Please advise - Coin Collector

[heasymo]

CC on Collective2

Collective 2 reliable?

The CC REAL balance shows a loss of about 2000 usd since September 1.

The green equity curve for Coincollector on this website shows a hook up at a higher level then on September 1.How can this be since the balance today is almost 2000 usd lower ??

What do you think?

Has anyone noticed the same?

 

[J Mattson]

Are you using Andy's Open Bar System to help catch some of these moves when the market is trending strongly? This is a very good hedge system to use with CC. As I said before, in the month of August I had almost the exact same amount of fills as was shown on the system's website. Have you talked to Markus about your situation?

Ivot, How long have you been working for Rockwell?

 

[ivot]

Ivot, How long have you been working for Rockwell?

You know Mattson, what kind of dumb #$& comment is that? Just because I give my point of view/experience with this system and it happens to be pretty positive overall, that makes me a Rockwell employee?? You want me to start saying the system stinks just because that's what you want to hear? Get a life. If you have something constructive to post, then do it. Your kind of comments don't add anything to this thread. I've used this system personally, both on paper and with real money, and I bet you've done neither.

 

[J Mattson]

You know Mattson, what kind of dumb #$& comment is that? Just because I give my point of view/experience with this system and it happens to be pretty positive overall, that makes me a Rockwell employee?? You want me to start saying the system stinks just because that's what you want to hear? Get a life. If you have something constructive to post, then do it. Your kind of comments don't add anything to this thread. I've used this system personally, both on paper and with real money, and I bet you've done neither.

My,My,
Someones shorts are on tight today? I'm glad that you are doing well using CC. My experience was not quite so good. Yes, I have traded both the 60 and 40 min program. Paper trading was fine. I ran both 40 and 60 (paper) for a month and made close to $4k. I then switched to real trading and lost $4k in 3 weeks. Missing fills was a big problem. I'm sorry to have upset you. Good luck with your trading. JM

 

[rogerha]

My,My,
Someones shorts are on tight today? I'm glad that you are doing well using CC. My experience was not quite so good. Yes, I have traded both the 60 and 40 min program. Paper trading was fine. I ran both 40 and 60 (paper) for a month and made close to $4k. I then switched to real trading and lost $4k in 3 weeks. Missing fills was a big problem. I'm sorry to have upset you. Good luck with your trading. JM

Yep, I have been running it for weeks now, and have lost about $3k on a single contract over the last 6-8 weeks or so. Its main issue is when is its Limit Orders which are automatically generated on Strategy Runner and placed with the broker. These Limit orders are ambitious, and a very large number of them have not been filled. The strategy then should call for the Limit Order to be converted to a market order after a pre-determined time if it hasn't been filled, which Strategy Runner does not do.

Just my 2p's worth !

 

[Directional]

any trading system that uses backtesting to produce a possible results set, but then uses limit orders for entry cannot in my opinion be relied upon, unless past results can accurately account for missed fills, which few trading platforms can do.

It is an unfortunate side effect of limit orders that you will always get filled on the losers, but much less often on the winners - simply because price has to trade through the limit before you lose, but if the entry was directionally right then price is obviously going to be weighted and moving away from your possible entry order.

If the use of limits is to try to eliminate slippage, it is a flawed ideal. Marketable orders are the only and best way to automate a directional trading system and to ensure that your backtesting is wholly accurate - assuming you then correctly account for a tick of slippage on entry and exit. Paying up the bid/ask spread on entry with a mkt order will cost you less than the missed but profitable trades where you try to enter using a limit order.

If the system is not robust enough to eat a tick slippage on entry and exit for a guaranteed fill, then of course, its not robust enough to trade at all.

 

[JonnyT]

Absolutely correct.

When I looked at coin collector it did seem rather 'thin' which is why I passed on it.

JonnyT

 

[Directional]

The biggest flaw of backtesting is where a trading system assumes a limit was filled when price merely touches the limit, rather than actually trading through it.

TradeStation is worst for this, amongst others.

Backtesting with limit orders where this happens, is guaranteed to produce totally unreliable results, and will skew your opinion of whether a system is in reality actually tradeable given the test results are not discounting for the trades you would in reality never have got filled on.

Since limit orders will only ever NOT give you a fill on winning trades - look at your test results - would the system still be profitable if say 30% of the winners were missing? what if it was more like 50%? suddenly all those losing trades, which you WILL get filled on 100% of the time, start producing much bigger drawdowns and I'd also bet that that positive equity curve will also in reality be negative.

When automating, you cannot afford not to use market orders. If the limits are used to acheive a certain price that might be a percentage of ATR or APR away from the current price, you just have to look at ways of placing market orders intrabar when price reaches that objective for entry, rather than placing a limit and hoping/praying for a fill.

As JonnyT says, if the system is too "thin" - ie: makes too small average return per trade after losses to eat the cost of market order entry, the possible rewards of the system are just too small vs the risks to make it worth trading.

 

[Directional]

Attached is the performance report for ES from the coin collector system, from their website.

The average trade value is only $36 - thats less than three ES ticks AND it doesnt include any commissions.

If you were to use market orders instead of limit orders and account for slippage as a tick per side, this would knock two ticks off the winning trades, and add two ticks onto each losing trade.

When you also consider that average trade value doesnt include commissions which would also need to be knocked off the winning trades and added onto the losers, AND that the average losing trade is already significantly bigger than the average winning trade - the win/loss ratio is 0.67 meaning that the average loser is 33% bigger than the average winner, this doesnt look like a robust system to me, in my opinion.

Take into consideration that missed profitable trades from using limit orders will cost you much more than adding two ticks for using a market order, and that you REALLY NEED every single one of those winning trades, since the win/loss ratio is already strongly against and you can quickly do the maths on the expectancy of trading a system like this.

 

[Directional]

just a few more musings:

Extrapolating the missing data from their performance report:
average losing trade = $75.74 (loss / #losing trades)
Gross loss (exc commissions) = $22,724 (net profit / profit factor)
Net gain = $35,450

total trades = 994
Assuming 69.9% accuracy, which seems rather high
Winners = 694
Losers = 300

Assuming a retail commission rate such as TradeStation's going rate of $5 RT, though I suspetc a broker using strategyRunner would probably charge more? 994 x $5 = $4,970 of commission to come off these numbers.

That makes the presumed gains down to $30,480 and the losses up to $27,694 so including commissions - of the 994 trades, 300 would be losers, each averaging $92.31 loss making for an average trade of just $30.66 after comms and a much uglier win/loss ratio of 0.47

Not much room for paying up two ticks to go to market, since this would also add two ticks to the losers, making an average gain of just $5.66 per trade, avg loser of $117.31

I would not be risking my money with these sort of ratios, and certainly not with limit orders.

 

[ivot]

just a few more musings:

Extrapolating the missing data from their performance report:
average losing trade = $75.74 (loss / #losing trades)
Gross loss (exc commissions) = $22,724 (net profit / profit factor)
Net gain = $35,450

total trades = 994
Assuming 69.9% accuracy, which seems rather high
Winners = 694
Losers = 300

Assuming a retail commission rate such as TradeStation's going rate of $5 RT, though I suspetc a broker using strategyRunner would probably charge more? 994 x $5 = $4,970 of commission to come off these numbers.

That makes the presumed gains down to $30,480 and the losses up to $27,694 so including commissions - of the 994 trades, 300 would be losers, each averaging $92.31 loss making for an average trade of just $30.66 after comms and a much uglier win/loss ratio of 0.47

Not much room for paying up two ticks to go to market, since this would also add two ticks to the losers, making an average gain of just $5.66 per trade, avg loser of $117.31

I would not be risking my money with these sort of ratios, and certainly not with limit orders.

Good analysis of the system Arbitrageur. I would assume a higher commission rate, probably something around 8. Just out of curiosity, what do you think of this link? http://www.collective2.com/cgi-perl...13472311&session=1641951041265989550059195025

This company is supposed to be independently verifying the results.

 

[Directional]

looks pretty good - is that the same system walkforward? why would the numbers on their own website be so different?

I dont know anything about CC or collective2, my previous comments are just speaking from my own experience in system development and automated trading

 

[rogerha]

looks pretty good - is that the same system walkforward? why would the numbers on their own website be so different?

I dont know anything about CC or collective2, my previous comments are just speaking from my own experience in system development and automated trading

Me Again ! Oddly enough, I think the main reason for the difference on Collective2 and Strategy Runner results on this system are because SR only places limit orders. The collective2 system would trade the system as it is meant to be traded, ie, wait x-amount of minutes for a limit order, if not filled, then it is replaced with a market order.

Obviously it hasn't helped me much, and largely I agree with Arbitrageur, however it would be unkind to dismiss the strategy entirely, technically it appears to me there are two strategies in play here, the SR version, and the Collective2 version, I would be careful not to confuse the two. Also, look very carefully how Collective2 manage their ficticious accounts, you wouldn;t trade like that in real life. Well you might, but most normal traders would not.

 

[ivot]

Arbitrageur, If I remember correctly, all the results on the Collective2 wesite are from forward testing.


Rogerha, below is what Rockwell's site says about the differences between CC and CC on Collective2. I think it's rather unusual how varied everyone's experience has been with the system.

Q: What is the difference between ordering CoinCollector on your website and ordering the CoinCollector on Collective2?


A:

If you order at rockwelltrading.com our systems are programmed into Strategy Runner, and you can either use the “demo version” of Strategy Runner to get the signals and then enter it into YOUR trading platform or you can use the “real version” and have Strategy Runner trading it for you. For real trading, you need to open an account that supports Strategy Runner. (Here is a list: http://www.strategyrunner.com/Content/Partners.htm )
If you subscribe to Collective2 you have even more options to trade our strategies:
You can receive emails with the current trades and open positions.
You can use a so-called Instant Trade Messenger on your screen that gives you the current trades
You can have a broker trade the strategy for you. There are a couple of brokers that accept orders from Collective2 .
Or you can integrate TradeBullet with one of the following front-ends to automatically place orders:
Interactive Brokers
MB Trading
Man Financial
Rand Financial
Open E Cry
Patsystems (J-Trader)
MyTrack and TrackTrade

 

[J Mattson]

Me Again ! Oddly enough, I think the main reason for the difference on Collective2 and Strategy Runner results on this system are because SR only places limit orders. The collective2 system would trade the system as it is meant to be traded, ie, wait x-amount of minutes for a limit order, if not filled, then it is replaced with a market order.

Obviously it hasn't helped me much, and largely I agree with Arbitrageur, however it would be unkind to dismiss the strategy entirely, technically it appears to me there are two strategies in play here, the SR version, and the Collective2 version, I would be careful not to confuse the two. Also, look very carefully how Collective2 manage their ficticious accounts, you wouldn;t trade like that in real life. Well you might, but most normal traders would not.

rogerha,
When I paper traded CC I did so using SR. The system appeared to run great. Trades were made with out problems. I believe that SR paper trades the limit orders as once touched the order is made. This was a far cry from trading SR real. I believe that Arbitrageur has it pegged the winning trades were small and the losers were larger.
Also, my trades did not resemble the Collective 2 web site.

 

[rogerha]

Lost another $1k yesterday (single contract) ! It begs the question of the legalities of selling a system that with the accompanying paperwork says it does one thing (ie. Change the Limit Order to a Market order if not filled), and actually doesn't do that, and perhaps this is one for the CFTC. I did e-mail the developers on that and they claimed that the Limit Order gave better results and that you could expect about 10-20% of those orders not to be filled. This would quite possibly be the case on Paper Trading, but these claims are way out there with the fairies in the real world.

In reality you can expect about 20% of those orders to be filled and 80% not to be filled.

 

[JonnyT]

The reason you are not getting filled is simply because to many people are operating the system and the market it trades cannot handle the volume without taking out several levels.

It is a victim of its own commercial success though I doubt the system vendor would agree with this..

You know what you have to do.

 

[Directional]

Mattson, Rogerha,
does the system lose money in the market, but in its backtest show that it MADE money for the same period traded?

 

[rogerha]

Mattson, Rogerha,
does the system lose money in the market, but in its backtest show that it MADE money for the same period traded?

Indeed Arbitrageur, for all of the reasons that you have stated earlier in the thread, backtested Limit Orders !

 

[rogerha]

The reason you are not getting filled is simply because to many people are operating the system and the market it trades cannot handle the volume without taking out several levels.

It is a victim of its own commercial success though I doubt the system vendor would agree with this..

You know what you have to do.

Not entirely sure that I agree with you. Some of the Limit orders are very ambitous particularly in a tending market. If the strategy was executed as it should be (ie. executing a market order when the Limit Order fails), then although slippage would obviously become a factor, the fill rate would increae dramatically.