I’ve seen you define it this way but I’m still not sure how you would use it as a timing statistic can you explain exactly what you mean?
Does this refer to your earlier comment that “long term holds when they are timed properly have outsized APD’s, as was the case with several buy hold systems with hold times out past 2-3 weeks” ?
So you’re saying a good APD indicates the system times it’s entries well?
Exactly.
I’m not entirely convinced a high APD would indicate that, or if it did I’m not sure it’s relevant for those type of systems anyway. To me if someone holds something for long periods of time, say buy and hold or trend followers, they make most of their money by being able to run winners so the entry is one of the least important parts of the system, it is largely irrelevant to them. Your entry signal merely dictates frequency of trades, not how much you make on the trade (reference also the ‘Palladium Watch’ thread discussing the futility of limit orders when one intends to hold something for over 6 months) so the timing of the entry is a fairly moot point, but I imagine it would be more relevant to scalping systems going for smaller gains.
I must say, I feel a little used over this whole issue. And though I almost hate to say it, I’ll bet Ross does too. Matthew, you Jon, Jules, and whatever other henchmen you have out there can do what you please. Just don’t play games.
Any computation that states something to the effect of being "low / less risk" should not be stated on any credible web site, trading system, etc.
Anyone with experience knows that trading / investing is [extremely] risky with potential for [substantial] monetary loss at any moment for any number of reasons.
Any indicator / computation that states otherwise is flat out misleading.
If APD demonstrates that a trader / system past positions to be on the winning side from entry for a longer period than a losing side, then that should be stated, and that alone. Avoiding any type of "safer / less risky" type of wording fairly serves those that are trying to make sense from the computation.
Playing games? We sought opinion on a subject where we knew others shared our concerns, we’ve had an open debate in a public forum, had over 130 posts over several threads over 10 days on the subject and submitted a final proposed text for everyone’s consideration in a democratic process, and rather than the more radical aims of yourself and others trying to remove APD completely we are merely trying to change the wording.
After you’ve had a little lie down trying to recover from being used maybe you should ask Ross what he would prefer, having APD remain intact with a more accurate description as we would propose or have it removed from the site completely as per your wishes.
Man up, Keith.
Agreed, and that’s why the proposed new wording has removed the references to degrees of risk, and so far no-one has disagreed with this.
Quit sneaking around. If Matthew told you to get a consensus modification for the APD wording, but that removing APD was off the table, and you had made that known at the start of the “democratic process”, I wouldn’t have made a single post. I don’t care what the wording is. But you, Jules, and Matthew failed to convey that little bit of information.
I’m sure the others that made arguments to eliminate APD had the same hopes as me: maybe we can convince Matthew to get rid of that meaningless stat.
Next time you’re going to start a “democratic process”, let everyone know how it’s rigged beforehand.
Matthew did no such thing, and he never said removing it was off the table, you’re putting words in his mouth.
This all started on the 24th when (ironically) Ross made a rather unprovoked attack on the Spyder system thread using the APD as his basis and the debate began there. Together with Jules we turned it into a conversation about how we might achieve a compromise once and for all and others joined in. It was then AND ONLY THEN that I contacted Matthew to bring his attention to the thread highlighting that we were having a surprisingly productive conversation (as at that point Ross was agreeing he could see our point) with the possibility of arriving at some kind of consensus, adding that he should feel free to comment, he responded by saying he thought it was good and applauds it and would try to accomodate any conclusions we reach. I had and still have no guarantees that anything will be adopted but it was only right to at least consult him in the process otherwise what are we doing it for, what could we possibly gain?
All your above statements are totally without foundation, this has been an open and honest process throughout with no hidden agenda and has involved as many people as possible. Inviting what now amounts to over 130 posts on the subject is hardly sneaking around, if I wanted to try and force something through without consulting anyone else I would have just contacted Matthew direct and suggested the changes there and then, believe me it would have been a lot easier for me to do so, but it wouldn’t have carried the validity and weight of consensus opinion that has developed by making it an open and yes, democratic, process.
Keith, I am not playing games. My opinion is from the beginning that APD has some positive features but also some limitations. I publicly pointed out one limitation in the Spyder thread. I suggested several alternative indices on this forum, but in each case I quickly realized that they had other limitations. After my failure to improve APD (and remove the present version) my opinion became that APD should stay but that the interpretation should change substantially, and that this should begin with the mouseover text. I have defendend this opinion from that point on. I also had some private communication with Jon and I said so on the forum in a reply to Beau. No surprise there.
So it is not the case that I started this discussion with a fixed opinion that was boiled in secret conclave with Jon and MK, if that is what you think. My opinion was formed independently and changed during the discussion from “improve APD” to “improve interpretation”.
Another matter is that during this discussion Jon has contacted MK to ask whether this whole disussion made any sense at all, since obviously if MK is not willing to change anything we can debate the topic untill we see blue in the face but that would all be futile. What I understand from this (but I had no direct contact with MK) is that a change of the mouseover text would be a possibility if we all agree.
I don’t know for sure how MK feels about removing APD, but it doesn’t seem to be an option since we have been over this many times when APD was introduced. Clearly he was not willing to remove it, but he did change the text. Even if removal is an option, I would be against it though. I think that Jon’s text is a fair compromis between the different opinions.
Matthew did no such thing, and he never said removing it was off the table, you’re putting words in his mouth.
Getting rid of APD isn’t on the ballot, but a choice between a new wording or retaining the existing one is. What’s your vote?
Jon, what is going on here. Who elected you.
Matthew always follows what goes on in the forums, and when he feels the time is right he steps in to make a decision if he feels it is warranted. But for some reason you decided to make a decision for him.
No-one has been elected to anything, and neither have I ever stated that I have been elected to anything, we have done this on our own volition and invited the views of others which most have given cordially. Anyone is free to start a thread on any topic and invite comment. That is what a forum is for.
I haven’t made any decision for MK nor have I claimed to, he is perfectly capable of making them himself, but to canvas opinion with everyone and propose changing something without at some point consulting the very person capable of changing it would be the ultimate in stupidity.
At this point, I will endorse the textual changes.
This entire argument from the anti-APD ranters is getting ridiculous. People getting angry for the sake of anger. If you don’t want to use it, don’t.
Thank you Ross, I appreciate you doing that as I know you had reservations, I believe the proposed text is a more accurate description and interpretation than the existing one.
Would you care to post the final text for me to see?
The APD Ratio (“Average Profit to Drawdown”) measures how much profit is made for each dollar of drawdown suffered while a trade is open.
A high APD means that on average, the system makes trades with high returns in comparison to drawdowns.
A low APD can result from averaging down or allowing open losses to run in the hope they recover to achieve relatively small profits. This is considered a high risk strategy.
Other possible causes of low APD are the system applies a hedging strategy, or the system has many small losing trades and less frequent but large winning trades. These strategies do not necessarily imply a high risk.
In summary: a high APD is good but a low APD is not always bad.
Where this system stands: 10.7% of systems have a better APD than Broadsword Macro’s 0.41 (and 89.3% have worse APD stats)
General Distribution of APD stats on C2:
85.1% of systems have an APD of less than 0.20 (low)
4.0% have an APD of between 0.20 and 0.40 (medium)
4.4% have an APD of between 0.40 and 1.00 (high)
6.5% of systems on C2 have an APD of 1.00 or better (exceptional)
Hi, everyone. First let me apologize for my silence during this very interesting discussion. My intent was not to seem remote and mysterious. Rather, I have been traveling these past few days, and – between travel and trying to keep up with normal day-to-day site operations – I haven’t had much bandwidth to follow this discussion carefully.
Indeed, I must admit sheepishly that I haven’t yet had a time to really thoroughly every single post in this thread, and so I am hesitant to interject now. However, it seems that tempers have flared, and partly this is because my intentions (as administrator of the Collective2 site) are unclear.
So let me step in and make a few very general points. I hope this will serve to keep the conversation on track (and, even better, will serve to soothe people’s nerves.)
Contrary to how it sometimes appears, my goal as the administrator of Collective2 isn’t to punish system developers or make them miserable. Nor is it my pleasure to play favorites and pick the winning and losing systems of this site. I am aware, of course, that my decisions do affect the livelihoods of system vendors, and so I try to be thoughtful about that, and not act capriciously.
But let me state very clearly my overriding objective. My goal is to help trading system subscribers – not vendors – make the optimal trade-off decision between risk and reward. There is no way around that trade-off decision. In the world of financial speculation, if you want higher returns, you necessarily take larger risks. Unfortunately, this is very easily forgotten by both system creators and subscribers.
Let me take this a step further and be even more specific. My main goal for Collective2 is that system subscribers never be surprised. By this, I mean the following. If a system subscriber wakes up one day and loses a lot of money while following a trading system, then of course in one sense this will be a surprise a very unpleasant one. But my goal as Admin of Collective2 is to make sure that, when a subscriber loses money, he says to himself, "Darn. That is very unpleasant. I hope to make up this loss over time. But I can’t say that I was unaware of the risk of this loss’ occurring."
I admit that, to a certain extent, this goal is somewhat impossible to achieve and that we can only work toward it in an iterative process. This is due to the imperfection of human beings (myself included!): Most of us are weak and greedy optimists; and most of us are bad at judging probability and risk. Taken together, these two facts combine to create a very dangerous situation for most traders.
So the goal of C2 in general and C2’s statistics and user-interface in particular must be to try to help human beings overcome their frailties: C2 must work to reveal risk and let subscribers know what they are "in for."
What about the APD?
It has been a long time since the APD was initially developed, and some of the events that led up to its creation have been lost in the fog of time. But I do remember the initial impetus for its creation. Before the APD, there were a large number of systems on Collective2 that seemed unusually attractive and risk-free (due to the way C2 stats were calculated and equity charts drawn periodically) and thus these systems garnered a lot of subscribers. However, many of these systems ultimately lost money, and so too did subscribers who followed them.
So one of our members, Ross, proposed a stat which seemed an effective way to reveal this class of systems. Over time, the name and description of this statistic changed, but the stat remained the same, and it has effectively served to at least call attention to some of the risks inherent in certain trading styles.
Unfortunately, there has been a trade-off, because the APD – while warning people about certain trading styles – also catches in its net a certain class of systems which may not be as risky as the APD indicates (and may, in fact, on a long-term basis, be less risky than those systems not flagged by the APD).
So what is the implication of this in terms of C2s “APD” and other statistics? Two important points here:
1) First, I dont pretend to be an expert – either at trading, statistics, or probability theory. While I never specifically asked anyone to start this discussion, nor appointed a secret cabal to come to a decision, I am incredibly grateful that the C2 community has taken it upon themselves to improve the site. Im glad that this discussion is happening, and is taking place in an open forum, so that C2s many users can contribute their expertise and experience to arriving at optimal solution.
2) Everything is “on the table.” Whether we as a community decide to jettison the APD altogether, or merely change the wording which describes it, or supplement it with another statistic, or maybe replace the simple math formula with a more subtle statistic/algorithm
it will be okay with me – providing that the path we choose helps achieve C2s goal (more on this in a moment).
I want to be clear that this message is not intended to stop the discussion, or even to slow it. My hope is that by posting a message as the C2 Admin, I can eliminate uncertainty about this process, and therefore encourage even more, broader participation.
May I therefore inject a few ideas into this thread, and see if they inspire some additional participation by the C2 Community?
First, because C2 is a giant computer program, the algorithm we use to determine statistics or ratings of systems does not need to be a simple algebraic equation. I say this only to remind you that the ultimate outcome of this discussion could be a more complex algorithm – a process, and not a formula.
For example (and here is an idea I have not reflected on in any way!): The algorithm could first determine the trading “style” of the system, and exempt from APD risk flagging any system that trades with a certain style (i.e… long-term trades, no averaging down into losses).
Or perhaps we jettison the simple APD number and change it into a set of words. Again, an idea I have not reflected on: Systems could be assigned Style ratings – e.g. long-term trend following, Diversified Portfolio, or whatever. Perhaps then risk can be assigned based on the historical performance of other systems in that class? (I know, its a crazy idea, but maybe theres something there?)
Anyway, these particular ideas may be dead ends, but I hope they serve the purpose of broadening the discussion.
Let me close this post by asking for the following. First, please dont stop the conversation, and please dont restrict it. But do remember the following, because its the most important point Im going to make in this post. Our goal as we try to come to an answer is not to protect system vendors, nor to favor one style of system vendor over another. Our goal is to maximize the transparency and usefulness of the site from the perspective of the system subscriber.
To say it another way: there is nothing wrong with high-risk systems. For many traders, a high-risk, high-reward system is an appropriate component of a financial portfolio. But traders must be aware that the system they select is a high-risk system. The mechanism to achieve that awareness to let Joe Trader (an average person, and not a mathematician) understand the choice he is making – is what we are trying to arrive at.
So democracy and openness is great, and Im all for it. But remember, as you contribute to the discussion, that the goal I, as admin of the site, am trying to achieve is to help the average trader make wise decisions about his money.
Please help me to help him.
-Matthew
Excellent post and thank you for clarifying the situation and making your thoughts known.
I believe the final draft of proposed text (in the post above yours) fulfills a lot of the functions you mention namely as a tool for subscribers it better educates them exactly what APD does and how it could be interpreted.
I am happy to simply adopt the changed wording. If others were to feel strongly enough to jettison APD completely so be it, but as I have stated previously I do think it has merit in highlighting systems it was designed for, the problem as you say is the others it catches in it’s net and it’s those that the new proposed text gives allowances for, so for me with the new text we have an agreeable compromise.
Interestingly there have been no major disagreements with the specific text, and Ross himself has now endorsed it, so it clearly doesn’t say anything factually incorrect, most peoples arguments now seem to be about the process by which we arrived at it or wanting to go one step further and just remove it completely, rather than the content of what is being proposed.
My summary of what has transpired to date is all those who specifically expressed a view on the text supported it, those that didn’t haven’t come up with an alternative one but instead argue to remove APD completely.
"My goal is to help trading system subscribers – not vendors "
Exactly. Vendors could argue against any stat that paints them in a bad light. And whether they like it or not, there will always be a stat that will do so, no matter how well they think they are trading.
For example, the fixation on winning percentage is well known as perhaps the LEAST important and most manipulable stat in the industry. What sets C2 apart from other timing services, is the suite of tools and statistics to let potential subs look under the hood at things that vendors would like to suppress.
The main metric for traders worth their salt, is Risk-Adjusted Return. However, given the variety of styles, instruments, methods used by vendors, that is difficult to grasp. Anything that helps moves the whole C2 site forward.
Thank you for clarifying what we can and what we can’t. Particularly I am glad to hear that everything is on the table and that more complex processes are possible too, because I think that’s what ultimately will be needed.
For now, however, I have no concrete suggestion that would be better than APD, so I still vote to keep it and to change the text as Jon suggested.
