23 replies to this topic

[ogm]

Being a hardcore systems guy, i can't emphasize enough, the importance of having a trading system. Having said that, i would place no importance on the results of the trading systems as they are highly variable and is a function of the market conditions. Past performance is no guarantee of future returns, as clichéd as it is, is a truism in the trading world. I would place more importance of consistency of the system rather than performance. Consistency is everything in this business. What i would do is test the system in various market conditions like

1) Low volatile uptrends
2) High volatile uptrends
3) Low volatile downtrends
4) High volatile downtrends
5) Low volatile range markets
6) High volatile range markets

Well no system can make money or perform equally well in all conditions. Just make sure the money you made in one market condition, is not all given away in another market condition. I have seen so many system developers getting carried away with results they get in a particular market condition as the system is optimized for that type of market - only to later give away all those gains back in another market condition and switch to another system (rinse and repeat). A system which works great in a certain market condition should at least breakeven or perform with minimal drawdown in a another market condition to which it isn't optimized to.

In the end, the bitter truth is, no matter what system you trade, there will be setbacks and drawdowns. It's all about emotionally dealing with those situations. Mechanical or otherwise, a trader has to deal with it. My 5 cents...

NAV,

I agree about consistency. Thats what I was trying to achieve. Fortunately the past 5-6 years have provided just about any market environment I can imagine. A dieing low volatility bull market of 2007-2008, followed by somewhat of a crash condition of 2008, followed by a birth of a new bull market, and all of its phases leading to what looks to be a dieing bull market once again. Not to mention all that overnight gapping insanity in both directions that we went through.

I did try to minimize drawdowns as much as I could. The stops that I've built in are tighter then they should be, With looser stops the % profitable is higher.

As for internals, It depends on how you use them, I guess. I've never liked TRIN for example. Too iffy. NYMO and NAMO are also hard to read, if you don't know what they do. And on Intraday charts they work differently from big picture daily charts. But some things do work. And VIX does work on 60 min charts very well.

As for non confirmation ... the intent of the system is not to catch every possible curve in the market, but to have high profitability ratio, whilie minimiziing risk. There are just as many fakeouts as good trades you can possibly miss. If you notice the longest period out of position for the system is 15 days, and its only 46% of the time in the market overall. It missed quite a few moves. But the drawdown is low. Average drawdown per trade is less then 8 points. As long as it remains that way, I'll be happy, even if it misses a few points here and there.

[ogm]

How about using your automated trading strategy, but you the trader has the final say as to whether the trade is executed. If you get a buy signal and the market has already made a move up then maybe you don't make that trade. If the chart by itself doesn't look right can you improve the performance of the system by picking your trades? Because of the constant monitoring it's really not an automated system anymore but maybe a tool to help your trades. That's what I've been trying lately. I'll find out if it works as well when the volume and volatility return.

I don't want to. The point of trading a system is not to make discretionary decisions. Because you are constantly influenced by news, opinions, etc. Its very hard not to remain influenced. Not many people can. Sometimes I can , sometimes I can't. Like everyone else. I've been trading for almost 15 years since the 90s, and to be honest this environment is the biggest mess I've ever seen. And besides a big chunk, if not most of the trading in the present markets is done by computers. So I'm bringing a gun to a gunfight. The next step of trading evolution.

by the way, if the market has made a move up, good chance it will continue moving up. The idea is not to catch the tops and bottoms. This isn't a predictive system, its reactive.
Constant monitoring is required to make sure the datafeeds are working and that software glitch doesn't occur and the orders get filled and you don't get stuck upside down because of that.

[ogm]

Ogm,

This looks to be a good product. Contact me directly at hedge1208@yahoo.com Hedge funds might be interested in it. What is the capacity?

Denleo

Denleo,

The capacity is high. Its designed to trade ES on 60 min chart. Usually a very liquid high volume market. 1,000 contracts per side, for example, is nothing for that. And thats about 4 mil per side. I also have a version that increases the position size if multiple confirmations start appearing. Profit factor of that version is almost 4. with % profitable near 70%. The version that I posted is 1 contract per side with no pyramiding.

It also trades YM. Can be easily adapted to TF too. I wouldn't do NQ, though because the index is dominated by 4-5 stocks, and that messes up the internals responsiveness. But a version adapted to that can be made.

But in the big picture, since everything moves about at the same time in the same direction, ES is sufficient and definitely has liquidity.


Sent you an email.

Edited by ogm, 19 August 2012 - 09:52 AM.

[arbman]

I ran such funds and my experience is similar to yours. You can trade about 1000 contracts per day and look for a few points. However, even if your allocation model is extremely well (buying and selling in chuncks), the big money is not made this way but rather buying and hedging for large swings. The main problem with all intraday methods is the market swings happen mostly overnight and when they happen during the day, most of the time you won't be stopped out on time and incur huge drawdowns without the proper hedges in place. I completely abondoned the intraday trading engine, but I adapted it to my daily swing system (not automated). I place only a few trades per month and the returns for the past 4-5 months is over 100% by never risking over 30% of the portfolio and in fact the portfolio leverage was also 4:1. But my drawdowns never exceeded 4-5% due to hedging. Essentially you need about 2% moves either way for the hedges to work or make a significant return to offset the cost of hedging etc. I also offset the time cost greatly by selling the near term options (not naked). These are much sane and easier managed, I get tons of time to think and take proper trades. HFT is more like for the firms to front run their customers, not necessarily for low risk market exposure for high returns. Proper long/short option exposure or equity portfolio should perform just fine according to the risk/return objective...

[ogm]

I ran such funds and my experience is similar to yours. You can trade about 1000 contracts per day and look for a few points. However, even if your allocation model is extremely well (buying and selling in chuncks), the big money is not made this way but rather buying and hedging for large swings. The main problem with all intraday methods is the market swings happen mostly overnight and when they happen during the day, most of the time you won't be stopped out on time and incur huge drawdowns without the proper hedges in place. I completely abondoned the intraday trading engine, but I adapted it to my daily swing system (not automated). I place only a few trades per month and the returns for the past 4-5 months is over 100% by never risking over 30% of the portfolio and in fact the portfolio leverage was also 4:1. But my drawdowns never exceeded 4-5% due to hedging. Essentially you need about 2% moves either way for the hedges to work or make a significant return to offset the cost of hedging etc. I also offset the time cost greatly by selling the near term options (not naked). These are much sane and easier managed, I get tons of time to think and take proper trades. HFT is more like for the firms to front run their customers, not necessarily for low risk market exposure for high returns. Proper long/short option exposure or equity portfolio should perform just fine according to the risk/return objective...

This isn't an intraday system per say. It makes decisions based on intraday charts, but within the context of larger timeframes. Average time in trade is almost 3 days.

100 trades per year on average isn't exactly HFT

I tried making systems on 5 and 15 min charts before, and I agree, they are less reliable.

[arbman]

I just want to clarify something, the automated intraday trading systems where you design it once and go to cash every day at the close for a few points to retire from trading is kind of a myth. Not that you will fail at it, but you do work very hard for it as there is always something that changes such as the volatility, the trend or the divergences that simply persist, or simply the liquidity. If anything the broker data feeds can have problems during the day too and you have to constantly baby sit everything. Anyway, the point is I worked much harder while overseeing the intraday trading systems than simply analyzing and designing my trades to place them at the right moment that I do now. I simply favor swing position trading with proper hedges to any other intraday trading style after experiencing both sides of trading for years...

[arbman]

If you are planning to trade up to 3 days with 1000 contracts, you definitely need to have an hedging system in place and it has to be FLAWLESS. You sound like describing a swing system to me, then I would simply place those positions manually.

[NAV]

I agree about consistency. Thats what I was trying to achieve. Fortunately the past 5-6 years have provided just about any market environment I can imagine. A dieing low volatility bull market of 2007-2008, followed by somewhat of a crash condition of 2008, followed by a birth of a new bull market, and all of its phases leading to what looks to be a dieing bull market once again. Not to mention all that overnight gapping insanity in both directions that we went through.

Yes, we have seen most types of market conditions since 2007. But not all.

2005 - 2007 low volatile bull market is one thing you should test with.
1998-2000 high volatile bull is another one.

As for internals, It depends on how you use them, I guess. I've never liked TRIN for example. Too iffy. NYMO and NAMO are also hard to read, if you don't know what they do. And on Intraday charts they work differently from big picture daily charts. But some things do work. And VIX does work on 60 min charts very well.

Well, it just didn't work for me. But i am not saying they don't work. I have much simpler price based tools.

As for non confirmation ... the intent of the system is not to catch every possible curve in the market, but to have high profitability ratio, whilie minimiziing risk. There are just as many fakeouts as good trades you can possibly miss. If you notice the longest period out of position for the system is 15 days, and its only 46% of the time in the market overall. It missed quite a few moves. But the drawdown is low. Average drawdown per trade is less then 8 points. As long as it remains that way, I'll be happy, even if it misses a few points here and there.

I agree.

BTW, the equity curve of your system really looks good...real smooth. I would be curious how it would look like in real testing. If you don't mind, share the results of the equity curve when you implement the system.

Good luck with your system.

[ogm]

If you are planning to trade up to 3 days with 1000 contracts, you definitely need to have an hedging system in place and it has to be FLAWLESS. You sound like describing a swing system to me, then I would simply place those positions manually.

It is a swing system. I'm not planning on trading 1000 contracts myself, was answering Denleo's question on how much volume it can handle. But to be honest I think it can easily handle 1,000 contracts on autopilot. 1,000 contracts isn't a market moving volume in ES futures. Not even on a 1 min chart There are build in stops, and yes, they do trigger once in a while. No way you can avoid a news related gap in the opposite direction that may trigger the stops, or something similar, or simply a bad entry that has to be reversed. But the system for the most part trades with the bigger trend. Especially on the short side. Short side isn't a mirror image of the long side, it may use the same indicators, but its a system in itself, with its own entries and exits. Actually both short and long sides of this system are 2 separate systems, built on the same indicators and principles but separately.

Its a game of probabilities. probability of a profitable entry is 65%. 35% you lose. If I make the system exit faster, it will be 80-85% profitable entries, but since you don't let your profits run, the bad trades will cripple the performance.

[ogm]

Yes, we have seen most types of market conditions since 2007. But not all.

2005 - 2007 low volatile bull market is one thing you should test with.
1998-2000 high volatile bull is another one.

True, but as long as the principle of a price making a move ( primary trigger) confirmed by internals (secondary ) holds, I think it will do just fine, maybe a little choppy here and there, but over the long run it will come out ahead.
Unfortunately tradestation doesn't have 60 min internals data for those periods, but price and VIX parts of the system work there too, I checked. Just couldn't get internals data.

BTW, the equity curve of your system really looks good...real smooth. I would be curious how it would look like in real testing. If you don't mind, share the results of the equity curve when you implement the system.

Good luck with your system.

Thanks. Will do.