Fully automated futures trading

marameo said:
My unit of risk per bet is 1% of my capital. Volatility determines how I should dimension the trade (2xATR). That normally results in a 20% equity being traded. Is that basically the 'annualized volatility target' mentioned in Systematic Trading?

Since I am trading trend following rules (skew>0) is it safe to assume 'pessimistic' Sharpe Ratio should be 0.30 and expected return as 0.30x20% = 6%?

Thanks
More...

I'm not sure what you mean by '20% equity being traded' or '2xATR' as 'dimension the trade', but I suspect you are doing something quite different. If you could write down more clearly an example of how you size a trade without any jargon (since we may be using diferent terms to mean the same thing) then I could tell you how it corresponds to my sizing method.

That's a pretty reasonable expected Sharpe Ratio, however at that Sharpe Ratio you shouldn't be running 20% risk; a maximum of 15%.

GAT
 
SPY etf last close price $311.81

ATR is 'average true range' indicator to calculate SPY volatility (I like this better than standard deviation). Currently is at 6.81

2 times ATR is 6.81 x 2 = 13.62 so my stoploss will be at $311.81 - 13.62 = $298,19

My trading capital is $10000 and I am willing to risk 1% of it ($100)

$100 : 13.62 = 7,34 number of SPY shares to buy. I round that number to 7.

I buy 7 x $311.81 = $2182,67 worth of SPY shares. I call this the 'equity being traded' which is 21,82% of my capital.

Is that 21,82% basically the 'annualized volatility target'?

If my stoploss gets hit I sell my position at 7 x $298,19 = $2087,33 and I realize a $100 loss (1% of my capital).

Those are TurtleTraders rules.

.
.
.

Now, suppose I risk 1% for each other asset classes such as Gold, Corn and Interest Rates.

Overall I risk 4% (4 different asset classes; 1% unit of risk each) and my 'equity being traded' will be around 80% of my trading capital (~20% for each asset class). Is that 80% the 'annualized volatility target' or is still ~20% since those asset classes have low correlation?

Thanks
 
Last edited:
marameo said:
SPY etf last close price $311.81

ATR is 'average true range' indicator to calculate SPY volatility (I like this better than standard deviation). Currently is at 6.81

2 times ATR is 6.81 x 2 = 13.62 so my stoploss will be at $311.81 - 13.62 = $298,19

My trading capital is $10000 and I am willing to risk 1% of it ($100)

$100 : 13.62 = 7,34 number of SPY shares to buy. I round that number to 7.

I buy 7 x $311.81 = $2182,67 worth of SPY shares. I call this the 'equity being traded' which is 21,82% of my capital.

Is that 21,82% basically the 'annualized volatility target'?

If my stoploss gets hit I sell my position at 7 x $298,19 = $2087,33 and I realize a $100 loss (1% of my capital).

Those are TurtleTraders rules.

.
.
.

Now, suppose I risk 1% for each other asset classes such as Gold, Corn and Interest Rates.

Overall I risk 4% (4 different asset classes; 1% unit of risk each) and my 'equity being traded' will be around 80% of my trading capital (~20% for each asset class). Is that 80% the 'annualized volatility target' or is still ~20% since those asset classes have low correlation?

Thanks
More...

Thanks, that's pretty detailed.

An ATR of 6.82 roughly corresponds to a annual standard deviation of ASD=95.5 (using my own estimate that the ratio of SD and ATR is ~14, but this is an average). As a % that's 30.7%. That's a fair bit higher than I get as a measure of standard deviation, but let's just use your number for consistency.

If you own 7.3 SPY shares worth $311 each that's 7.3*311*30/7% = ~$700 of standard deviation on your position. Your trading capital is $10,000 so you're taking a risk of $700/$10K = 7% on that position as a proportion of your capital.

As a general formula your risk will be ASD%*# shares*share price / account value, and as your # shares = 1% * account value / SLM*ATR and ASD% = ATR*14 / share price that works out to (ATR*14/share price)*(1% * account value / SLM* ATR) * share price / account value

= 14 * 1% / SLM = 7% (for an SLM of 2)

If you have 4 positions on, then assuming they are perfectly correlated, your risk would be 4*7% = 28%. In fact (a) if they are from different asset classes there is likely to be a fair bit of diversification, and (b) this only applies if you always have a position on in every asset class. So more realistically your annual expected risk on your account as a whole is probably around 20%. That's a little higher than the 15% I'd advocate for an expected SR of 0.30.

A stop of 2*ATR translates to roughly 0.14 of ASD, which is an average holding period of about 10 days. For SPY ETF that comes out at around 0.06SR units. That's just about on the edge of what is sensible - I wouldn't trade any faster. And SPY is quite a cheap ETF. You may want to consider a wider stoploss

Now the problem is if you do that, you're risk will increase. Which is why I don't like the idea of using your stop loss to calculate your position size. Stop losses should be set purely to determine holding period, indepedent of risk target.

Instead if I rearrange the calculation above, I get:

Expected risk = # of shares * ATR * 14 / portfolio size = ~7%
If I rearrange and substitute target for expected risk:

# of shares = target risk * portfolio size / (ATR * 14)

There is more on this, if you're interested, in my latest book Leveraged Trading. Of course I'm happy to continue answering your questions if you don't want to buy yet another book :)

GAT
 
Last edited:
How does that change should I pyramide (adding to the winners)?

Let's go back to a few weeks ago when SPY was at 301.60 and average true range at 7.76

I buy 6 shares at 301.60 = $1809.6 (18% of my capital)

Stop loss is at 286.08 (2 times ATR = 15.52).

I will add one more unit of risk (1%) when market goes up 1 time the average true range (301.60 + 7.76)

Few weeks later SPY is at 311.81 and average true range is 6.81

I buy 7 shares at 311.81 = $2182,67 (as shown in previous post)

I know own 13 shares at $3992,27 which is around 40% of my capital.

What is my 'annualized volatility target'?

Stop loss in now at 298,19 for the entire position.

I will pyramid a maximum of 3 units of risk.

Thanks

.
.

PS: I will indeed buy your latest book even if I am afraid I am not a huge fan of leverage.

PSS: my ATR is based on a 10 weekly period. Do the ratio of SD and ATR at ~14 still hold?
 
Last edited:
marameo said:
Those are TurtleTraders rules.
More...
marameo, have you done an actual backtest on those rules you posted? The last time I checked, the original turtle trader rules haven't worked since the early 90s.

With a backtest you can also trivially extract those metrics you wanted such as annualized volatility.
 
Those are rules about risk, position sizing and volatility. TurtleTraders than traded trendfollowing rules within this structure as they were fully systematic.
 
GAT

I would like to add this question about your blog post on futures and minimum capital requirement.

Basically, below 10000 EUR only V2X (VSTOXX) futures can be traded. I fully agree with this.

Yet, how about relative value (statitical arb) trades?

Long /ZT short /ZN (2s10s curve steepening)
Short /V2X DEC 2019 long /V2X FEB 2020 (sell risk premium)

Thanks
 
marameo said:
How does that change should I pyramide (adding to the winners)?

Let's go back to a few weeks ago when SPY was at 301.60 and average true range at 7.76

I buy 6 shares at 301.60 = $1809.6 (18% of my capital)

Stop loss is at 286.08 (2 times ATR = 15.52).

I will add one more unit of risk (1%) when market goes up 1 time the average true range (301.60 + 7.76)

Few weeks later SPY is at 311.81 and average true range is 6.81

I buy 7 shares at 311.81 = $2182,67 (as shown in previous post)

I know own 13 shares at $3992,27 which is around 40% of my capital.

What is my 'annualized volatility target'?

Stop loss in now at 298,19 for the entire position.

I will pyramid a maximum of 3 units of risk.

Thanks
More...

Each of your units will increase your risk by the corresponding amount. So if you start with 7%, then you will have 14%, and finally 21%.

When I talk about vol targets, they are expected vol targets for your average position

So if you are pyramidding to 3 units, then it's probably reasonable to assume your average is 2 units. Which means all the numbers in my previous post have to be doubled: Your SPY risk is probably 14%, and your 4 asset portfolio risk is around 20%.

Adding in 3 more asset classes, your expected average risk is more like 40%. That's now starting to feel a bit high.

marameo said:
.
.

PS: I will indeed buy your latest book even if I am afraid I am not a huge fan of leverage.

PSS: my ATR is based on a 10 weekly period. Do the ratio of SD and ATR at ~14 still hold?
More...

Yes, as long as you are still measuring a daily ATR and taking an average, rather than measuring the range over the 10 weeks (given your ATR is 6.8 units, I'd say it's clearly the former.

GAT
 
marameo said:
GAT

I would like to add this question about your blog post on futures and minimum capital requirement.

Basically, below 10000 EUR only V2X (VSTOXX) futures can be traded. I fully agree with this.

Yet, how about relative value (statitical arb) trades?

Long /ZT short /ZN (2s10s curve steepening)
Short /V2X DEC 2019 long /V2X FEB 2020 (sell risk premium)

Thanks
More...

Let's look at the steepener since I'm personally more familiar with this trade. We'll compare this with an outright position on 2's or 10's. Let's assume an arbitrary 25% annual vol target.

Minimum capital = minimum exposure * instrument risk% / target risk% = risk$ / target risk %

[formula from my new book, Leveraged Trading]

For 10 year futures, minimum exposure is 1 contract worth ~$130K. Annual vol is currently ~4% i.e. $5.2K. Minimum capital = $20.8

For 2 year futures, minimum exposure is 1 contract worth ~$214K. Annual vol is currently ~1.2% i.e. $2.6K. Minimum capital = $10.4K

Duration neutral position on 2's 10's will be ~1:2. I reckon annual vol in $ terms of this position is $2.8K (makes sense as about half the 10 year outright). Minimum capital = $11.2K

That's less than the 10 year, but slightly more than the 2 year. That's because we need to buy twice as many contracts in the 2 year than when we do the outright trade.

For the vol calendar trade, (a) we will do a 1:1 ratio*, (b) the correlation is higher so risk even lower. So yes the minimum capital will be even lower here than the outright. I don't have the correlation figure to hand to work it out exactly.

GAT

* As a 1:1 trade this isn't very well balanced as the vol is quite a bit lower on the further out leg, hence there will be some residual exposure to the nearer leg. So if you have the money you'd be better off doing a 4:3 or 3:2 ratio; I don't have the exact figures to hand.
 
globalarbtrader said:
[formula from my new book, Leveraged Trading]

For 10 year futures [...] Annual vol is currently ~4%

For 2 year futures [...] Annual vol is currently ~1.2%

[...]

* As a 1:1 trade this isn't very well balanced as the vol is quite a bit lower on the further out leg, hence there will be some residual exposure to the nearer leg.
More...

Vol is quite a bit higher on the further out?

Thanks ever so much
 
You must log in or register to reply here.
Back
Top