Reality check or can you count?

Let's analyze System1 from this trading system comparison page. This page compares three e-mini trading systems, all designed for ES, the S&P 500 e-mini futures market.

As you can see, in 2005 this system made 159.75 ES pts. It accomplished this with exactly 497 trades, which amounts to about 41 trades per month on average. The profit I quoted is hypothetical.

Hypothetical profits are for vendors (that's where the word 'hype' derives from). As traders we should be interested in actual profits. Let us estimate it then. The system uses limit orders which can cause 'non-fill' slippage which happens when a trade really does not take place because its order is not filled, but the profit from such a phantom trade is taken into account inflating the system hypothetical profits. Moreover, whenever you cannot exit with a limit order (at the session close), you experience regular slippage that happens to virtually all market orders and to a lesser extent to stop orders depending on the market liquidity and the number of contracts one trades. This kind of slippage in ES usually does not exceed 1 tick per trade for market orders and can be conservatively estimated at 0.5 tick on average for stop orders for the 1-5 lot orders.

In the past, in 2004-2005, you could have this system traded for you by a broker at Lions Futures. It is from this broker that I got the data about the actual system performance. It turns out that during the period of 5 months, from July 2004 till November 2004, the actual profit the system generated was slightly over 43 pts lower than the hypothetical one, or to put it in other words the actual monthly profit was on average about 8.71 pts lower than the profit reported by the vendor. During those 5 months the system would take about 45 trades per month on average.

Considering that in 2005 the system would call a bit fewer trades (41) per month than during the sample period (45) let us assume, probably rather optimistically, the average correction to the hypothetical monthly profit to be 8.5 pts. Multiplying the last number by 12 and subtracting the result (102) from the hypothetical annual profit for 2005 we obtain 57.75 pts or $2887.5 before any commissions. With the commissions included (assuming the round turn of $5 per contract) we get $402.5 (=$2887.5-$2485) per contract.

Suppose that you traded 5 contracts in 2005. That would make you $2012.5. At the same time you would pay your broker well over $12,000 and your system provider would earn $3,000 if you were to pay him on a monthly basis (quarterly payments would not make a big difference).

And here we come to the punch line which is this: after toiling like a stupid monkey for a whole year you would not have made enough to even cover the system subscription fees, much less your other bills!

Now, you might think that I chose a particularly lousy system to have fun knocking it. Well, actually, the really sad thing here is not my penchant for knocking others (which I might possess too but probably not to a greater extent than any other guy), but the fact that it is not necessarily such a bad system (relatively speaking), meaning that many others are even worse. If I really wanted to have a field day, I would have chosen System2 from the very same comparison page. Obviously, that would have been rather malicious of me.

Let me say it again, this time quite seriously: while I do think that this system's performance in 2005 was far from stellar, I also think that I can correctly describe it as average and so rather representative of what you can expect from many trading systems like that.

What is not reasonable and justifiable though is the subscription fee which is absolutely not commensurate with the system's actual performance. I should add that this fee gives you access to other things as well, but they have nothing to do with trading this market! Why would you want to pay more if the other things are of no use to you?

The moral of this story is quite simple: when it comes to trading systems, mechanical or otherwise, it's not all gold that glitters. After a careful analysis, such as the one I just performed, many systems with great looking equity curves may actually turn out to be losing money! This comment applies particularly to systems that trade rather frequently. I am not a big believer in mechanical systems that trade more often than 20 times a month, especially when they use limit orders to initiate or close their positions.