Ladies and gentlemen and boys and girls… I give you the keys to the kingdom (for this system anyway).
As per the spreadsheet below I’ve calculated the correct position sizes to be taking with this trading system based on Risk-Based Position Sizing as detailed in earlier posts.
A 5% risk per trade is assumed.
A $100K trading account is assumed.
VERY IMPORTANT NOTE TO ANYBODY THAT SEES THIS AND DECIDES TO IMPLEMENT IT FOR WHATEVER REASON:
If using these calculations be aware that if an instrument is NOT denominated in the base currency of your account or if you’re not trading with a spread betting broker: you MUST convert the currency of the instrument being traded into the base currency of your account. This is VERY important. This is not something I have to concern myself with at my broker as all prices are quoted to me in the base currency of my account. This is really simple to deal with. But at other brokers the DAX, for example, will be quoted in EUR. You MUST convert EUR to USD (or whatever the base currency of your account) otherwise you’re going to land in very hot water. In addition to this: you MUST know what the tick size and tick value is at your broker. This can differ sometimes. And in order to calculate positions sizes and rate each instrument you will need to normalize the values depending on the tick size and tick value. In other words: you need to normalize the basic increment of ATR in $ across all instruments.
Regarding the spreadsheet:
First of all the position sizes are shown. And these are on a per trade basis. As it pertains to the TPS scaling-in: the figures are for FULL positions NOT per SIGNAL. Big difference between the two. In other words and based on the current figures: you’d be wanting to end up with or eventually scale-in to 5 FULL positions on the Dow. NO MORE.
In addition: you want to be trading the top four to six instruments that are the most volatile. In other words: you want to be trading those with the highest ATR values.
Taking the above even one step further: you’d also want to be trading the top four to six instruments that not only are volatile but that also have directional movement. Think about it: the more directional movement the sooner these trades will be over after their pullbacks. So if this is something that you wish to factor in then I’m just multiplying ATR by ADXR. This is a stripped down version of Wilder’s CSI as noted i.e. I’ve not included margin requirements and trading costs as I think it’s a bit of an overkill for our purposes here.
I can attest to the merits of the above. For the last few years I’ve been trading anything and everything that generates a signal. But I’ve noticed over the years that some instruments during certain periods are just not worth trading. They hog margin and seem to go nowhere so profits are less. Since implementing this I’ve seen a noticeable change in this regard.
In addition and since implementing the risk and position size calculations the wild intraday swings in floating P&L has all but disappeared. In addition: these calculations seem to normalize the instruments against each other (which is obvious given the different positions sizes indicated). What’s more: one should never even come close to a margin call on this basis.
In addition: just based on last week I’m confident that using a fixed stop is the way to go and on this basis will not interfere with the trading system itself. This fixed stop is for those times where these markets begin to trend against you without even so much as pausing to look back (hence not even giving you a signal to exit at a loss). Take a look at some charts and you’ll clearly see what I mean i.e. it’s happened before and it will happen again (and would not surprise me if these coming weeks do not serve as a prime example in real time). This fixed stop is for those times when things just get totally out of hand.
Lastly and as I stated earlier in another post: this entire method of position size and risk management seems to hold together almost in a perfect balance with this trading system.
So there you have it. Not the tidiest of spreadsheets. Sorry about that. Serves my purposes though.
These calculations should be done weekly over a weekend and the figures would be valid for the coming week’s trades. Given that I’m using the default 14-day period for ATR this should be accurate enough. One could, I suppose, use a 7-day period for ATR and ADXR. Maybe even a 5-day period (possibly makes sense as these trades theoretically should only last about that long under normal market conditions). This would of course speed up the reaction time for the calculations. As to how much of a difference it may make: I know not. Just thought I’d mention it is all.
Just another note that I believe is of interest. In just doing these calculations of late AND, oddly enough, doing something similar some years ago but for a different reason: the S&P 500 almost always is top of the pops. I believe this is proof of concept enough and why it’s the go to instrument for those that trade the indices. It is currenlty only surpassed by Italy and Honk Kong (depending which of the two rating systems you implement i.e. ATR only or ATR and ADXR). But most would not trade those indices I don’t think so it would be of no concern to them I suppose. As far as the three US indices are concerned: very rarely does the Dow top theS&P. The NASDAQ is always the slower of the three (this in spite of the fact that I do firmly believe that the NASDAQ actually leads the market due to its component stocks and sectors) (although this is a moot point really i.e. not something I’d base a trading decision on).
With regard to the last paragraph above: a 5-day period is too short. It would have you rotating in and out of instruments far too often. Just too much work really. Seems to me that maybe a 10-day ATR and ADXR could be the sweet spot. And I’m willing to bet this is why Jeff used a 10-period ATR as well. And Wilder does actually state the the period for ANY indicator should be equal to one half of the period under review. Given that the markets are closed over weekends then a 10-period interval makes sense I suppose. It does of course beg the question as to why every single one of Wilder’s indicators defaults to 14-periods of course (seem to remember reading something in his Delta Phenomenon about this but don’t quite recall at this time).