Real Performance Measure

Forgive me if such a measure is already available.



The three main reasons I like to see actual trades from autotraded accounts are.



1. Gives me a glimpse as to how many people or scale the system is being traded.

2. More information about the type and magnitude of the slippage

3. To see if anyone has actually made or lost money following the system.



I understand that publishing how subs a system has is not likely, so I am not asking for that, even though I would like to see it as a sub and not as a vendor.



My suggestion is for #3, can we have a measure or indication of the system performance only using real trades?



A. Equity curve with only real trades (at least one autotrade) scaled/normalized to the system scale.



B. Equity curve with only real trades, showing the actual gains in USD for all autotraders combined.



C. You could to the same as in A or B with a single performance metric (instead of a curve) like APY or the like.



Jim



Jim,



How does showing the actual trades from autotraded accounts answer any of your 3 questions. EOD or EOW (or longer) systems may have a small % (or no) of subscribers that autotrade.



You mention “real trades” and traders actually making money. Again, I’m wondering how the autotrading stats would provie you with this information? I may be wrong but It was my understanding that you can autotrade to a test or “play” account. If so, autotrading stats have nothing to do with “real” money or trades.



C2 already provides a good number of performance stats as well as every trade the system performs. I don’t see how autotrading performance stats would enhance what is already available.



Am I missing something?

Demo autotrading is limited and I don’t think it is part of the numbers that were displayed on the old site as “show actual trades in brokerage accounts” nor are demo trades used to calculate slippage numbers. As a result, I am only talking about real verified trades, not “play” trades.



The real trade information is a “final word” as to whether a system is “tradeable” at C2 via autotrading. That is the main reason for the request. If I can easily see time periods of comparable performance with real money and with hypothetical money, I can get a good idea whether the system is really working or if/when a system becomes over-traded or over-subscribed.



As you mention, this is most relevant for faster trading systems (day trading) that aim to capture small moves and much less relevant for systems that hold a position for days or weeks and aim to capture larger moves. Plus, for slower systems with defined entry times (EOD or EOW), manual trading is a option, so the autotrade stats will not be complete.



Nearly all the systems I have traded on C2 would be difficult to manually trade, so the autotrade stats are very relevant to me.



An example of a system that I believe seems to not work with real money is “daily income”. I know there are other reasons why one would not trade this system, but what I find odd is why every time it gets an autotrader, they lose money and quit, then the system starts gaining again.



I hope that helps.



Jim

It’s important to remember that we automatically change the hypothetical fills you see on a system’s C2 trading record to the volume-weighted average fill price achieved in real live AutoTrader accounts. (These should still be considered hypothetical prices, since no single trading account actually achieves the exact results reported on C2. However we think these kind of hypothetical numbers – essentially average real-life fill prices – are nominally more useful than purely imaginary quote feed numbers).



This changing of C2 fill prices to the actual AutoTrade fills from live (not demo) brokerage accounts happens automatically, by default, for any system with AutoTraders. This explains why some systems start to perform worse as more people AutoTrade it: because real-life reveals that you can’t really buy at the bid and sell at the ask, which is essentially what some of these scalping systems try to do. These kind of systems tend to look good in hypothetical backtesting, but are not really tradable at any sort of significant volume in real life.

It seems to me that for hypothetical fills, one should assume a buy at the ask and a sell at the bid and not the other way around.



I like the change to real data once available. Essentially what I am asking is that there be some way to easily distinguish between the real and hypothetical fill data at an equity curve level or with separate performance numbers.



Jim

>> It seems to me that for hypothetical fills, one should assume a buy at the ask and a sell at the bid and not the other way around.



Yes, of course, and that is what we assume, but I was being a bit facetious. What I meant was that a lot of these scalping systems that look so good on paper will post a limit order buy at the bid, and then a limit order sell at the ask. On some kind of hypothetical basis, you can actually make these kinds of systems work and look good, because there is always one or two units traded at those levels. But in real life, of course, the chance that you are the trader who gets those lucky fills, both buy and sell, is extremely remote.

I see, I suppose the "Keep after worse case slippage" stat addresses this issue. As it would be very poor for such a system, like daily income.



Jim