Klaus,
There is also a historical context here. Gen 1 autotrading worked pretty much as you described: Enter a trade when the signal is given, not otherwise. This caused various problems. One problem was that a limit order can be filled for some but not all subscribers. Matthew and Francis made several improvements to this, like the feature to convert limit orders to market orders. This again caused other problems, like the possibility that the converted order is sent while the original limit order is being filled just after the moment that the cancellation signal was sent and just before the moment that the cancellation signal was received by the broker - leaving the subscriber with a double position. Each improvement seemed to cause a new problem. After some years of frustration subscribers argued strongly in favor of an auto-sync feature that guarantees that the subscribers account will always have the same positions as the systems account at C2. The idea was that one should not track the mutations, but rather the total outcome of all previous mutations (i.e. the position sizes). This is implemented in Gen 2 and Gen 3 autotrading.
Judging from the number of forum posts I believe that the problems with autotrading have been greatly reduced since auto-sync has been implemented rigorously. This is why I and others will not easily agree when you suggest something that violates the basic principle of auto-synchronization.
My following point will probably be less convincing. If a vendor keeps a profitable trade open, we must assume that he believes that the trade will become even more profitable. That is, the vendor must believe that the trade has still a positive expected value in comparison to the present value. Otherwise he would close it right now. If you do not want to enter this trade, then you basically say that you don’t trust this judgement of the vendor. Why do you subscribe to a system if you don’t trust the vendor’s latest judgement? Probably you have some trading philosophy for that, but it is not the task of C2 to make this judgement for you.
I have no objection against an optional feature like the one you and John ask. But given the basic principle of auto-sync I can imagine that it is hard to implement without opening opening another can of worms.
Dear Jules,
first of all: yes I know the whole topic is not easy and that there has been put a serious amount of effort into it already. I also track the general discussions here on C2 a lot, so I am fully aware of these problems and issues.
Regarding the first point: you are right there is a possibility of the signal conversion, which you have to deal with also if you do full auto-sync.
(There you deal with this, e.g, by immediately selling again.)
However, the two things have actually nothing to do with each other.
One dimension is whether you only look at the status or whether you look at the signals, while the other is whether you look at the whole portfolio or only the part opened since you entered.
So, you can do auto-sync, but only for the portion that became relevant since one subscribed. (perhaps this formulation is clearer)
Thus, I would argue that your argument misses the point here.
Regarding the second argument, sorry, this seems to me very unconvincing (so you are right).
If we talk about mechanical systems, they are optimized to provide good results when you enter when they provide a signal and exit when the signal is given. The same holds for all kinds of indicator-based trading, you simply cannot from a fundamental point of view make a reasonable assumption what happens if you enter late.
For example one of the good systems currently here is "Gen 2 - Mini S&P" it opens and closes at certain times, explicitly there are NO assumptions made, about entering in between (Kevin, I hope I am not misrepresenting you.)
Your argument might be valid if a person gives all the signals and does not use any kind of statistical tools and optimization… - and these are certainly systems I am avoiding here completely 
Also from a strategic point of view: every trader (should) take into account risk and return of a position, so he buys the potential for higher return with the risk for drawdown (as going higher is never sure). However, this equation is completely different as soon as you enter late. Thus, what is a completely rationale assessment for the provider is not an appropriate assessment for a subscriber entering late…
Jules, we do certainly have common ground here regarding such a feature being optional
Or to go even beyond (but this would make it it even more complex): to have the possibility to select the instruments to synchronize while showing the current (this would have to be real-time) P/L per instrument, then I can deselect everything with a positive P/L 
But I do not want to make it to complex for Matt…
Cheers
Klaus
I wasn’t sure if you were aware of these previous discussions, this is why I mentioned them.
Your example of Gen 2 - Mini S&P convinced me that at least in some cases your suggestion is logical indeed.
Correct me if I am wrong, but is it not already possible to select instruments that should be discarded in synchronization?
Jules,
That was a gen2 feature that has been disabled because too few people used it and Autotrading is optimized for hands-free trading. Sync is by far the most complex part of Autotrading, and the laws of causality apply in spades: any new feature of tweak has a consequence and it’s not always possible to foresee what that will be. Some features are not worth the risk if they are used by only 1% of autotraders.
Not synchronizing open positions when a system is being configured for autotrading is something that has been on our to-do list for many months. We’ll get to it eventually, hopefully soon.
Francis
This issue just won’t go away…
Step back a moment and look at all the time and effort devoted to (1) discussing Auto-trade functionality, (2) Auto-trade system design, (3) Auto-trade syncronization/integration issues, (4) education regarding Auto-trade for non-professional Users, (5) casual users and system developers not grasping what auto-trade does and how it does it.
The amount of time, effort, errors, complaints on this site is HUGE. So - whether it’s a system design flaw, a bug, user education - it DOESN’T matter - it’s TOO much. IT DOESN’T MATTER IF IT IS DESIGNED AND PROGRAMMED CORRECTLY - IF IT DOESN’T PRODUCE THE DESIRED RESULT FOR END USERS - IT FAILS. IF IT IS TOO COMPLICATED FOR END-USERS TO PROPERLY USE - IT FAILS.
Software isn’t supposed to be this difficult to USE. So - as a professional software designer/manager - I have to say it currently FAILS. Sorry. I applaud the efforts the designer went to to design it well and implement it in a fashion that casual users can successfully use it in, but it’s still NOT performing at the bottom line OFTEN ENOUGH for Production and wide-spread use.
Designing auto-trade software that simultaneously can accurately and repeatedly handle a myriad of equities, trading markets, software platforms, etc. IS A COMPLEX UNDERTAKING. This version of auto-trade is good - but eventually, not good enough. Hence the HUGE amount of time spent on this on this site. This is like a car that works, but doesn’t work well enough often enough. Ready for PRIME-TIME? Not yet. Should you trust it? Beware if you do…
Sorry - the proof is right in front of us - all over this site. Keep working on it, Matthew. When it is ready for Prime-Time, then more of us will use it. We’ll know when it is ready for Prime-Time when the number of complaints, integration issues, syncronization issues and the like significantly drop from these forums. Period.
If a start-up software company came to me for strategic business advice and showed me this version of Auto-Trade and wanted to bring it to market in its current state, after seeing this many problems in Alpha and Beta testing, I could not recommend that they bring it to market right now. Too many issues.
So - if you “sign over” control of your hard-earned money to a product in this shape, be forewarned and take personal responsibility. There is plenty of evidence of numerous complaints/problems scattered all over this site NOT to understand that Auto-Trade can create real-world money problems for you.
Someday, it will probably be capable - but not right now.
My 2 cents.
Skip
Any person who turns on autotrading, and is shocked at the losses incurred on day 1 is already doomed to failure.
The software works as intended, and as it should: The client equity curve will match the C2 equity curve (with commissions / slippage). That is the intended function of this entire website. Everything else is splitting hairs.
If a subscriber thinks a vendor has a huge market edge on the actual entry / exit of any given trade, they are sorely mistaken. Tradeable edges are TENDANCIES – they aren’t absolute, or very predictable. In every legitimate case I have tested (ie, using actual trades instead of backtesting, and yes, including Kevin’s system), the net +/- effect of jumping fully into existing positions at any given point in time will produce an outcome very slightly worse than chance.
In the past I have run monte carlo simulations based on jumping into positions mid-trade, as part of my own system analysis. I don’t do so anymore, because the results were definitive. If the system works in real-time, guess what – there is no significant difference in account returns based on your “jumping in” point.
That’s why I state to deselect the option for auto sync on my daily system. The delay would cause a serious statistical disdavantage in the short run.
This may be how the system was designed to work but I consider this a MAJOR design flaw. Under no circumstances is it in the best interest to autosync the customer into old positions even if those are potential winning positions. The risk is too great. I also think the risks of doing so are not very well explained or presented. Any reasonable person would expect that autotrading would begin only on newly initiated trades or any allocation changes would be done on newly initiated trades.
The only reason I can think of that collective 2 would want to do this is to protect their statistical accounting. In other words statistical accounting is being weighed more than the individual customer’s potential for extreme losses upon immediate autosync.
In my honest opinion I think if a court were hearing this case they would find on the side of the above customer since the damages were not the result of the trade recommendations but the flawed implementation (due to flawed design) on the side of collective2.
My suggestion is to allow an option and it should be the default option to begin autotrade only on newly opened positions, and any allocation changes on newly opened positions.
I agree with Greg’s end conclusion that autosync should be an option, but…
"Under no circumstances is it in the best interest to autosync the customer into old positions even if those are potential winning positions. The risk is too great."
I disagree. I can think of at least one circumstance where customer risk is actually reduced by joining an in-progress trade. I take advantage of this circumstance (when possible) every time I put a new system into production. It has ALWAYS reduced my risk, and increased my equity.
"I also think the risks of doing so are not very well explained or presented."
What more do you personally need than this statement? “Because of C2’s Auto-Synch technology, Collective2 will place market orders (as soon as the market is open) to match the positions that are currently open in [SYSTEMNAME]. This means you will receive the current market prices to open these positions.”
"Any reasonable person would expect that autotrading would begin only on newly initiated trades or any allocation changes would be done on newly initiated trades."
I’m a reasonable person, and I totally disagree. If I was joining a system that traded only every 3 months (or every month, or every week, take your pick), and had trades that lasted for six months, I would not want to wait until the next set of signals - I would want to be synched up now. Otherwise, why not just wait to subscribe?
"In my honest opinion I think if a court were hearing this case they would find on the side of the above customer since the damages were not the result of the trade recommendations but the flawed implementation (due to flawed design) on the side of collective2."
Are you a lawyer? If you are, I know you would never say this. Someone who blows through a stop sign and gets in an accident can’t say it is a design flaw that they were allowed to enter the intersection even though a car was coming. It is the driver’s responsibility to read the sign, make a decision (proceed or not) and live with the consequences.
I find much of this whole discussion amusing, especially now, since the gentleman (John) who brought this up earlier stated:
"I hope my experience can help prevent others from suffering similar losses, if they fail to carefully read the instructions, as I did."
I again appluad John for taking responsibility for his actions, instead of looking to place blame elsewhere.
Regardless of auto-sych, no auto-synch, or allowing someone to choose which option to use, I bet people will still complain, if their choice ends up being the wrong one.
