where probability of success increased. Lot of “ifs” but for example…in buysetup may add a condition where if NDX and SPX both below 5 day ema or both had dropped more than 3% from highest close in last 5 days or one index may be up from prev close and one may be down etc etc….if I was able to identify a consistent time where probability of win or size of win increased then on these days I would trade on margin or increased margin (eg instead of 25 positions at 5% may go to 25 at 8%)

Am I straying into curve fitting territory again or simply refining a strategy ? I suspect the former but I keep hearing on podcasts that its important to increase size when probabilities increased !

[ChrisVirides]

I do not think its curve fitting, but until you test it its just an untested hypothesis

[Nick Radge]

Agree with Paul. It intuitively makes sense although the challenge will be detecting these ‘so called’ higher probability times.

Good luck with that…. :woohoo:

[Anonymous]

Scott McNab wrote:

… but I keep hearing on podcasts that its important to increase size when probabilities increased !

i took up howard bandy’s offer of a free book so im looking forward to seeing what insights he has about it

[JulianCohen]

Darryl Vink wrote:

Scott McNab wrote:

… but I keep hearing on podcasts that its important to increase size when probabilities increased !

i took up howard bandy’s offer of a free book so im looking forward to seeing what insights he has about it

Me too. I’m amazed he was happy to ship it free too.

[TrentRothall]

Darryl Vink wrote:

Scott McNab wrote:

… but I keep hearing on podcasts that its important to increase size when probabilities increased !

i took up howard bandy’s offer of a free book so im looking forward to seeing what insights he has about it

Let us know how it is!

[Anonymous]

In an attempt to try something slightly different or go down another rabbit hole I came up with the following and would appreciate some feedback on whether this is a case of curve fitting or not.

I was trying to get away from selection bias. The system has simple rules that use 2 MA’s and ATR for stretch so nothing complicated. I initially set some parameters and tested it on 20 ETF’s, major indices of the world markets. Because the universe is so small and the trade frequency is low I used max 10 positions at 20% of equity using 50% margin. This first attempt gave me decent results. (I used the ETF’s because I knew I would be limited on the universe, so that’s why I started there)

I then tried this across different universes but could not get the results close to the tests on the ETF’s.

The reason it works is that the correlation between ETF’s in the universe is high so when one makes money they all tend to make money and the win % is high – 70%. When there are losing trades no surprise they move together, but the exit means the losses are manageable. So there are many winning streaks and then some losing streaks, the DD is around 20% and the CAGR is similar around 22-24%. Average trade duration is about a week.

The reason that it makes money is also the reason that it does not work as well when tested on other universes, so its strength on the set of ETF’s becomes it’s weakness across a broader universe.

I optimised and stress tested the system and the original parameters I used were the values that were the most robust so no problem there, the stress testing did not show any adverse results.

Have I found something that I would be comfortable trading on an initial attempt and the fact that it did not hold true across different universes make it curve fitting that I inadvertently stumbled across?

Thoughts.

[Nick Radge]

What kinds of ETFs?

[ScottMcNab]

Hi Miguel,
I will be interested in the response of Nick and other more experienced members of the forum to your post. The only thing I can contribute that may be of interest is the following: I am trading ASX on 10 positions at 10% and that can lead to some big swings …other system I trade is in US with 25 positions at 8% which seems to have fewer adrenaline jolts….when I read 10 positions at 20% my heart rate went up a bit…but of course everyone has different comfort levels but it is something to consider.

[LeeDanello]

Julian Cohen wrote:

Darryl Vink wrote:

Scott McNab wrote:

… but I keep hearing on podcasts that its important to increase size when probabilities increased !

i took up howard bandy’s offer of a free book so im looking forward to seeing what insights he has about it

Me too. I’m amazed he was happy to ship it free too.

Tell me Julian what is the book like?

I read his Mean Reversion book and found it a bit esoteric. I didn’t find it practical enough. I concluded that at the end of the day that I was looking too deeply into a shallow pool. I think I’ll go back and read Holy Grails again.

[Anonymous]

What kinds of ETFs?
Nick I added 2 & 3 X leveraged Bull and Bear ETF’s on the main US Indices and some international markets.

SSO – 2 X S&P
UPRO – 3 X S&P
SDS – 2 X Inverse S&P
SPXU – 3 X Inverse S&P
DXD – 2 X Inverse Dow
SDOW – 3 X Inverse Dow etc. etc
QLD
TQQQ
QID
SQQQ
TNA
TWM
TZA
EEM – Emerging markets
EFA – Foreign large cap
EPP – Pacific ex Japan
IEV – Europe
ILF – Latin America

18 in total not 20 as I previously said.

Scott – you are 100% correct, but there are not many items in the universe so getting a full allocation does not happen often, it does happen but 6-7 is more common (that is by doing a quick manual scan, I need to figure out the code to do that for me). The long MA acts as a stop so there is some safety although anything can happen as we have seen. The universe also has bull and bear ETF’s lowering the number that will be traded out of the universe at any time, 10 Bull ETF’s and 8 Bear so to get a full allocation of 10, all 10 would have to align at the same time so the risk is not as extreme as it may sound but again it can happen. I have also not tested further by trying a safety stop before the long MA exit, that may reduce DD, but I have not tried different things unless there is some merit in exploring this further.

[Author]

Posts