# Statistics issues

Dear C2,

these are the stats from my newly listed system:

4 wks

49

1.6 days

73.5%

6.0

\$570

248.0%

I can’t see from this window what the 49 means but:

Av trade length: 1.6 days. In fact, I have put on three positions, rolled over one of them (today). One of the three is in fact an addition to an existing position so it is debatable if it is really one trade or two. But in any case, they have been in place for far longer than 1.6 days, but obviously the recent rollovers have pulled that out of wack.

The net P/L on the TY trade (short 3 TY) is accounted all together on the trade lists of course because they are all in the same contract. The other trade is long 2 CLJ5 which rolled into long 2 CLK5 today. I made a mistake and placed two separate orders to buy the May and sell the March (1 contract each time instead of doing 2 contracts the first time!).

So now I have 6 trades listed with average length of 1.6 days when in fact there are two positions, neither one of which has closed out and both of which have been held for much longer than 1.6 days.

Now: A W/L ratio of 73.5% even though all positions have so far been profitable - even the newly rolled over positions but perhaps that will be updated later. But this stat shows also that it will be wrong at some point when there are both winners and losers.

And average profit of \$570.00 presumably because the total profits are divided by 6 instead of 2. But wait: an average profit of \$570 x 6 trades = \$3420.00. But the total open profits are \$10,000 so I have no idea how to make sense of that one. It is simply wrong whether you do it by position (2) or by your trade calcs (6).

The annualised return I presume is correct, but given that so much of the rest of it is misleading… and that an annualised return after 3 weeks is not all that meaningful…

Suggestion:

1 de-emphasise W/L by trade and rather track it by position if you can with futures accounts. Since they are tracked all together (2 CL, 3 TY) this shouldn’t be too hard. If you can’t, then this should be explained clearly somewhere on the website, especially over the longer-term systems for which this is often an issue. Perhaps you can add a ‘rollover’ option in the trade entering zone so that it gets recorded differently?

2 Show the % drawdown and % profit on the entirety of the account vs. breaking it down by individual trades. The whole account is more important than the indiv. trades anyway. If you want to show average profit, then show av profit per month of the whole system vs per trade since this one is inaccurate for some systems.

I am sure your method of calculation works for some systems. But since it obviously doesn’t work for mine which only does standard things like buying, selling, exiting and rolling over, I feel this is ‘unfair’ and at the least misleading.

The suggestions I have made will not penalise my system which adds into existing position and rolls over - like nearly all futures systems - whilst giving, I think, a better picture of overall performance for all systems being tracked. So nothing is lost to any other developer by the above suggestions.

In any case, the fact that the nr of trades, W/L ratio, profit-per-trade and holding time are all wrong for my system presumably means they are wrong for many other systems, so now I don’t really trust the stats, only the equity line. Everything else might be quite misleading…

I appreciate this is a new operation; I am glad you are there and glad to be involved, otherwise I wouldn’t have signed on! But there are many inconsistencies and hopefully these will be addressed soon. Otherwise I suspect that many people who first check it out from your many Google ads etc. ( for which thanks and congratulations are offered!), will take a quick look, find it hard to navigate (which it is) and compare systems etc. and then never come back.

As I suggested earlier, having 3 week track records showcased on the opening page for people clicking in from the ads gives the wrong impression, even though I was delighted to be featured there!

All best,

A. Howes, VIVALDI seasonal trends system.

In addition, I don’t think 3 week trach records qualify for the best trading system catagory either. I would only put my money on a system that has much longer period of above average performance. I would really like to see systems with at least one year track records of above average performance as being a requirement although I understand that not many systems would qualify since most trading systems on this site are younger than that. The bottom line is that the best trading system catagory should contain trading systems that people can subscribe to and feel confident in. Whereas a system that has only been up for a few weeks with an annualized gain of 1000% is surely not realistic in the long run and will probably realize a large draw down sometime in the near future. My vote would be for the best trading systems catagory to contain trading systems people can feel confident in that will give them a good return with minimum draw down.

For simulation trading, high % gain is absolutely sustainable.

I’m not talking about simulation trading. I am talking about traders subscribing to a trading system that they plan on using to trade real accounts. In that respect I would think that this website would feel the obligation to the trader by promoting trading systems with long term proven track records to help protect the trader who has put their trust in the trading system ranking algorithyms of this website.

Alan: I agree with everything you have said in terms of length of track record, and for real vs. for simulation.

C2 tracks systems real-time. So even if a system has been around for a while, it only tracks from the day trades get entered into the C2 tracking system. This is not a problem with C2, or the system or anything else. It is just a fact.

That said, if a system stays with C2 over the longer term, then hopefully at some point there will be section for systems at C2 for over 1 yr, 2 yrs etc. because as you say, their track record becomes much more meaningful for those who value such things.

Also, you can use the C2 display to check into a system further by visiting the developer website, seeing if there are other track records available etc. vs. just the C2 display alone, especially with a system recently being tracked (like mine for example!).

I personally think it would be ideal if there was a link on C2 to a hypothetical track record that can be compared to C2 tracked results, but I suspect this is much too hard for Matthew to put together, especially since many of the systems here are not mechanical, rather advisor-services who are calling trades into the tracking system and for their subscribers. So there are two very different approaches going on.

In this regard, it would be better also if mechanical systems were listed separately from the discretionary ones. A mechanical system can be traded in many different ways at different brokerage houses, but an advisory-based system needs a different setup arrangement, usually the subscriber being available to place the trades etc.

A. Howes, VIVALDI Seasonal Trends.

I think there are 3 types of trading: discretionary, technical, and strategy-based.

A discretionary trader uses a combination of expert judgment and non-quantifiable data to determine when to enter and exit the market.

A technical trader uses a combination of indicators, hotlines, newsletters and objective rules to enter and exit the market.

A strategy trader uses a combination of expert judgment and technical indicators based on a strategy (for eg., search for value stocks as opposed to growth stocks)a method of trading that uses objective entry and exit criteria that have been validated by historical testing on quantifiable data.

Strategy traders are restricted by a set of rules. These rules make up what is known as the strategy. As a strategy trader, you will not deviate from your strategys rules at all, unless you have decided to use a different strategy altogether which might result in a different trading style (long-term, short-term etc.). When your strategy tells you to buy, you buy and when it tells you to sell, and exactly how much your strategy tells you to.

As a strategy trader, you’ve spent a lot of time and research in creating those rules. Your rules have been hand-designed by you and tested and re-tested on years of historical data. This testing has given you positive results and the conviction that lets you know its time to take your strategy into the future. Your emotions might still fly as high and low as the market, but at least they are not causing you to make bad trading decisions.

He has realized the value of quantifiable data and back testing, and starts to put on trades with the confidence that comes with knowing the historical track record of the same strategy for the last several years.

A strategy trader thus moves into a mode of acquiring and testing quantifiable data as it relates to historical price activity. This is a marked difference from a technical trader, who tries to correlate data to price but usually through observation, and from the purely discretionary trader, who doesnt use quantifiable data at all or feels he needs to in order to make money.

Technical analysis helps investors filter out all the blather and hype about markets and the economy. It’s based on hard, objective data from the market itself – not the half-baked opinions of talking heads on TV who have their own axes to grind.

I’ve also learned that every indicator requires interpretation to determine the levels that correspond with bullish conditions and levels that are bearish. No two individuals will analyze a group of indicators in precisely the same manner. It is a subjective analysis (though not purely discretionary) that requires attention each and every day.

In my opinion, both the purely mechanical and purely discretionary approaches to trading are inferior to an approach based on a sound strategy.

That is why there’s no substitute for expert judgment based on a strategy (managed entirely at my discretion – using all of my technical indicators and my own investment and trading judgment, applied across a broad array of asset classes - strategy trading) in investing.

I believe that most sysems at C2 can be classified as neither purely discretionary nor purely mechanical, but strategy based.

rgds, Pal

Midas Long-Term Value

Midas Short-Term Value

Well, Pal, you put a lot in there.

I think the best traders are those that trade according to their character for one thing. Some are on the floor and go by gut and make tons. Most of us are not on the floor and need some sort of data-based (usually charts) system.

The next level is someone who uses a combination of indicators and rules but uses discretion or ‘feel’ as to when to pull the trigger on a signal - many of which are mechanically generated, or, just as important, which markets to get into and which to avoid based on factors extraneous to system rules. i.e. the grains are moving now, I’ll apply my system/approach to them and de-emphasise equities which aren’t going anywhere for a while.

As for mechanical, I actually thought at first that since this is - in part - a system-ranking service that most of the systems were mechanical, but of course since they are entered (by voluntary or involuntary discretion ) by the developers, it appears that most of them are discretionary.

I am not sure about your distinction between technical and strategic and won’t comment. But some rules-based systems are indeed strategic and some are not.

My system VIVALDI here at C2 is purely mechanical. I believe the best results would be obtained by using discretion on top of the system but that is not what I will be submitting here since I am using C2 to have a third party verified track record for a mechanical system.

I do agree with you that a good system should be strategic, although sometimes strategy and tactics get blurred in this realm. One strategy I had in my system, for example, was to make the core rules a combination of non-correlated factors - seasonal, trend-following and a filter. So there seasonal statistics in terms of times to buy, sell or avoid; then a trend-following oscillator (or indicator if you will); then another indicator based on ADX that acts as confirmation or denial for trend following. These three core elements, although they all use closing price as referent, are non-correlated in terms of how they evaluate market action. They do not ‘see’ each other’s point of view in other words.

I have also been working on a COT-based system but haven’t found a way to make one that is really solid, i.e. not overly dependent on curve-fitting. Similarly, I have recently coded in systems based on volume with similar issues, but both look very promising. At some point, I would love to blend them into the existing VIVALDI core tactics, as part of the overall strategy of combiniing valid systems together as long as they use non-correlated rules. When several of them agree, probabilities of success go up. The higher the historical probability over long periods of test data, in my opinion, the less likely degradation of the system.

Now most indicator-based trading rules are very similar in principle. Lining up stochastics, adx, volume, weekly trend, pivot points etc. As long as most of them are basically giving a different read on the markets, then when they co-incide they are meaningful, so I am not saying anything new.

But the best ‘system’ involves rules-based discretionary trading. I myself cannot do it because I am far too emotional with trading, which is partly - if not mainly - why I got involved in system design, especially after finding out that so many out there really aren’t all they are cracked up to be. But no mechanical sytem can substitute for an experienced trader who combines solid strategy, or rules-based analysis and entry/exit levels with some sort of discretion, along with the ability to make fast decisions and not agonise about them one way or the other.

Humans always beat machines…

Agree, human always beat machine.

The best the machines could do is to force us to a draw…

Humans always beat machines… in the long-term, but the machines have an distinct advantage; they have more stamina; and are getting more powerful every day; but so are the humans…

The most powerful neural-net in the world is, was, will be, had to be, the mental neural-net we are born with, because it has taken millions of evolutionary years to shape it. So, an occassional draw, nevertheless a humiliating one, as evidenced by the draw between IBM’s Big Blue and the then world chess champion…

Are the machines poised to take over the universe sometime in the future as portrayed in the movie Terminator? Machines lack a soul. The soul or the ego is the faculty that thinks, acts, feels and judges independently based on preceptual (sensory) data. A body without a soul is a corpse. A soul without a body, a ghost…