Dozu, for purposes of comparison here are the summary results of the testing I did on a very simple system for the NQs last year (2000). The testing was not rigorous, it did not include every day of the month and the exclusions were random, not based on systematic predefined conditions. Also the entries and exits were based on interpolation from charts rather than rigorously calculated. No allowance was made for slippage and commissions. The column headings are Month, Days Included, Number of Trades, Gross Points, Average Points per Trade April, 11, 54, 1721, 31.9 May, 16, 69, 520, 7.5 June, 19, 70, 542, 7.7 July, 15, 88, 732, 8.3 Aug, 23, 132, 560, 4.2 Sep, 4, 19, 159, 8.4 The results looked pretty darn good to me, although let me emphasize I have never traded a future in my life and I am not an expert in system development or evaluation. In fact, sufficiently good that I was willing to give it a try with real money. Unfortunately the Canadian regulators just then cut off access to the broker of my choice, so I didn't. I had another look at it recently, but in nowhere near such detail. My first observation was that the index is now only half what it was. So it is likely that my average points per trade figure would be cut in half. However, slippage and commissions would be unchanged, so we are now perhaps looking at a net edge per trade in the 2 point range. With no experience I am not sure how good that guess is. Not only that, but the limited amount of testing I did using more recent data suggests that the big wins are fewer and farther between due to the choppier nature of the markets, so my average gain might be less still. My conclusion was that I had better do a hell of a lot more back testing over a wider range of market conditions if I am seriously going to consider trading my rules. Perhaps the regulators did me a favour by stopping me. But you can be sure I will be seriously interested again if and when we revisit 4000 on the index.
Dozu, I took a break for supper in the middle of my message, so I did not see yours until I posted mine. It looks like we have come to very similar conclusions re market levels and conditions.
I read a book recently that summed trading up best in that anyone can look at the center of a chart and say I would have bought or sold here. The same seems to be true for most systems in general, until the execution goes through and a profit is booked, can you really look back at the chart and say it worked? I traded part time for years, and in Aug 2000 left my job and decided to do it full time. That was timing mistake number 1!!! But with that behind me, and a rather painful year of mistakes, I have found a few things to be true. If I watch, my indicators always work perfectly. Once I buy, they may work, but my mind starts reeling and greed and fear set in. If I sell, the price runs off without me. I tell myself to not be such and wuss, I hold, the bids go to lunch and I get stopped out. If it is a really good set up, I can't get an execution at the ask. The most recent was the market bottom, I did take some positions due to my indicators, but was scared silly and traded out for small profits (and a few losses) way too soon. Who new? I have found, that the best for me is to have several things that work reasonably well, including sleeping at night. Best of luck to all.
Here are the Dec contract results for the NQ system I trade. The system trades on a 1m chart, and uses a smoothed RSI (RSX from Jurik) for entry signals. Exits are a fixed 15pt target and a 30min donchian trailing stop, with a 12 pt crash stop. I have backtested this system over about 18 months and about 2000 trades (system takes an average of 6 trades per day). I have traded the system live for about 300 trades. Live results are similar to backtest, except for slippage on the entry, which is a market order. Stop exits are slipped on *maybe* one out of 20 trades...Globex is sweet. No commissions are included in the numbers below. Comments appreciated. -Jeff TradeStation Strategy Performance Report - RSX1 NQ Z1-1 min. Performance Summary: All Trades Total Net Profit $13,300.00 Gross Profit $27,600.00 Gross Loss ($14,300.00) Total # of trades 301 Percent profitable 46.51% Number winning trades 140 Number losing trades 161 Largest winning trade $300.00 Largest losing trade ($240.00) Average winning trade $197.14 Average losing trade ($88.82) Ratio avg win/avg loss 2.22 Avg trade (win & loss) $44.19 Max consec. Winners 6 Max consec. losers 8 Avg # bars in winners 31 Avg # bars in losers 16 Max intraday drawdown ($1,200.00) Profit Factor 1.93 Max # contracts held 1 Account size required $1,200.00 Return on account 1108.33%
Hi Jeff, Congrats on some great looking results for your system! Certainly 300 or so trades gives you a pretty good starting point for comparison. It sounds like this is doing well for you. As the summer markets are different, I'm wondering if you've tested this on any Sept contracts. Are your results similar? I have some similar short term systems, and I've found that I lose about 1 tick on a market entry, especially those systems where I'm entering at an extreme. Using RSX would suggest to me that you are basing your entries on extremes as well, although if you are entering countertrend (i.e. on the RSX move into rather than out of an extreme overbought or oversold), then you are probably seeing some positive slippage, which is great. If you had only the backtested results, without your actual trades for comparison, I'd be very concerned about the $44/trade average (2 NQ pts). Commission is not a big deal these days, but slippage is, especially on shorter term systems. In fast markets, it is very likely that someone could see a couple of points slippage which would wipe out that average. Just out of curiosity, and I certainly don't want you to divulge your system, so please ignore this if it's too much detail: Do you use any kind of a short term trend filter for your entries? (e.g. if the 5 min LinRegSlope is negative, then only take the Sell signals, or use the LinRegSlope to modify/shift your trigger levels)? I've switched back and forth between minute and tick based charts. I've found that especially in volatile markets, a 200 or 300 tick chart (nominally equal to 1 min in ES, NQ) produces a smoother trend with fewer wide ranging bars, and works better in my systems. You may try that if you haven't already.
Dave, Thanks for your reply. I have both tested and traded the system in the summer. It sucks I traded it this summer through a painful drawdown and subsequent recovery. But it didn't feel very good I can tell you that. I only had 1 summer of backtest results before this year, so I wasn't properly prepared. Next summer, I will simply pass on trading most of June and July. August seems ok. The system enters trades as RSX moves out of the extremes, using a 30wma as a trend filter. I.e. WMA must be above WMA[1] for longs, reverse for shorts. Sometimes, this does have me entering during fast moves, so there is some slippage. But "slippage" for a market order is hard to define anyway. By definition, when you bang the MKT key, you get what you get. Its hard to call it "slippage", unlike getting slipped on a stop order. But having said that, I rarely get fills that I think are unfair. I can't remember the last time I was filled over 1 point away from where I thought I should be. I have searched and searched for a good filter to keep the system out of chop. But everything I have tried decreases the bottom line and does not provide enough benefit to $avg trade or $max DD. So I am still taking all the trades and searching more. I really like tick charts, but I end up sticking with minute charts so I can overlay other data when necessary. For example, one of the filters I tried was using TRINQ. Plus, until recently, I was on quote.com for data. I frequently didn't trust their data to keep up a good feed, which plays havoc with tick charts LOL. I can laugh now, since I've switched to esignal. QCharts was a nice application, but I just couldn't handle the bad data quality anymore.
That system did well moving into the December holiday period. I knew it was going to be choppy during the holiday, so I stopped trading the system. In January, I began monitoring its trades again, and was going to resume trading when it returned to "normal". Hmmm....It never happened! I'm not sure what character of the market changed, but it did. Since Jan 1, the system is actually up a couple thousand, but with an ugly drawdown in the middle. -Jeff PS: If anyone is interested, I have a great NQ trading system that I have just released to the public. I will only sell copies to the first 20 customers, so call now!