Averaging Down - Beware

Many trading systems average down. This may work for a period of time but eventually it will cost them. I only noticed they average down by mistake as my mouse went over the BTO/STO trade symbol.

Does not averaging down amplify the risk to an already losing position unless the vendor is 100% sure that the market will rebound? I dont think so, nothing is 100% in trading.

Beware subscribers as most vendors do not disclose this strategy in thier systems description. When evaluating a trading system move the MOUSE OVER the BTO/STO trade SYMBOL to check if the strategy incorporates an AVERAGE DOWN philosophy.

There have been a lot of discussions about this. I think that averaging down is not necessarily dangerous, but it can create a large position and that is risky. It depends also on the size of the initial position and how many times it was multiplied. So I agree that when a system averaged down, it is wise to investigate how much risk was taken, but the averaging down alone is not sufficient to dismiss the system.

100% totally agree. Another comment is the time interval and difference between successive positions may not be feasible or as profitable for subscribers in real life trading.

Averaging down is a technique of buying into a postion that is going down. Possibly the most important straigie to utilize when trading is to buy positions that are going up. Dollar cost averaging is a term that was created by wallstreet insiders. Although the idea can be twisted around and presented as a wise thing to do. It is still buying a postion that is going down.

Do you refer to the usual slippage problems? Do you think that this is more a problem when the system averages down than it is in other trades? I would think that since we are talking about a situation where the value of the position decreases, a delay at the subscribers side will easily give him a better fill than C2.

Excellent point, never thought of it that way.

Averaging down is feasible strategy if and only if traded in a risk controlled manner and a stop in mind.


The subsequent position added should be smaller than the original position size and also have a smaller stop than the original position stops.

Different types of averaging down exist, ie one has to separate

a) controlled "scaling in" a position, whereas the total maximal position size is predetermined before even opening the first position.

This is done in order to get a better entry price while not risking more than initially planned (max positionsize * max stoploss in pips/ticks).

In my opinion, this is a reasonable strategy, and there is nothing wrong doing this.

Nevertheless, it is absolutely important that maximum positin size as declared in the system description is never exceeded.

b) uncontrolled adding to a position by taking unreasonable position size with regard to the equity. Exceeding of position size as declared in system description, in most cases combined with exceeding of max stoploss.

While there is nothing wrong with a), you should avoid systems where you find trading behaviour as described in b).

True. The drawback to a potential subscriber is that if the vendor averages down frequently in a trade then he/she would have to be alert and active even after the initial trade has opened which could be a real life trading problem. Would auto-trading remove the subscriber from this predicament?

Yes, autotrading would remove such problems.

Especially when trading Forex-Systems, I would strongly encourage auto-trading, since this is a 24 hours market, where a trade can - by definition - be opened and closed at virtually any time from Monday till Friday.

A subscriber should be aware of this fact in the first place, and if he or she comes to the conclusion that a system with trading hours from 9 till 4 EST is better (and where entries and exits can be watched closely ), then perhaps a Stock-index trading system with no overnight positions is more convenient.

That is why the APD ratio was added. Systems that average down rather than have a competitive average usually have a considerably lower APD value than other systems. In other words, averaging down tends to increase the max DD for that trade.

From the track record table, for all rows where max DD has a value:

APD = (sum the net profits column) divided by (sum the max DD column)

Is it fair to say that systems with a high APD are better? Never heard of APD outside this website, but seems like an excellent useful indicator. Some systems have a low APD but a high sharpe ratio. Thats strange and I will have to do more research on this.

APD was how I screened systems on my own. Many systems here were scalpers or averaged down heavily or would hang on until a losing trade turned positive to make their records look better, and I noticed that they had large drawdowns on some trades, where they were "holding and hoping."

I just contributed it to the site owner. Others have also contributed things as well, to make this a better site.

Generally, systems that outperform due to trading savvy will have decent APDs. Those who outperform due to refusing to take a loss or those just entering and waiting until a trade shows a profit will have poor APDs. It also seems to have some value against scalping systems (which tend to have poor returns when autotraded).

Congratulations, wont be long before APD is used outside this website as well. Why are the majority of systems under the APD 1.0 threshold. Does this mean most systems average down or ride out large drawdowns and quickly take a relative small profit. On the surface this appears to be inefficient trading or is an APD of 1.0 too high a standard.

A higher APD means that you have to bear less risk to make xy amount of profits.

IE if you have 2 systems with an ending balance of 150k (both starting with 100k), and system1 has an ADP of 0.2, while system 2 of 0.7, it means that the overall risk of system2 was smaller.

So with system2, a lesser amount of stress (drawdown) led to the same result.

So in a way, the higher the ADP, the better.

From time to time, you can see that Sharpe ratio is high (above 4), but still ADP is poor (smaller than 0.4). My explanation of this occurence it that sharpe is calculated on closed equity only. So it is possible to average down a lot during a short trade (with a high intra-trade drawdown), and close the trade with a very small profit. The result would be that Sharpe ratio remains high (since the intra-trade drawdown is not recorded, while the trade is closed at a small profit), and equity curve remains smooth.

However, such a trading “method” will reduce overall ADP for the system, especially if this is constantly done by the system vendor.

My observation (correct me if I’m wrong) is that sharpe ratio does not necessarily show poor trading methods (averaging down and risking a high amount of money for small profits), while ADP does.

So look for systems with high ADP (bigger than 0.5), and high sharpe (higher than 3.5).

You can also calculate (avg win$ /avg loss$) x (profitable trades %/100).

If the reading is above 1.1, this is a good sign as well according to my observations.

To make an example:

My system X-Ray-signals.com YEN (96 trades until now) currently shows

ADP of 0.68,

Sharpe of 4.648 and

1.20 ((avg win$ /avg loss$) x (profitable trades %/100))

Of course there are some systems around here at C2 with better stats, but not a lot.

To do a quick raw evaluation of a system then I could multiply APD x Sharpe x Expectancy. The higher the result the better. Can this site do this type of filtering. Thanks everyone. Very informative.

Just if anyone out there does not know how expectancy is calculated:

Expectancy = (Probability of Win x Average Win) - (Probability of Loss x Average Loss)

I would suggest not to use expectancy ($) in the evaluation, but - as indicated above -

(AVG Win$ / AVG Loss$) x (X %Profitable Trades/100)

Lets call this number WinFactor.

because the result - when multiplying

ADP x Sharpe x WinFactor

will be position-size neutral then. IE it would not matter if I traded 10 contracts or 1 contract. In contrast, when doing the calculation with the expectancy in $ number (instead of WinFactor), it would make a big difference.

Trading is very simple and it is very dificult. The dificult thing about trading is keeping it simple. Traders that buy positions that are going up make more money than traders that buy positions that are going down. Simply put. Buy appreciating assetts.