The Gateway to Algorithmic and Automated Trading

The art of the fast exit

Published in Automated Trader Magazine Issue 31 Q4 2013

If anyone deserves the label 'market veteran', it may be Jim Moore. Since getting into the business in the late 1960s, he's worked with some of the biggest names around. He also embraced algorithmic trading early on. Adam Cox talks to Jim about his trading techniques and investment philosophy.

Jim Moore, TIF Fund Management

Jim Moore, TIF Fund Management

Jim Moore, head of TIF Fund Management, has worked with some of the biggest names in the business over the years, from Sandy Weill to Paul Tudor Jones. He was also an early advocate of algorithmic trading in the 1990s and the largest trading account in his company is fully automated. As Adam Cox learns, Jim may have been in the market for more than 40 years, but he always wants to be ready to get out in 40 seconds.

Adam: You have a system you developed over many years. Let's talk about how it came to be.

Jim: Basically, we don't believe in buy and hold. We believe that the markets are so volatile today and commissions are so low that it is cheaper for us to exit a losing trade, and look for a safer place to re-enter the market. We are not members of or associated with any broker-dealer and we get no share of commissions, so it's best for our clients for us to keep the commissions as low as we possibly can, and still get good executions of course. Essentially, we protect our capital by trading very actively.

We trade about 80 times a day, 40 are typically new entries, and 40 are liquidations. And that's the underlying concept. It's specifically oriented towards very, very low risk, and active trading. Our average trade for a loss is about 97 minutes, our average trade for a gain is a little under three days, and therefore our average trade is about 1.4 days.

Adam: Regarding the losers, it sounds like you have tight stop-losses and as soon as you get to a certain point you want to get out.

Jim: It might well start losing money right from the outset. And that's okay, but when it gets to a certain point, we exit. Our average loss is less than 1/30th of one percent of our AUM.

Adam: So if you have such small tolerance for downside, does that not add new risk in terms of knocking you out of a lot of positions early on?

Jim: I'm a pilot, and the one thing I can tell you about pilots who are old is a) they're not too bold, and b) they really in most cases would prefer to be on the ground wishing they were in the air, than in the air in a bad storm wishing they were on the ground. We built this program with the exact same philosophy; we would rather be out of a trade, looking for a safe place to enter, rather than be in a bad trade, without a logical place to exit. If you anchor properly, you can be in and out with very little loss and look for another trade in the same market or else proceed to the next opportunity.

Adam: How does the system determine a safe point of entry for a new market?

Jim: Now you're asking for our secret.

Adam: Can you talk about it in general terms?

Jim: I will answer that to some degree. The program designer has looked at the trades and said 'Look, there's a place to enter where the risk is low, and I'd rather enter it and try, and if it doesn't work, okay, so I'm out for a very small loss'.

Here's the fallacy. I don't know how much you watch the promotional word in our industry today, but a vast number of people claim to be right at 65, 75, 85, even 92% of the time. Two pieces of literature I've read in the last week said that. Believe it or not, our program makes money and we're right on only 42% of our trades. People are typically amazed! We count a break-even as a loss, because it was an inefficient use of our money. We committed capital that didn't work for our investors. We might not have lost money but we didn't make any money, therefore it's an inefficient use of our capital. Secondly, why 42%? Why so low? That's not common. Actually, it is common for floor traders. And that's how we think. Basically we take our losses so rapidly, that it brings down that average. But our average gain, what we get on a winning trade, is more than two times what we lose on a losing trade. If you work out that on a non-leveraged basis, no leverage at all, you're still making 9% a year.

Adam: What can you describe about the program?

Jim: It's got 13 long algorithms, it's got 13 short algorithms, and we're developing more all the time. Now we're working more specifically on targets. We look at money in $2.5 million increments.

On that amount of money, we will basically make around 80 trades a day, and yet our commissions at the end of the month are miniscule, because we've got such low rates.

But to answer your question, what we've done is to change what we know are the algorithms that we think will work, and then said, maybe we should have targets. Because, we'll be up 10 or 12 or 14 thousand dollars in a day, and end up - with the algorithm the way it's working - making only three or four thousand. So we're thinking maybe it's not a bad idea to use some targeting. If something gets near a target and starts to fade from it, get out and give it another look-see later.

Adam: Is the system using some kind of pattern-recognition?

Jim: There's no pattern recognition but there are certainly activity patterns that we see. The 13 long algorithms and the 13 short algorithms are looking for either trends, or trend reversals, or intraday reversal that have some rationale to them - not necessarily fundamental, this is really mathematical what we've done.

Adam: This system has evolved over many, many years. When you make changes, what is the process? What sort of back-testing do you do, how do you go about optimising?

Jim: It's pretty interesting work. The 13 original algorithms have never been changed - they're very simple algorithms. However, the markets do change. Market activity does change. And volatility does change. Volume has changed dramatically as I'm sure everyone's aware. So what the designer has done is taken this concept and said 'How do I apply it and adjust to the market?' We've looked at our performance when we started the program, when we changed from futures to securities. And the algorithms matched the signals very well and get out of the losses very well. That's never really changed. But we said, 'What could we possibly change to minimise our risks?' If the algorithms aren't going to change our risks, the only thing we can really change is position size.

We had a very large hedge fund invested with us, and not like what they saw for a very short period of time. We thought, 'You know, maybe they're right and we're wrong'. We didn't want to go about this thinking: we are God's gift to trading. Such arrogance is what gets traders in trouble. We looked at it and realised we were applying very large position sizes to very low-priced securities. But we didn't do that in the futures market; we would apply a very small position size to very high-priced securities.

So we began to experiment, by bringing the lower-priced position sizing down and slightly raising the upper ones, this work was done when Apple and Google were both around $700. And we said, 'Holy crow, this looks good'. We were able to go back and take all of the identical trades we made, no changes whatsoever, no changes in the algorithm, no changes in the method of entry.

By changing the position size; reducing the size of the lower priced securities positions and slightly increasing the size of the higher priced securities, we changed the rate of return dramatically. It's been the reason for the program's success.

We didn't have to go back and do back-testing the way most people do it and apply an algorithm to try to find another signal. We knew the signals worked. We just need to adjust position sizes. We can look someone straight in the eye and say, 'Look, these are all the same 66,000 trades going all the way back to 2007'.

We didn't have to apply an unreal trade execution or size. We just had to reduce the sizing of most of the trades. And that told us that we knew we could have gotten the execution. There was no 'curve fitting' because we weren't making up any execution that we might not have been able to execute in the market, because we were basically in most cases reducing the trade size - 90% of the trades were reduced in size. So we could have gotten it all, couldn't we? We could say to people, 'Wait a minute, we traded 10,000 shares yet we only should have traded 2,000 so we obviously could have executed at this price'.

Adam: So your proof of concept was based on real trades and trades that you know that you could have executed because they were smaller than your actual execution size.

Jim: Absolutely correct. And that all stemmed from one hedge fund giving us a large amount of money and losing less than 1%! We made money the first day we traded for them, then we lost for six straight days. We made on the eighth day and then we lost for seven straight days. Our total loss was less than 90 basis points. But they stopped the trading and did us the best favour possible.

Adam: So did they stick with you then or pull out?

Jim: They pulled out and told us to fix it. Now they're back talking to us.

Adam: The basic program is about 27 years old, is that right?

Jim: The initial research was done in 1987. The actual first trading started in 1993.

We feel that the algorithms are not a secret. There's nothing exceptional about the algorithms, they are just tools. We think the real value of the program design is the Portfolio Selection Process.

We presently trade 1,200 very active, very liquid US securities. The computer determines which will likely be the most successful trades. It misses some trades but that is a function of the amount of capital available. It would miss fewer trades if we had more capital.

On a $2.5 million dollar increment, the largest trade that we make is 225 shares. And of these 13 algorithms, we have a self-imposed restriction of a maximum of four signals either long or short. This limits any potential over-concentration in any one market.

For example, on every $2.5 million increment, the largest position we take is 900 shares. Say a $10 stock that's trading somewhere between $10 and up to $50. We would trade 225 shares per trade. If there were four separate signals in the same direction, a maximum of four trades would be generated. Generally the return would be around 9% a year, non-leveraged.

The portfolio selector is the real brains of the program. It's not the algorithms. As the designer said, 'I could publish the algorithms. They're trend followers, they're counter-trend followers. They're not the secret of the program.' If there's a secret, it's a) portfolio selection and b) money management - the discipline to follow the signals. The money management is key.

Adam: In the 1990s, it was all futures before you switched to securities in the past decade. In recent times, I presume it wouldn't have been feasible before bid-offer spreads on securities started to come down quite dramatically, given the low loss risk tolerance.

Jim: It's a combination, The bid-ask spread coming down, and lower commissions. The commissions are key. We pay very low commissions.

Adam: Whereas 15 years ago, the commissions would have eaten up a lot of the profit.

Jim: Oh, absolutely. And the brokerage firms would have gotten rich and our clients would have gotten poor. This is not an indictment of brokerage firms - they're in business to make a profit but not at our clients' expense.

Adam: You mentioned that you were all floor traders. Can you talk about your background?

Jim: I was a national sales manager of Shearson back in the 70s. And then I worked for Sandy Weill and I worked for a number of chairmen for Merrill. Years ago, I created something called Cotton Trading Partners, a fabulously successful floor trading fund. Paul Tudor Jones was our first investor, and John Hirsch, who was the chairman of Cotton before Paul, was the trader. It was a very successful fund. It was, I think, one of the top ranked funds for a number of years. But then Johnny stopped trading it and we closed it down and we merged the name into Triumph Fund. And so now Triumph Fund, which used to be 16 separate traders and 16 separate funds, is, is now just one fund with about eight or nine traders.

Adam: How is the overall group performance going right now?

Jim: I think we're up about 9% on the year. August was not very good but our option boys came through like gang busters. It's funny, we have quite a good mix. The mix is really something. When the futures and options are making it, and the equities are losing, our heads are still above water. Might only be up to our nose, but we're still above water.

Adam: The choice of markets now: futures, options and securities. And the futures covers everything from interest rates to commodities?

Jim: Anything you trade that's got volume. We don't trade the things that don't have volume because we know that exiting a loss is equally important as entering at the proper place.

Adam: So what's the most exotic thing you've traded? Have you done anything like weather derivatives, for instance?

Jim: No. We don't do anything we don't know very well and has the volume to support our entrance and exit needs. We have tested our program, and in the past we've been able to get out of the entire position within 40 seconds.

TIF Fund Management LLC

- The group's flagship Triumph Fund offers exposure to accounts run by different traders

- The program with the largest amount of capital invested is fully automated; first developed to trade futures, it now primarily trades securities

- The automated account is up more than 10% this year and has generated positive annual returns every year since the year of inception in 2007, ranging from 6% to 37%, with a maximum drawdown of less than 5%

- Those returns represent a dramatic outperformance of the S&P 500, whose annual performance ranged from -37% to +26% in the same period

- TIF assets under management were as high as $190 million but Moore reduced the fund while taking a break from trading to recover from a hip operation

- Now he's rebuilding the business