Dear Shaggs,
fully agree on multiple points:
first: I never said one should extend JBT in that way - I only said,
it would be nice to...
- while I am afraid this goes so far beyond the current design, that
it would actually be a different project.
second: of course backtesting is a major issue. This also implies that
portfolio strategies must be there and
these must be backtested as well.
In the meanwhile I thought about it more and I would say, if (given
the above points) a portfolio extension
would be integrated, then we can differentiate two situations.
A) Multiple strategies on the same instrument:
We can always replace that by making the decision logic of the
various strategies by indicators and then
define a single strategy that integrates the values and does the
trading.
Disadvantage: this is less structured, which is a pity.
Position management, money at risk, etc. are different things from
generating entry and exit
signals. There are good reasons for conceptually separating this.
Advantage: every tool in JBT can be applied on this, if we simply
integrate PM into the
strategy itself.
B) Multiple strategies on different instrument:
Here, I actually do not see a way how this can be done well with
JBT.
This is a pity, but of course having a working single instrument
strategy is the first thing :)
Portfolio theory is a major aspect of sound trading..
However, I believe, thinking about portfolio theory and the relation
to
JBT has intrinsic value. It helps to emphasize the notion of risk and
how to address
this. So, perhaps a next step could be to develop volatility
indicators as a step to this.
Also other aspects like trading hours, and in general position
management are conceptually
different things and thinking deeply about it, also helps strategy
development..
Cheers
Klaus
On 5 Jun., 03:24, ShaggsTheStud <[email protected]> wrote:
> > the idea that is coming up here, seems to be very interesting: having
> > portfolios from multiple instruments.
> > I believe this would be a great addition.
> > Let's suppose you have to strategies on different instruments:
> > but you want to trade only one: if there would be portfolio management
> > in jbooktrader, you
> > could take the first signal and ignore the second by the other
> > strategy, if it arrives, while you are in
> > the market. - Or if we add some sort of reliability indicator to the
> > strategies, you could allocate money to the
> > different strategies based on the expected quality of the signals.
> > This would be very interesting.
>
> > Also one could have multiple strategies on the same asset. The
> > portfolio could trade a multiple of those, but
> > at most 3 (e.g.,). So, a portfolio could add across strategies money
> > and risk management. Something which
> > can at this point not be done with jbooktrader.
> > But I am afraid, this would mean a VERY significant extension and
> > modification of the software...
>
> I personally find the ideas both fascinating and terrible, and there is a
> very logical reason why: When you design a strategy in JBT, you test it
> with whatever position sizing strategy you build into the strategy. The
> back test results assume that you trade the position rule VERY STRICTLY.
> When you deviate from this rule, you are deviating from the strategy, and
> the backtest results are useless.
>
> For example, let us assume you are trading 1 strategy on, say, currency,
> such as EUR/USD. Let's assume you are trading a second strategy with ES as
> the instrument. What you are proposing, then, is that when the ES strategy
> wants to set a position, you ignore the EUR/USD strategy. You might try
> this and find that it works extremely well. You might also try this and
> find it disastrous, because the markets have some kind of interlinkage where
> the signal you got on ES is bad, and its one of your loser trades, but
> meanwhile the ES strategy would start winning big. Many strategies tend to
> have a lot of small losing trades, and big winning trades - and you really
> don't want to miss those winning traders.
>
> Basically you want to trade a strategy as tested, always, including position
> sizing, and position timing. You cannot whimsically pick and chose when to
> trade it - or how big to size your position. If you define a rule in your
> testing, you need to stick to that rule.
>
> To properly do what you propose, you need to build a multi-instrument
> strategy (JBT is NOT designed to do that) and test it - one that has the
> full relative position sizing rules built into it.
>
> I personally think you guys are getting way ahead of yourselves. Start by
> designing strategies that work on ES, and ES only, and plan on trading one
> strategy. If you get that far, you are doing really well, and then you
> should figure out what the next step for the platform is, code it, and
> donate the code back. If you can't get past that first step, figure out why
> not - maybe we need to make the tools we have better before we can move to
> new fancy features. If we can't chop down a tree, why are we worrying
> expanding our logging truck fleet?
--
You received this message because you are subscribed to the Google Groups
"JBookTrader" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/jbooktrader?hl=en.
