New Position Priority (for up to 25 position sizes) PosSizer request
Author: Sammy_G
Creation Date: 8/15/2011 9:07 PM
profile picture

Sammy_G

#1
Hi Eugene,

Could you please create a new Position Priority PosSizer that can specify up to 25 different position sizes? It will be useful for testing entire portfolios! In fact, this will open up a new era for WLPro whereby users could publish & discuss the merits of entire portfolios comprised of different asset classes; its a powerful concept.

The current Position Priority PosSizer only allows one to specify up to 6 pos sizes which is too little for a large portfolio. One of the most famous portfolios out there uses more than 6 pos sizes but the community would need the PosSizer first in order to implement it.

Please, Eugene, please please.
profile picture

Eugene

#2
Hi Sammy,

Having spent some considerable time today reviewing this request, my verdict is that the current PosSizer's design isn't suitable for the job (can't extend it to these lengths). Thereby a complete re-design of the code from scratch is required, meaning new architecture, new controls, and somewhat changing business logic. It's too involved now that I've got a queue of projects requested by other customers (open source fundamental provider, market replay tool, IB data provider etc). Sorry.

As a Q&D workaround, how about creating a new Combination strategy, adding your strategy 4(5) times, selecting the Priority PosSizer for each but configuring the priorities differently to cover the range of 24(25) sizes in the following manner:

#1 Child Strategy 1: 0.01-0.05, 0.06-0.1, 0.11-0.14, and so on

...continuing to break up the range of priorities in 25 groups by re-using the PosSizer several times...

#6 Child Strategy 6: ...0.86-0.89, 0.9-0.94, 0.95-1.0

It's only an example but hopefully the idea is clear.
profile picture

Sammy_G

#3
Using combination strategies is a cumbersome way of doing it (besides, I don't even use v6.2 as it killed some of my strategies). Its far simpler to use the PosSizer, and would make it possible for users of all skill levels to implement it.

If you can't do 25 sizes, can you do 20? Even 15?

Edit:
Eugene, please review this post. I floated the idea on 7/1/11; you liked the idea (see your response dt 7/2/11). I thought you would be coming out with the PosSizer in Aug; when you didn't, I had to create this post.
profile picture

Eugene

#4
QUOTE:
If you can't do 25 sizes, can you do 20? Even 15?

Even 15 would be a monster both visually and in code: the existing approach was not meant to be extendible. It must be redesigned from scratch to move forward but I'm not available for this single-user solution, sorry.

My reasonable estimate (and last word) is 10 sizes for the existing PosSizer (hopefully nobody should lose their saved settings).
QUOTE:
I floated the idea on 7/1/11; you liked the idea (see your response dt 7/2/11).

Yes, I really liked your idea of using the PosSizer to assign various amounts/percentages for different ETFs to help the topic starter. And as promised, this use case was reflected on the Wiki help page. But never did I mention that I'm going to expand this PosSizer.
QUOTE:
I thought you would be coming out with the PosSizer in Aug;

In July/August, I've come with two Traders' Tips, a brand new library (MS123 IndexDefinitions), 2 provider fixes, 2 new chart styles, a new performance visualizer, support for Combination strategies in a large extension. Not to mention support, forum and some other activities.
profile picture

Sammy_G

#5
There's no denying you work hard, Eugene.

I disagree its a single-user solution. Its really for the entire community. Anybody with substantial assets at some point or another will need to consider fixed-asset-allocation portfolios (e.f. Fidelity has multiple such portfolios), for atleast part of their funds, if for no other reason than low-stress maintenance. Besides, with the right PosSizer one can easily tweak percentages to one's needs and check out the results.

I am sorry I couldn't make you see the value in it.
profile picture

Eugene

#6
Not quite so. As you could have noticed, position sizing is my passion so I might see some potential value in your idea.

However, the reality comes as a spree of requests and ideas on my to do list. Already filled up 'til the end of the year (let's be optimistic). But there are priorities higher than my community-driven development.

No matter how good the idea could look, 10 is the cap of sizes for the current layout and to make it both look (and code) properly, a significant effort is required. Almost anything can be improved and this process can go on forever (if memory serves, we started with 4 sizes) but currently there's not enough "bang for the buck" in rewriting from scratch what already works for a potential enhancement that may be accepted by a relatively small group of users.

P.S. But 10 sizes there will be.
profile picture

Sammy_G

#7
Do whatever you are comfortable with, Eugene. Thanks.
profile picture

Eugene

#8
Alas, I'm not comfortable with redesigning the PosSizer due to high workload. (Although employing a DataGridView could make unlimited sizes possible and reduce the poor code at least 5x times.)

I'm comfortable with 10 entries (max) in the current design though:

MS123 PosSizer Library v.2011.09
Priority Adjustment