Sharky's Trading Diary Has Moved!

Please update your bookmarks to:
www.thinkingstuff.com/blog.aspx.

New Survey Heats Up

The new survey asks:

I'd love to subscribe to this fantastic software, but...
It's too expensive
I couldn't work out how to use it
I'm waiting for v1.0 to be released
It doesn't connect to the broker I use

After 9 days of constant battle between the 4 options, we have a winner starting to emerge.

Out of the 1 person who answered the survey so far, a total of 1 person (100%) said that they couldn't work out how to use it.

Now, if I only had extensive online help files, I could understand that someone couldn't work it out. If I only had extensive online help files, plus help sections attached to the more tricky windows, I could understand that someone couldn't work it out.

But if I have extensive online help files, plus help sections attached to the more tricky windows, plus free and unlimited support, I can't understand it.

If you're having trouble, ask me a question. Live chat is available.

(And if that 1 person is the same 1 person who tried to use the Contact Us page while it was broken - sincere apologies).

Contact Us Now Works!

Dear anybody who tried to use the Contact Us page during the past week,

Although a confirmation message told you that the message was sent, it was not sent. But if you try again now, it will be.

Apologies all round.

No Gridding

Gridding is where you believe the price to be moving in one direction, so you set up a whole bunch of limit orders at regular intervals (let's say every 10 pips) both above and below the current price.

It means you're buying more and more regardless of where the price is going in the short-term. The theory is that because the price is eventually going to go in your favour, you want to keep buying more and more at hopefully cheaper prices. When the big move comes, you rake in the dough.

And just to finish off the gridding explanation, I think also when one limit order is taken up, you're supposed to place another one so that you've always got 10 limit orders at any one time. Or maybe it's 10 limit orders on either side of the current price. Or whatever.

It's a lot like dollar-cost-averaging, which I already said I don't like. The reason I don't like either of these is because prices don't always come back. Sure you're buying cheaper than what you originally got in at, and this is fantastic if the price does rebound, but it's no help at all if the price never makes it back to these "cheap" prices. In fact you're just multiplying your losses.

Now when I say "No Gridding" in the title, I don't really mean to say to you that you shouldn't do gridding. I wouldn't. But then I wouldn't set up a website for people to post videos and 2 years later rake in $1.5 billion.

What I mean by "No Gridding" in this context, is that the Thinking Stuff software will not be able to support gridding. Sure you could have made 20 different trading systems which were set to place their entry orders at "Close Minus 10 Pips", "Close Minus 20 Pips", etc, but even so I made a change in v0.8.3 that just won't allow it.

Gain Capital has a heart attack if you try to put a Long limit order below the current price. But Oanda allows it, and such a limit order will be taken up when the price pushes *down* through the limit order price, rather than pushing up through it as normal.

I had this as a bug - a limit order had been placed, and due to some latency between retrieving the price and placing the order, the price had already shot up above where the Long limit order was put. What to do in this case? The problem is of course, that the algorithm which works out if we should be in a trade will think that we should be in a trade - a Long limit order was placed during the previous bar, and the current price is above where it was placed. "We're in a trade!", the algorithm will think to itself. And then it will discover that it should be in a Long trade, but actually there's no trade and still a Long order. "Huh?" it'll say.

So, I decided the best thing to do would be to cancel the order altogether, and when the current bar finishes, recalculate where the order should be placed.

i.e. No Gridding.

I know I didn't have to say all that, because it's in the Change Log. And I know that everyone reads the Change Log because it's so fantastically interesting.

v0.8.3 - Market Orders

From How Does Thinking Stuff Decrease Risk?:

No Market Orders

All orders placed are limit orders.

It requires that the price has enough momentum to bust through where your limit order is set in order to get you in to a trade.

Also, with Gain Capital at least, you cannot set a Stop Loss at the same time as opening a market order. A Stop Loss can be added later, however that option is not decreasing the risk involved - for a short time your entire capital is theoretically vulnerable.

While potentially controversial, it is our opinion that using market orders is a sign that you are fearful of missing out on money, and greed wants you to win as much money as you can on the trade, and therefore have to get in now! now! now!

Whereas a more detached, unemotional approach would be to place a limit order, and then allow the price to prove it has momentum by continuing far enough in the direction you wanted so that your limit order is taken up.

In Thinking Stuff, it's (1) all the entry rules are met; and (2) the limit order is set; and (3) the price must then continue in the direction you wanted in order to get you in the trade.

Unless you are implementing a scalping technique, missing out on a couple of pips should not be enough to break your trading system (although we are happy to read through rebuttles to our argument).

To contradict that completely, Thinking Stuff now has market orders. Why? Because someone said they'd subscribe if it had market orders. So now it has market orders.

Yes, I'm a programming whore!

Incidentally, that person did in fact want to do a scalping-like technique.

Market orders are still not allowed for Gain Capital though - only Oanda, because Oanda allows stop losses to be put at the same time as creating the market order.

An Entry Value of "Enter At Market" overrides any other Entry Value setting you might have.

In backtesting, the entry point is assumed to be the Open price of the next bar. This may or may not be true in real trading.

I also did a number of bug fixes.

See the full list of changes at the Change Log. See what's coming next at the Road Map. Also read through the Known Bugs so you know what you're getting. And then go and purchase.

Oh, v0.8.3 was going to be Release Candidate 1, but due to the request for market orders I wanted to get this out as soon as possible. Most likely v0.8.4 will become RC1.

Most likely.

Survey Results

Just putting some surveys to bed. Thanks to all those many people who contributed.

What else do you want Thinking Stuff to do?
0% - Auto-trading of shares.
29% - Auto-trading of futures.
29% - Add more trading rules for us to choose from
43% - Just fix the bugs that are there now - then do new stuff

Thinking Stuff currently connects to Gain Capital and Oanda. If you had to choose, which other company should it connect to?
50% - Interactive Brokers
25% - FXCM
25% - InterbankFX

Total number of respondents was 7 for the first question, and a whopping 8 for the second. Yes siree, obviously a hot topic for discussion around the ol' water cooler.

To respond, yes, I've drawn a line in the sand for what will be included in v1.0, and that is not much else besides fixing the remaining bugs. No saying I don't listen to the voices of the 7 survey contributors.

If you would like more trading rules to choose from, I guess the next step is to tell me what they are. Now that I've incorporated TA-Lib, it will be very easy to create rules based on any of the indicators listed here.

And gosh, I'd love to have it connect with Interactive Brokers, but it's not so simple. IB offers shares and futures and options and currencies, whereas the entire framework of my software deals solely in currency at the moment. There's a set number of currencies, they work in pips, and have no volume. Shares have volume, work in cents, and there's thousands of different stocks, being added to and subtracted from all the time. So to offer auto-trading in the other financial instruments would mean a major overhaul taking many months.

But that's my problem, not yours. There are 4 people who might be very unhappy right about now. Apologies. Make that 8, because FXCM doesn't have a suitable API, and InterbankFX doesn't have examples in VB.Net.

The next survey (available here) asks:

I'd love to subscribe to this fantastic software, but...
It's too expensive
I couldn't work out how to use it
I'm waiting for v1.0 to be released
It doesn't connect to the broker I use

p.s. None of those answers are acceptable - go buy it :-)

Carry Trading

Carry Trading is where you buy and hold a currency in order to accrue interest payments.

For example, and this and all figures are taken from the Excel file produced at ElectricSavant.com, which in turn takes its figures from Oanda, if you buy 10,000 units of AUD/JPY at 1:50 margin, you get paid interest at 250% p.a.

250%!!

That's if the price of AUD/JPY remained static for that year. If AUD/JPY went higher, you make even more. If AUD/JPY goes lower, you might still make money if the interest payments are more than the amount you lose in terms of pips.

From that Excel file, the top ten are:
AUDJPY, Long: 250%
GBPJPY, Long: 208%
USDJPY, Long: 200%
GBPCHF, Long: 168%
USDCHF, Long: 161%
CADJPY, Long: 153%
EURAUD, Short: 130%
NZDUSD, Long: 122%
EURJPY, Long: 100%
EURGBP, Short: 87%

So I set about finding if there was some kind of fool-proof way to rake in the interest, and hedge against any potential drop in price.

The thing to note is that if you do a perfect hedge - that is you go long 10,000 units of AUD/JPY in one account, and go short 10,000 units of AUD/JPY in another, then you'll be paying your broker interest. Because here's the reverse of the top ten above:
AUDJPY, Short: -286%
GBPJPY, Short: -237%
USDJPY, Short: -239%
GBPCHF, Short: -192%
USDCHF, Short: -195%
CADJPY, Short: -189%
EURAUD, Long: -168%
NZDUSD, Short: -170%
EURJPY, Short: -137%
EURGBP, Long: -119%

The 250% interest from your long AUDJPY trade is wiped out by the -286% interest from your short.

So, what some people do is create a "Carry Basket" - a group of currencies that combined pay a high amount of interest, but the price is as flat as possible. Every day or week or so, the Carry Trader shifts money around so that the ratio of currencies owned remains fairly constant.

One such basket I read about was:
EURCHF, Long: 61%
GBPCHF, Long: 168%
EURGBP, Short: 87%
EURAUD, Short: 130%

(Actually the ratio was to buy 2x EURAUD than the others, but let's keep things simple).

What I did was, using Daily bars back to the start of 2005, worked out for each day how much interest was accrued, and how much money was gained/lost due to the fluctuations in price. Then I added those figures together to see how much money would have been made in a buy and hold strategy.

Note that I did not use compounding. Everyday the same 10,000 units of each was kept.

The end result is quite fantastic - $4500 in profit, when the margin required to buy those 40,000 units (20K in CHF, 10K in GBP, 10K in AUD) at 1:50 was about $1200. As with all these images, click on it to see the full-size version.

The interest received was around $1500. Meaning $3000 came from price movement.

You can see from these charts, that the big moves in profit all come from the big moves of price in our favour (that's the yellow sections):

EURCHF:

GBPCHF:

EURGBP:

EURAUD:

Let's instead start everything at exactly the worst time - 22-June-05. Markets do change personality, and I assume it's going to happen as soon as I place my first trade using a new system that backtested well. The cosmos might want me to succeed, but it also has a sense of humour.

Now the chart is not so good:

While the interest certainly does lessen the impact of the price going in the wrong direction, it doesn't make up for it altogether. It would have been great to earn all that interest, but the only good part about it for more than a year was that without it, you would have been losing more than you were.

If you add compounding to the mix, I'm not sure the results would be much different. Buying 2x more EURAUD than the others didn't change the shape of the chart either.

So what does this mean? What's the conclusion?

I think that for some currencies, the interest is too much to ignore. However I also think that price fluctuations will always outweigh the amount of interest earned. Interest would be the icing on the cake for trades that went well, and take the bite out of trades that don't.

So... if you want to trade long-term in currency and rake in some interest, how about developing a trading system that only trades in the direction that pays out interest, and base that system on Daily or Weekly bars.

Following the normal strategy - you backtest over historical data, taking into consideration the amount of pips earned and the number of days in the trade. More is better for both. Then instead of buying and holding, and adjusting once per week, you just trade your system - place order, set stop loss. Repeat. Simple.

When I read about Carry Trading in forums, I see the words "sell when you make money", "buy more when the price is going down", "dollar-cost average". None of these make sense to me. These tactics assume the price is always going to rebound, so you're buying while it's "cheap". I don't like that kind of assumption.

v0.8.2 - Candlesticks

I incorporated the open-source project of TA-Lib, and all candlestick calculations come from that. There are quite a few (see the full list here), but a lot of them use gaps and therefore those that do aren't so useful for currency trading.

Sorry for the delay in getting this smallish release out. I had to study up for a job interview, and then I researched Carry Trading to see if and how I could use it. Carry Trading is where you essentially buy and hold a currency for the (fantastic) interest you receive. I'll put up my thoughts soon.

Also fixed in this release is an embarrassing error that didn't allow any new trading systems to be added. I hope any newcomers that tried v0.8.1 will try out v0.8.2.

See the full list of changes at the Change Log. See what's coming next at the Road Map. Also read through the Known Bugs so you know what you're getting. And then go and purchase. I should really get around to putting up the Alerts prices.

Change of Email Address

I've decided to ditch this email address:

I'm a switchin' completely to Gmail, which gives me more than 2.5Gb of space, and better spam-handling capabilities.

The old address will continue to work for a while. But, instead please use the last bit of the above (minus .com of course) at Gmail.com.

If in doubt, use the Contact Us page.

v0.8.1 Available

Go get it. And buy it. And give me some feedback.