but how is that supposed to work when a system holds a portfolio of stocks with an average holding time of a few weeks?
but how is that supposed to work when a system holds a portfolio of stocks with an average holding time of a few weeks?
I have suggested the fixed size account several times, and others have advocated it too. There are obvious disadvantages to this too, however. I remember from a chat that Matthew explained to me that this was more or less the way C2 was initially set up, but then everybody started complaining because it is not realistic, it doesn’t show money management, etc. Thus it was changed and changing it back would require an major redesign, so obviously he won’t do it.
One conceptual problem with a fixed account size is that the system can loose again and again and the account size will always be reset at 100K. That corresponds to a trader who keeps depositing money to his account even though it is flushed through the toilet every time. So he must have almost infinite financial reserves. That is not realistic.
Your are right, Jules. This was how C2 was initially designed, but we jettisoned the fixed-account size idea after endless user complaint, and hours spent each day explaining why it was necessary.
While I doubt I’ll ever re-implement this feature, I can probably work on something almost as good: how about a dynamic graph that shows what the system equity would look like if each trade had been sized as either (1) a fixed dollar value, or (2) a fixed percent of the account at the time the trade was opened?
This would essentially show the system’s performance without any money management.
The question is how to handle traders who “leg into” trades (that is, keep adding to a position, incrementally). For these kinds of trades, we need to decide whether to set each trade leg at the per-trade dollar amount, or - alternatively - to work backwards so that the largest open position of a legged-in trade equals the per-trade amount. In either case, there will be unfortunate side-effects… but I’m not sure I have a full grasp of what they are yet.
Matthew
Again, I’m not sure how those two options would work when multiple positions are open at the same time. Suppose a system has always traded a max of 4 open positions at a time. It then makes sense to set the “a fixed percent of the account at the time the trade was opened” at 25% of equity. But what then happens as the systems suddenly has 10 open positions at a time. If you stick to 25%, an unrealistic situation would occur as this would normally produce a margin call. If you retrospectively change the fixed percentage from 25 to 10, all previous positions would need to be resized to 10% as well.
I think that the suggestion will overcomplicated current stats that is already complicated. Estimation of portfolio system is complicated. why to not choose simplified/understandable way for majority? (I like it in ADP 
)
Something like new feature: "If I trade the system at…"
Start date/end date
Trading capital
Rules (The rules already exists for autotraders. It’s number of contracts/ percentage per trade and so on)
So a subscriber will define the info and personally receive C2’s simulation of “If I…”.
Ideally full stats as it’s already provided by C2, but with defined by the subscriber limitations of trading. I mean ideally there should be personalized trades history/stats/advanced stats.
Eu
P.S. It’s not even should be done online. I think everybody will accept a delay for calculation of his/her own “If I…” stats.
I like Eu’s idea the most as the guiding principle. That is: consider only the “What if” scenarios that can be set up with autotrading. This answers your questions immediately: Currently the most important parameters are
- scale,
- maximum and minimum dollar size per opening signal, and
- maximum and minimum number of units per opening signal.
So this should also be the parameters of the simulation. We can’t restrict the total position size if there are multiple legs, so there is no need to simulate that. If this produces a margin call, so be it - good for the user to know.
If this is too complex I would suggest:
(1) fixed dollar size per trade OR
(2) fixed number of contracts per trade.
(that is, you program 1 AND 2 but let the user choose which of these options he wants to simulate). These scenarios happen if you set the min and max parameter equal.
In addition I believe that it is also important to let the user specify the initial account size, because rounding of fractions in option 2 can have unexpected effects if you trade futures (I think).
I didn’t really think through this, but I believe there would still be some problems because some advanced statistics assume compounding. But it would definitely make it easier to compare a compounding system with a non-compounding system, and it would also be more relevant for someone who plans to autotrade with fixed sizes.
On a second note, I think it would be a very good idea to let autotraders control the total position size instead of the size per signal, but that is of course a totally different discussion.
Personally, I’d prefer a fixed account size with a profit/loss graph. However, I would much rather keep it as-is instead of providing a complex solution. Complex solutions tend to have complex side-effects…
Hans.
I think generating what-if scenarios might put some burden on the servers. Every time a subscriber enters a new set of parameters, the program would have to recalculate the equity curve.
@Hans:
I’m not sure which of these things you think are too complex, but the simple options (1) and (2) of my post are are not more complex than what Matthew proposed. They are a further specification of his suggestion.
@Science Trader:
It should certainly not interfere with the trading operations, but I agree with Eu that there is no need to do this online
About a year a ago there was a long discussion between Sam and Ross versus me and Pal, and although we disagreed on something, we didn’t disagree that it is good to display some kind of fixed size scenario.
